{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 100,
  "global_step": 4375,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 5.666074737401576,
      "learning_rate": 4.5454545454545457e-07,
      "loss": 4.9489,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.2919499668519006,
      "learning_rate": 9.090909090909091e-07,
      "loss": 5.0149,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.751572636916668,
      "learning_rate": 1.3636363636363636e-06,
      "loss": 4.7185,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.987975245760819,
      "learning_rate": 1.8181818181818183e-06,
      "loss": 5.3279,
      "step": 4
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.188234053045029,
      "learning_rate": 2.2727272727272728e-06,
      "loss": 4.944,
      "step": 5
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.986833659078025,
      "learning_rate": 2.7272727272727272e-06,
      "loss": 5.04,
      "step": 6
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.509408766722836,
      "learning_rate": 3.181818181818182e-06,
      "loss": 4.9769,
      "step": 7
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.133340991785179,
      "learning_rate": 3.6363636363636366e-06,
      "loss": 5.0269,
      "step": 8
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.748086777344607,
      "learning_rate": 4.090909090909091e-06,
      "loss": 4.9199,
      "step": 9
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.570807269122426,
      "learning_rate": 4.5454545454545455e-06,
      "loss": 5.0529,
      "step": 10
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.482662865356133,
      "learning_rate": 4.9999999999999996e-06,
      "loss": 4.7209,
      "step": 11
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.19866038833127,
      "learning_rate": 5.4545454545454545e-06,
      "loss": 4.9518,
      "step": 12
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.516359721753553,
      "learning_rate": 5.909090909090909e-06,
      "loss": 4.595,
      "step": 13
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.542556314716988,
      "learning_rate": 6.363636363636364e-06,
      "loss": 4.559,
      "step": 14
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.8277529104697567,
      "learning_rate": 6.818181818181818e-06,
      "loss": 4.2907,
      "step": 15
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.835940064085838,
      "learning_rate": 7.272727272727273e-06,
      "loss": 4.1489,
      "step": 16
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.681144900293683,
      "learning_rate": 7.727272727272727e-06,
      "loss": 4.2694,
      "step": 17
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.4638454527836124,
      "learning_rate": 8.181818181818181e-06,
      "loss": 3.9363,
      "step": 18
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.5054715074257823,
      "learning_rate": 8.636363636363637e-06,
      "loss": 3.8886,
      "step": 19
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.296415352777394,
      "learning_rate": 9.090909090909091e-06,
      "loss": 3.7577,
      "step": 20
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0231359784820504,
      "learning_rate": 9.545454545454545e-06,
      "loss": 3.6206,
      "step": 21
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3632420632933506,
      "learning_rate": 9.999999999999999e-06,
      "loss": 3.5292,
      "step": 22
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3699548154483416,
      "learning_rate": 1.0454545454545455e-05,
      "loss": 3.4068,
      "step": 23
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2418478722029422,
      "learning_rate": 1.0909090909090909e-05,
      "loss": 3.3779,
      "step": 24
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9249590022502725,
      "learning_rate": 1.1363636363636365e-05,
      "loss": 3.3395,
      "step": 25
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9206740699894223,
      "learning_rate": 1.1818181818181819e-05,
      "loss": 3.2138,
      "step": 26
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7916478427215432,
      "learning_rate": 1.2272727272727274e-05,
      "loss": 3.145,
      "step": 27
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7797929288426864,
      "learning_rate": 1.2727272727272728e-05,
      "loss": 3.0872,
      "step": 28
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8734126257257293,
      "learning_rate": 1.3181818181818183e-05,
      "loss": 3.0906,
      "step": 29
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8780769632590906,
      "learning_rate": 1.3636363636363637e-05,
      "loss": 3.0253,
      "step": 30
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.855742707297749,
      "learning_rate": 1.4090909090909092e-05,
      "loss": 3.0877,
      "step": 31
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1113923463860287,
      "learning_rate": 1.4545454545454546e-05,
      "loss": 3.0506,
      "step": 32
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6877610348129302,
      "learning_rate": 1.5e-05,
      "loss": 2.7857,
      "step": 33
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6117094686866662,
      "learning_rate": 1.5454545454545454e-05,
      "loss": 2.9063,
      "step": 34
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6904534882009097,
      "learning_rate": 1.590909090909091e-05,
      "loss": 2.9631,
      "step": 35
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9875276410046034,
      "learning_rate": 1.6363636363636363e-05,
      "loss": 2.8355,
      "step": 36
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6526449756719266,
      "learning_rate": 1.6818181818181818e-05,
      "loss": 2.9284,
      "step": 37
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4916417180228962,
      "learning_rate": 1.7272727272727274e-05,
      "loss": 2.9465,
      "step": 38
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.421629820404107,
      "learning_rate": 1.772727272727273e-05,
      "loss": 2.9681,
      "step": 39
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5952739314883118,
      "learning_rate": 1.8181818181818182e-05,
      "loss": 2.7704,
      "step": 40
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3114358517716649,
      "learning_rate": 1.8636363636363638e-05,
      "loss": 2.7069,
      "step": 41
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0880582930147953,
      "learning_rate": 1.909090909090909e-05,
      "loss": 2.8095,
      "step": 42
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2374133575437458,
      "learning_rate": 1.9545454545454546e-05,
      "loss": 2.7479,
      "step": 43
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1629161488331603,
      "learning_rate": 1.9999999999999998e-05,
      "loss": 2.7731,
      "step": 44
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2016954296141704,
      "learning_rate": 2.0454545454545454e-05,
      "loss": 2.7667,
      "step": 45
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.093492531631802,
      "learning_rate": 2.090909090909091e-05,
      "loss": 2.6245,
      "step": 46
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.053066989842202,
      "learning_rate": 2.1363636363636365e-05,
      "loss": 2.8324,
      "step": 47
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9569339469223753,
      "learning_rate": 2.1818181818181818e-05,
      "loss": 2.8369,
      "step": 48
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1797084963687388,
      "learning_rate": 2.2272727272727274e-05,
      "loss": 2.5519,
      "step": 49
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1135764975019935,
      "learning_rate": 2.272727272727273e-05,
      "loss": 2.655,
      "step": 50
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9349891746596473,
      "learning_rate": 2.318181818181818e-05,
      "loss": 2.663,
      "step": 51
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9968515632287099,
      "learning_rate": 2.3636363636363637e-05,
      "loss": 2.621,
      "step": 52
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.093897302228508,
      "learning_rate": 2.409090909090909e-05,
      "loss": 2.6238,
      "step": 53
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.020514616508283,
      "learning_rate": 2.454545454545455e-05,
      "loss": 2.6575,
      "step": 54
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0697292365000932,
      "learning_rate": 2.5e-05,
      "loss": 2.8655,
      "step": 55
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9692039158239506,
      "learning_rate": 2.5454545454545457e-05,
      "loss": 2.7467,
      "step": 56
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0386785172299398,
      "learning_rate": 2.590909090909091e-05,
      "loss": 2.7367,
      "step": 57
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8616749384257502,
      "learning_rate": 2.6363636363636365e-05,
      "loss": 2.4579,
      "step": 58
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1399396219228857,
      "learning_rate": 2.6818181818181817e-05,
      "loss": 2.5703,
      "step": 59
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.032464369002327,
      "learning_rate": 2.7272727272727273e-05,
      "loss": 2.6518,
      "step": 60
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9446867564311866,
      "learning_rate": 2.7727272727272725e-05,
      "loss": 2.6079,
      "step": 61
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0379055661708492,
      "learning_rate": 2.8181818181818185e-05,
      "loss": 2.8108,
      "step": 62
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9814908964204778,
      "learning_rate": 2.8636363636363637e-05,
      "loss": 2.6142,
      "step": 63
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9989573832474283,
      "learning_rate": 2.9090909090909093e-05,
      "loss": 2.562,
      "step": 64
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.010728619202395,
      "learning_rate": 2.9545454545454545e-05,
      "loss": 2.6645,
      "step": 65
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9967509134169045,
      "learning_rate": 3e-05,
      "loss": 2.6174,
      "step": 66
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8394813591733346,
      "learning_rate": 3.0454545454545456e-05,
      "loss": 2.4729,
      "step": 67
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.944231022403772,
      "learning_rate": 3.090909090909091e-05,
      "loss": 2.6105,
      "step": 68
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.020203641132312,
      "learning_rate": 3.1363636363636365e-05,
      "loss": 2.4051,
      "step": 69
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8681209281957354,
      "learning_rate": 3.181818181818182e-05,
      "loss": 2.665,
      "step": 70
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.894846947288812,
      "learning_rate": 3.227272727272727e-05,
      "loss": 2.6227,
      "step": 71
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9339903543314794,
      "learning_rate": 3.2727272727272725e-05,
      "loss": 2.6498,
      "step": 72
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9970702610649745,
      "learning_rate": 3.318181818181818e-05,
      "loss": 2.6025,
      "step": 73
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.929755753129639,
      "learning_rate": 3.3636363636363636e-05,
      "loss": 2.5056,
      "step": 74
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9433273396227504,
      "learning_rate": 3.409090909090909e-05,
      "loss": 2.4232,
      "step": 75
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0146374992207807,
      "learning_rate": 3.454545454545455e-05,
      "loss": 2.564,
      "step": 76
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9740360745085043,
      "learning_rate": 3.5000000000000004e-05,
      "loss": 2.4814,
      "step": 77
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9694068247987073,
      "learning_rate": 3.545454545454546e-05,
      "loss": 2.3345,
      "step": 78
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0597171009976596,
      "learning_rate": 3.590909090909091e-05,
      "loss": 2.5792,
      "step": 79
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9807323404319628,
      "learning_rate": 3.6363636363636364e-05,
      "loss": 2.4772,
      "step": 80
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0116964977111054,
      "learning_rate": 3.681818181818182e-05,
      "loss": 2.4512,
      "step": 81
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9969900431665562,
      "learning_rate": 3.7272727272727276e-05,
      "loss": 2.5098,
      "step": 82
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.088979946991267,
      "learning_rate": 3.772727272727273e-05,
      "loss": 2.561,
      "step": 83
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1059738290380796,
      "learning_rate": 3.818181818181818e-05,
      "loss": 2.4441,
      "step": 84
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9422440657938245,
      "learning_rate": 3.8636363636363636e-05,
      "loss": 2.4861,
      "step": 85
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.054730170241907,
      "learning_rate": 3.909090909090909e-05,
      "loss": 2.5474,
      "step": 86
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0768437082940046,
      "learning_rate": 3.954545454545455e-05,
      "loss": 2.497,
      "step": 87
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1522097326137126,
      "learning_rate": 3.9999999999999996e-05,
      "loss": 2.5083,
      "step": 88
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1000620696418693,
      "learning_rate": 4.045454545454545e-05,
      "loss": 2.5089,
      "step": 89
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9977370626937576,
      "learning_rate": 4.090909090909091e-05,
      "loss": 2.5286,
      "step": 90
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9696332687354436,
      "learning_rate": 4.136363636363637e-05,
      "loss": 2.446,
      "step": 91
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0503409350482367,
      "learning_rate": 4.181818181818182e-05,
      "loss": 2.5586,
      "step": 92
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.048721234345832,
      "learning_rate": 4.2272727272727275e-05,
      "loss": 2.4738,
      "step": 93
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1120970116781703,
      "learning_rate": 4.272727272727273e-05,
      "loss": 2.4832,
      "step": 94
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.062783405433584,
      "learning_rate": 4.318181818181819e-05,
      "loss": 2.4649,
      "step": 95
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0142326000861608,
      "learning_rate": 4.3636363636363636e-05,
      "loss": 2.5139,
      "step": 96
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1253553425952154,
      "learning_rate": 4.409090909090909e-05,
      "loss": 2.4966,
      "step": 97
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.118016183642498,
      "learning_rate": 4.454545454545455e-05,
      "loss": 2.4838,
      "step": 98
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1742726884764112,
      "learning_rate": 4.5e-05,
      "loss": 2.5067,
      "step": 99
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0144076645419544,
      "learning_rate": 4.545454545454546e-05,
      "loss": 2.4972,
      "step": 100
    },
    {
      "epoch": 0.11,
      "eval_blimp_filtered_avg": 0.7274626865671642,
      "eval_blimp_filtered_std": 0.0047905142640717905,
      "step": 100
    },
    {
      "epoch": 0.11,
      "eval_blimp_supplement_avg": 0.8275862068965517,
      "eval_blimp_supplement_std": 0.01640543527809189,
      "step": 100
    },
    {
      "epoch": 0.11,
      "eval_vqa_filtered_avg": 0.57,
      "eval_vqa_filtered_std": 0.04975698519562428,
      "step": 100
    },
    {
      "epoch": 0.11,
      "eval_winoground_filtered_avg": 0.65,
      "eval_winoground_filtered_std": 0.0479372485441102,
      "step": 100
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1273541276116053,
      "learning_rate": 4.590909090909091e-05,
      "loss": 2.5532,
      "step": 101
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1171734574660281,
      "learning_rate": 4.636363636363636e-05,
      "loss": 2.5632,
      "step": 102
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0491689581210082,
      "learning_rate": 4.681818181818182e-05,
      "loss": 2.4285,
      "step": 103
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1261894441434208,
      "learning_rate": 4.7272727272727275e-05,
      "loss": 2.5595,
      "step": 104
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0599746217144759,
      "learning_rate": 4.7727272727272724e-05,
      "loss": 2.4516,
      "step": 105
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.10593891869951,
      "learning_rate": 4.818181818181818e-05,
      "loss": 2.4866,
      "step": 106
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2305455062679165,
      "learning_rate": 4.8636363636363635e-05,
      "loss": 2.6422,
      "step": 107
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0930241499725917,
      "learning_rate": 4.90909090909091e-05,
      "loss": 2.5072,
      "step": 108
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2169825837849546,
      "learning_rate": 4.954545454545455e-05,
      "loss": 2.5056,
      "step": 109
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0541364682132532,
      "learning_rate": 5e-05,
      "loss": 2.4479,
      "step": 110
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0205212646675448,
      "learning_rate": 5.045454545454546e-05,
      "loss": 2.4582,
      "step": 111
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.068366182160805,
      "learning_rate": 5.0909090909090914e-05,
      "loss": 2.4894,
      "step": 112
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1659048182096312,
      "learning_rate": 5.136363636363636e-05,
      "loss": 2.4584,
      "step": 113
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0932581387668807,
      "learning_rate": 5.181818181818182e-05,
      "loss": 2.542,
      "step": 114
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1686365225877713,
      "learning_rate": 5.2272727272727274e-05,
      "loss": 2.5723,
      "step": 115
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0966599349184811,
      "learning_rate": 5.272727272727273e-05,
      "loss": 2.4659,
      "step": 116
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.179477838976062,
      "learning_rate": 5.3181818181818186e-05,
      "loss": 2.4867,
      "step": 117
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0805034439285617,
      "learning_rate": 5.3636363636363635e-05,
      "loss": 2.3385,
      "step": 118
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2494906553215728,
      "learning_rate": 5.409090909090909e-05,
      "loss": 2.5247,
      "step": 119
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.042400057888782,
      "learning_rate": 5.4545454545454546e-05,
      "loss": 2.5425,
      "step": 120
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0605946963560247,
      "learning_rate": 5.5e-05,
      "loss": 2.3139,
      "step": 121
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.014511061492036,
      "learning_rate": 5.545454545454545e-05,
      "loss": 2.4022,
      "step": 122
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.184976727787792,
      "learning_rate": 5.590909090909091e-05,
      "loss": 2.4046,
      "step": 123
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0449841428606084,
      "learning_rate": 5.636363636363637e-05,
      "loss": 2.4336,
      "step": 124
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1209176942061008,
      "learning_rate": 5.6818181818181825e-05,
      "loss": 2.4314,
      "step": 125
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1117666144655272,
      "learning_rate": 5.7272727272727274e-05,
      "loss": 2.528,
      "step": 126
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0676964307854058,
      "learning_rate": 5.772727272727273e-05,
      "loss": 2.5095,
      "step": 127
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1613586225884904,
      "learning_rate": 5.8181818181818185e-05,
      "loss": 2.3891,
      "step": 128
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.038109121066311,
      "learning_rate": 5.863636363636364e-05,
      "loss": 2.3851,
      "step": 129
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.076876386427059,
      "learning_rate": 5.909090909090909e-05,
      "loss": 2.4217,
      "step": 130
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.153425460759264,
      "learning_rate": 5.9545454545454546e-05,
      "loss": 2.3575,
      "step": 131
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0392383055809626,
      "learning_rate": 6e-05,
      "loss": 2.2935,
      "step": 132
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1190484947397057,
      "learning_rate": 5.9999991776722975e-05,
      "loss": 2.4972,
      "step": 133
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1265319407373804,
      "learning_rate": 5.99999671068964e-05,
      "loss": 2.4079,
      "step": 134
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.189109299451847,
      "learning_rate": 5.99999259905338e-05,
      "loss": 2.5243,
      "step": 135
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9700023245299823,
      "learning_rate": 5.999986842765774e-05,
      "loss": 2.403,
      "step": 136
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0887147191524427,
      "learning_rate": 5.999979441829974e-05,
      "loss": 2.3991,
      "step": 137
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0684165040605376,
      "learning_rate": 5.999970396250039e-05,
      "loss": 2.4112,
      "step": 138
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0653370291130542,
      "learning_rate": 5.999959706030929e-05,
      "loss": 2.5204,
      "step": 139
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0904541943435344,
      "learning_rate": 5.999947371178503e-05,
      "loss": 2.3432,
      "step": 140
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0884057443338868,
      "learning_rate": 5.9999333916995236e-05,
      "loss": 2.4796,
      "step": 141
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1318220141614306,
      "learning_rate": 5.9999177676016545e-05,
      "loss": 2.3227,
      "step": 142
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.125313225213161,
      "learning_rate": 5.999900498893462e-05,
      "loss": 2.4278,
      "step": 143
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1064368272548104,
      "learning_rate": 5.999881585584412e-05,
      "loss": 2.2275,
      "step": 144
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0944131071442729,
      "learning_rate": 5.999861027684874e-05,
      "loss": 2.2289,
      "step": 145
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3051854898101996,
      "learning_rate": 5.9998388252061174e-05,
      "loss": 2.5097,
      "step": 146
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9842108977243476,
      "learning_rate": 5.999814978160315e-05,
      "loss": 2.3345,
      "step": 147
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.042682991973711,
      "learning_rate": 5.999789486560539e-05,
      "loss": 2.348,
      "step": 148
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1262837797783634,
      "learning_rate": 5.999762350420766e-05,
      "loss": 2.423,
      "step": 149
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0760065917629242,
      "learning_rate": 5.999733569755871e-05,
      "loss": 2.2787,
      "step": 150
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0018161928457874,
      "learning_rate": 5.999703144581632e-05,
      "loss": 2.3417,
      "step": 151
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9842313585667043,
      "learning_rate": 5.99967107491473e-05,
      "loss": 2.422,
      "step": 152
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.069843855735458,
      "learning_rate": 5.999637360772745e-05,
      "loss": 2.4472,
      "step": 153
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.061664526284171,
      "learning_rate": 5.99960200217416e-05,
      "loss": 2.2256,
      "step": 154
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.026950552113965,
      "learning_rate": 5.9995649991383604e-05,
      "loss": 2.3875,
      "step": 155
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.094631573722599,
      "learning_rate": 5.99952635168563e-05,
      "loss": 2.3085,
      "step": 156
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.124214698256979,
      "learning_rate": 5.999486059837158e-05,
      "loss": 2.3316,
      "step": 157
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0186219855412997,
      "learning_rate": 5.9994441236150314e-05,
      "loss": 2.2396,
      "step": 158
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1671289267698,
      "learning_rate": 5.999400543042241e-05,
      "loss": 2.3694,
      "step": 159
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1392554137260569,
      "learning_rate": 5.999355318142679e-05,
      "loss": 2.3229,
      "step": 160
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1753181950880074,
      "learning_rate": 5.9993084489411384e-05,
      "loss": 2.4072,
      "step": 161
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0967423106372336,
      "learning_rate": 5.999259935463313e-05,
      "loss": 2.2813,
      "step": 162
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0523225924102768,
      "learning_rate": 5.9992097777358e-05,
      "loss": 2.3725,
      "step": 163
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1120082461107597,
      "learning_rate": 5.999157975786096e-05,
      "loss": 2.2576,
      "step": 164
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.204476428607002,
      "learning_rate": 5.999104529642599e-05,
      "loss": 2.345,
      "step": 165
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0050691726851548,
      "learning_rate": 5.9990494393346107e-05,
      "loss": 2.291,
      "step": 166
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0429501131094006,
      "learning_rate": 5.9989927048923315e-05,
      "loss": 2.3028,
      "step": 167
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1055583307308723,
      "learning_rate": 5.9989343263468646e-05,
      "loss": 2.3541,
      "step": 168
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2061753139721523,
      "learning_rate": 5.9988743037302147e-05,
      "loss": 2.3629,
      "step": 169
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0440810760940546,
      "learning_rate": 5.9988126370752874e-05,
      "loss": 2.3361,
      "step": 170
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0312277272913766,
      "learning_rate": 5.998749326415889e-05,
      "loss": 2.3939,
      "step": 171
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0772246584199698,
      "learning_rate": 5.998684371786727e-05,
      "loss": 2.4284,
      "step": 172
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9395965824982881,
      "learning_rate": 5.9986177732234107e-05,
      "loss": 2.3081,
      "step": 173
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.122056526745441,
      "learning_rate": 5.998549530762453e-05,
      "loss": 2.3294,
      "step": 174
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9817503244834807,
      "learning_rate": 5.9984796444412625e-05,
      "loss": 2.231,
      "step": 175
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9823601410420255,
      "learning_rate": 5.998408114298154e-05,
      "loss": 2.3022,
      "step": 176
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0323042500265465,
      "learning_rate": 5.998334940372342e-05,
      "loss": 2.3916,
      "step": 177
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.058506401220237,
      "learning_rate": 5.998260122703941e-05,
      "loss": 2.4341,
      "step": 178
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1093361580601528,
      "learning_rate": 5.998183661333967e-05,
      "loss": 2.4446,
      "step": 179
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1563076187025434,
      "learning_rate": 5.998105556304338e-05,
      "loss": 2.3011,
      "step": 180
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0933857262491447,
      "learning_rate": 5.998025807657873e-05,
      "loss": 2.2742,
      "step": 181
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0177321366532233,
      "learning_rate": 5.997944415438291e-05,
      "loss": 2.3912,
      "step": 182
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1250579028120269,
      "learning_rate": 5.997861379690213e-05,
      "loss": 2.3848,
      "step": 183
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.242416625180627,
      "learning_rate": 5.997776700459162e-05,
      "loss": 2.3462,
      "step": 184
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.142513926411368,
      "learning_rate": 5.997690377791558e-05,
      "loss": 2.2248,
      "step": 185
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1387757398738751,
      "learning_rate": 5.997602411734727e-05,
      "loss": 2.3005,
      "step": 186
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1749224844669794,
      "learning_rate": 5.997512802336893e-05,
      "loss": 2.3053,
      "step": 187
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1623640381571816,
      "learning_rate": 5.997421549647181e-05,
      "loss": 2.2825,
      "step": 188
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1490006846981147,
      "learning_rate": 5.9973286537156175e-05,
      "loss": 2.3228,
      "step": 189
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0139022027885631,
      "learning_rate": 5.997234114593131e-05,
      "loss": 2.2505,
      "step": 190
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3018753970119439,
      "learning_rate": 5.9971379323315473e-05,
      "loss": 2.3308,
      "step": 191
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2074986676092052,
      "learning_rate": 5.9970401069835976e-05,
      "loss": 2.3977,
      "step": 192
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1238775141594575,
      "learning_rate": 5.99694063860291e-05,
      "loss": 2.3697,
      "step": 193
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0464728625371833,
      "learning_rate": 5.996839527244016e-05,
      "loss": 2.3266,
      "step": 194
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0914336011175665,
      "learning_rate": 5.9967367729623455e-05,
      "loss": 2.3164,
      "step": 195
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.030405593664866,
      "learning_rate": 5.996632375814231e-05,
      "loss": 2.1582,
      "step": 196
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0518107891968786,
      "learning_rate": 5.9965263358569056e-05,
      "loss": 2.2346,
      "step": 197
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3007385181652185,
      "learning_rate": 5.996418653148501e-05,
      "loss": 2.4055,
      "step": 198
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0303937232441158,
      "learning_rate": 5.996309327748052e-05,
      "loss": 2.3056,
      "step": 199
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0925089442955271,
      "learning_rate": 5.996198359715492e-05,
      "loss": 2.2144,
      "step": 200
    },
    {
      "epoch": 0.23,
      "eval_blimp_filtered_avg": 0.7225373134328358,
      "eval_blimp_filtered_std": 0.004889470727597283,
      "step": 200
    },
    {
      "epoch": 0.23,
      "eval_blimp_supplement_avg": 0.8211206896551724,
      "eval_blimp_supplement_std": 0.01661446637197484,
      "step": 200
    },
    {
      "epoch": 0.23,
      "eval_vqa_filtered_avg": 0.54,
      "eval_vqa_filtered_std": 0.05009082659620333,
      "step": 200
    },
    {
      "epoch": 0.23,
      "eval_winoground_filtered_avg": 0.65,
      "eval_winoground_filtered_std": 0.0479372485441102,
      "step": 200
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9774564043078218,
      "learning_rate": 5.996085749111657e-05,
      "loss": 2.1919,
      "step": 201
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0834888661267623,
      "learning_rate": 5.9959714959982805e-05,
      "loss": 2.1489,
      "step": 202
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2096929643775005,
      "learning_rate": 5.9958556004379986e-05,
      "loss": 2.3292,
      "step": 203
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1864682192289415,
      "learning_rate": 5.9957380624943485e-05,
      "loss": 2.4413,
      "step": 204
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1568179218150323,
      "learning_rate": 5.995618882231765e-05,
      "loss": 2.4453,
      "step": 205
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1261822986709908,
      "learning_rate": 5.9954980597155866e-05,
      "loss": 2.2424,
      "step": 206
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0814065569548783,
      "learning_rate": 5.9953755950120495e-05,
      "loss": 2.2828,
      "step": 207
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0205970750218325,
      "learning_rate": 5.995251488188291e-05,
      "loss": 2.2397,
      "step": 208
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0902737123504063,
      "learning_rate": 5.9951257393123494e-05,
      "loss": 2.2229,
      "step": 209
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9800168339748077,
      "learning_rate": 5.994998348453162e-05,
      "loss": 2.2933,
      "step": 210
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.065792569949779,
      "learning_rate": 5.9948693156805665e-05,
      "loss": 2.3238,
      "step": 211
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1335113346570447,
      "learning_rate": 5.994738641065302e-05,
      "loss": 2.2316,
      "step": 212
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0652986945071463,
      "learning_rate": 5.994606324679007e-05,
      "loss": 2.2709,
      "step": 213
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0209365615182717,
      "learning_rate": 5.994472366594219e-05,
      "loss": 2.1434,
      "step": 214
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1523016598615023,
      "learning_rate": 5.9943367668843753e-05,
      "loss": 2.3397,
      "step": 215
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0987264877288985,
      "learning_rate": 5.9941995256238155e-05,
      "loss": 2.2445,
      "step": 216
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2002588514035513,
      "learning_rate": 5.9940606428877775e-05,
      "loss": 2.288,
      "step": 217
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2115691248785068,
      "learning_rate": 5.9939201187524e-05,
      "loss": 2.3704,
      "step": 218
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.105314116744233,
      "learning_rate": 5.9937779532947204e-05,
      "loss": 2.2471,
      "step": 219
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1323258012756534,
      "learning_rate": 5.993634146592676e-05,
      "loss": 2.3089,
      "step": 220
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0081099929631019,
      "learning_rate": 5.993488698725104e-05,
      "loss": 2.2958,
      "step": 221
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1059959669054553,
      "learning_rate": 5.993341609771743e-05,
      "loss": 2.2383,
      "step": 222
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.040881173418402,
      "learning_rate": 5.9931928798132296e-05,
      "loss": 2.3723,
      "step": 223
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0557557257610786,
      "learning_rate": 5.9930425089310984e-05,
      "loss": 2.3489,
      "step": 224
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0736150683069619,
      "learning_rate": 5.9928904972077886e-05,
      "loss": 2.236,
      "step": 225
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1507076094437076,
      "learning_rate": 5.992736844726633e-05,
      "loss": 2.4594,
      "step": 226
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1051708872109018,
      "learning_rate": 5.992581551571869e-05,
      "loss": 2.333,
      "step": 227
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0224729635383674,
      "learning_rate": 5.992424617828629e-05,
      "loss": 2.1403,
      "step": 228
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0216535629349073,
      "learning_rate": 5.9922660435829485e-05,
      "loss": 2.1591,
      "step": 229
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1483210870657317,
      "learning_rate": 5.99210582892176e-05,
      "loss": 2.1785,
      "step": 230
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0087915059639985,
      "learning_rate": 5.991943973932898e-05,
      "loss": 2.3214,
      "step": 231
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.08948079698862,
      "learning_rate": 5.9917804787050914e-05,
      "loss": 2.1347,
      "step": 232
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.11956580438021,
      "learning_rate": 5.991615343327973e-05,
      "loss": 2.2124,
      "step": 233
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.190319347650538,
      "learning_rate": 5.991448567892073e-05,
      "loss": 2.2456,
      "step": 234
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0684476418406392,
      "learning_rate": 5.9912801524888214e-05,
      "loss": 2.1197,
      "step": 235
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1465792638120405,
      "learning_rate": 5.9911100972105456e-05,
      "loss": 2.1833,
      "step": 236
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1750831075941757,
      "learning_rate": 5.9909384021504726e-05,
      "loss": 2.3122,
      "step": 237
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0775462654801933,
      "learning_rate": 5.99076506740273e-05,
      "loss": 2.2652,
      "step": 238
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2527246572529314,
      "learning_rate": 5.990590093062343e-05,
      "loss": 2.3132,
      "step": 239
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1482511540772267,
      "learning_rate": 5.9904134792252346e-05,
      "loss": 2.1421,
      "step": 240
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.117561564932975,
      "learning_rate": 5.99023522598823e-05,
      "loss": 2.3,
      "step": 241
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0892760020227148,
      "learning_rate": 5.990055333449048e-05,
      "loss": 2.0887,
      "step": 242
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1898857140618853,
      "learning_rate": 5.989873801706311e-05,
      "loss": 2.2697,
      "step": 243
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0509849793230774,
      "learning_rate": 5.9896906308595376e-05,
      "loss": 2.2294,
      "step": 244
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1186819126565046,
      "learning_rate": 5.989505821009146e-05,
      "loss": 2.2654,
      "step": 245
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.082499570175458,
      "learning_rate": 5.989319372256451e-05,
      "loss": 2.1699,
      "step": 246
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1187002772579915,
      "learning_rate": 5.989131284703668e-05,
      "loss": 2.4093,
      "step": 247
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1603168320290662,
      "learning_rate": 5.988941558453911e-05,
      "loss": 2.3569,
      "step": 248
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1044081134854236,
      "learning_rate": 5.9887501936111894e-05,
      "loss": 2.1611,
      "step": 249
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0879899268263526,
      "learning_rate": 5.9885571902804146e-05,
      "loss": 2.2531,
      "step": 250
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1397570480160115,
      "learning_rate": 5.9883625485673946e-05,
      "loss": 2.1373,
      "step": 251
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1784867749575945,
      "learning_rate": 5.988166268578834e-05,
      "loss": 2.2628,
      "step": 252
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.225213982124372,
      "learning_rate": 5.9879683504223395e-05,
      "loss": 2.3129,
      "step": 253
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.181865919227222,
      "learning_rate": 5.987768794206411e-05,
      "loss": 2.2652,
      "step": 254
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2161219631754998,
      "learning_rate": 5.987567600040451e-05,
      "loss": 2.3543,
      "step": 255
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1947710047398614,
      "learning_rate": 5.987364768034756e-05,
      "loss": 2.1938,
      "step": 256
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0618407183706957,
      "learning_rate": 5.987160298300524e-05,
      "loss": 2.1429,
      "step": 257
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0399906501228942,
      "learning_rate": 5.986954190949847e-05,
      "loss": 2.148,
      "step": 258
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1317931495510045,
      "learning_rate": 5.98674644609572e-05,
      "loss": 2.1883,
      "step": 259
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.110875129493364,
      "learning_rate": 5.986537063852029e-05,
      "loss": 2.2273,
      "step": 260
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1348850460997628,
      "learning_rate": 5.986326044333563e-05,
      "loss": 2.1542,
      "step": 261
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.283268703907935,
      "learning_rate": 5.9861133876560074e-05,
      "loss": 2.4837,
      "step": 262
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0521701496095341,
      "learning_rate": 5.985899093935944e-05,
      "loss": 2.1017,
      "step": 263
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2027431025451878,
      "learning_rate": 5.985683163290852e-05,
      "loss": 2.3164,
      "step": 264
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1053839497123576,
      "learning_rate": 5.9854655958391086e-05,
      "loss": 2.1026,
      "step": 265
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1152665674419877,
      "learning_rate": 5.985246391699988e-05,
      "loss": 2.2367,
      "step": 266
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.141515196802707,
      "learning_rate": 5.9850255509936645e-05,
      "loss": 2.1184,
      "step": 267
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1778257569195774,
      "learning_rate": 5.984803073841204e-05,
      "loss": 2.1501,
      "step": 268
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1545429032384975,
      "learning_rate": 5.984578960364574e-05,
      "loss": 2.1564,
      "step": 269
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2240148307102687,
      "learning_rate": 5.984353210686638e-05,
      "loss": 2.2854,
      "step": 270
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2438344514722743,
      "learning_rate": 5.984125824931154e-05,
      "loss": 2.3312,
      "step": 271
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.123011773334512,
      "learning_rate": 5.9838968032227815e-05,
      "loss": 2.1459,
      "step": 272
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1370376736847747,
      "learning_rate": 5.9836661456870736e-05,
      "loss": 2.13,
      "step": 273
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.096613310891092,
      "learning_rate": 5.9834338524504805e-05,
      "loss": 2.181,
      "step": 274
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1320434077269517,
      "learning_rate": 5.983199923640351e-05,
      "loss": 2.1542,
      "step": 275
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1874736136873716,
      "learning_rate": 5.9829643593849275e-05,
      "loss": 2.1856,
      "step": 276
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1897684090213205,
      "learning_rate": 5.982727159813352e-05,
      "loss": 2.272,
      "step": 277
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1901616386701075,
      "learning_rate": 5.9824883250556605e-05,
      "loss": 2.2043,
      "step": 278
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1342387495628472,
      "learning_rate": 5.982247855242787e-05,
      "loss": 2.3175,
      "step": 279
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1474148510568256,
      "learning_rate": 5.9820057505065623e-05,
      "loss": 2.1896,
      "step": 280
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1948113738963217,
      "learning_rate": 5.981762010979713e-05,
      "loss": 2.235,
      "step": 281
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1585334878276063,
      "learning_rate": 5.9815166367958596e-05,
      "loss": 2.0705,
      "step": 282
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1008914602392745,
      "learning_rate": 5.9812696280895214e-05,
      "loss": 2.1212,
      "step": 283
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.24715432810679,
      "learning_rate": 5.981020984996115e-05,
      "loss": 2.2543,
      "step": 284
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.183186436857975,
      "learning_rate": 5.98077070765195e-05,
      "loss": 2.1951,
      "step": 285
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1423213991741925,
      "learning_rate": 5.9805187961942324e-05,
      "loss": 2.1953,
      "step": 286
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.285513922092449,
      "learning_rate": 5.980265250761064e-05,
      "loss": 2.1728,
      "step": 287
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1458569290491756,
      "learning_rate": 5.980010071491445e-05,
      "loss": 2.0942,
      "step": 288
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1278554403465886,
      "learning_rate": 5.979753258525269e-05,
      "loss": 2.0714,
      "step": 289
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1440221238405026,
      "learning_rate": 5.979494812003325e-05,
      "loss": 2.1106,
      "step": 290
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0762174264188988,
      "learning_rate": 5.979234732067299e-05,
      "loss": 2.211,
      "step": 291
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2187100995125957,
      "learning_rate": 5.9789730188597696e-05,
      "loss": 2.1062,
      "step": 292
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.342862039719802,
      "learning_rate": 5.978709672524216e-05,
      "loss": 2.246,
      "step": 293
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1604606229835743,
      "learning_rate": 5.978444693205006e-05,
      "loss": 2.1777,
      "step": 294
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.218760660947507,
      "learning_rate": 5.978178081047409e-05,
      "loss": 2.15,
      "step": 295
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.127875610286464,
      "learning_rate": 5.977909836197586e-05,
      "loss": 2.1054,
      "step": 296
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1111358707611307,
      "learning_rate": 5.977639958802593e-05,
      "loss": 2.112,
      "step": 297
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1858026042878147,
      "learning_rate": 5.977368449010382e-05,
      "loss": 2.2359,
      "step": 298
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2212194516536041,
      "learning_rate": 5.9770953069698e-05,
      "loss": 2.4041,
      "step": 299
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1671822733181283,
      "learning_rate": 5.976820532830588e-05,
      "loss": 2.2609,
      "step": 300
    },
    {
      "epoch": 0.34,
      "eval_blimp_filtered_avg": 0.7274626865671642,
      "eval_blimp_filtered_std": 0.004907225497658721,
      "step": 300
    },
    {
      "epoch": 0.34,
      "eval_blimp_supplement_avg": 0.8168103448275862,
      "eval_blimp_supplement_std": 0.01675018164655828,
      "step": 300
    },
    {
      "epoch": 0.34,
      "eval_vqa_filtered_avg": 0.5,
      "eval_vqa_filtered_std": 0.050251890762960605,
      "step": 300
    },
    {
      "epoch": 0.34,
      "eval_winoground_filtered_avg": 0.65,
      "eval_winoground_filtered_std": 0.047937248544110196,
      "step": 300
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0847875574826866,
      "learning_rate": 5.9765441267433835e-05,
      "loss": 1.9666,
      "step": 301
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.109806132289268,
      "learning_rate": 5.976266088859716e-05,
      "loss": 2.0893,
      "step": 302
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1996277857286073,
      "learning_rate": 5.975986419332011e-05,
      "loss": 2.1989,
      "step": 303
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1973945861206536,
      "learning_rate": 5.97570511831359e-05,
      "loss": 2.1722,
      "step": 304
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.128376863419922,
      "learning_rate": 5.975422185958666e-05,
      "loss": 2.0475,
      "step": 305
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2433650432753733,
      "learning_rate": 5.975137622422349e-05,
      "loss": 2.247,
      "step": 306
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3286418562033495,
      "learning_rate": 5.9748514278606406e-05,
      "loss": 2.2246,
      "step": 307
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0544782125959788,
      "learning_rate": 5.974563602430438e-05,
      "loss": 2.0213,
      "step": 308
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1501757727720534,
      "learning_rate": 5.974274146289534e-05,
      "loss": 2.1744,
      "step": 309
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1850727651312196,
      "learning_rate": 5.973983059596612e-05,
      "loss": 2.1633,
      "step": 310
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1533497302157019,
      "learning_rate": 5.973690342511252e-05,
      "loss": 2.1379,
      "step": 311
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2458040675384363,
      "learning_rate": 5.973395995193927e-05,
      "loss": 2.1021,
      "step": 312
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1499015179189143,
      "learning_rate": 5.9731000178060035e-05,
      "loss": 1.9215,
      "step": 313
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.170676281785747,
      "learning_rate": 5.9728024105097416e-05,
      "loss": 2.1104,
      "step": 314
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.236996698807854,
      "learning_rate": 5.972503173468295e-05,
      "loss": 2.1732,
      "step": 315
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2135206121835782,
      "learning_rate": 5.972202306845712e-05,
      "loss": 2.1196,
      "step": 316
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.258804543361553,
      "learning_rate": 5.971899810806932e-05,
      "loss": 2.232,
      "step": 317
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.345738395583365,
      "learning_rate": 5.971595685517789e-05,
      "loss": 2.1146,
      "step": 318
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3059724353136946,
      "learning_rate": 5.9712899311450104e-05,
      "loss": 2.0892,
      "step": 319
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3008793373797627,
      "learning_rate": 5.970982547856217e-05,
      "loss": 2.1434,
      "step": 320
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3751998821281481,
      "learning_rate": 5.9706735358199206e-05,
      "loss": 2.2704,
      "step": 321
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3185191067300899,
      "learning_rate": 5.970362895205529e-05,
      "loss": 2.2937,
      "step": 322
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2166686487958462,
      "learning_rate": 5.970050626183339e-05,
      "loss": 2.1679,
      "step": 323
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1167523759042668,
      "learning_rate": 5.9697367289245446e-05,
      "loss": 2.0494,
      "step": 324
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1939485878792584,
      "learning_rate": 5.969421203601229e-05,
      "loss": 2.194,
      "step": 325
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1907805206484845,
      "learning_rate": 5.969104050386368e-05,
      "loss": 2.2172,
      "step": 326
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.266801546755305,
      "learning_rate": 5.968785269453833e-05,
      "loss": 2.0786,
      "step": 327
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1480888100957087,
      "learning_rate": 5.9684648609783836e-05,
      "loss": 2.1797,
      "step": 328
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1283251769983795,
      "learning_rate": 5.9681428251356753e-05,
      "loss": 1.9401,
      "step": 329
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2962642409394396,
      "learning_rate": 5.967819162102252e-05,
      "loss": 2.1035,
      "step": 330
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3530695462174926,
      "learning_rate": 5.967493872055554e-05,
      "loss": 2.206,
      "step": 331
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2461356579430483,
      "learning_rate": 5.96716695517391e-05,
      "loss": 2.0571,
      "step": 332
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2843386319862842,
      "learning_rate": 5.966838411636543e-05,
      "loss": 1.9377,
      "step": 333
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4120078242293248,
      "learning_rate": 5.966508241623564e-05,
      "loss": 2.1966,
      "step": 334
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3580034000333634,
      "learning_rate": 5.966176445315982e-05,
      "loss": 2.238,
      "step": 335
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2822613248496186,
      "learning_rate": 5.9658430228956916e-05,
      "loss": 2.0911,
      "step": 336
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2502759684084332,
      "learning_rate": 5.9655079745454805e-05,
      "loss": 2.1809,
      "step": 337
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1286118605880757,
      "learning_rate": 5.9651713004490304e-05,
      "loss": 2.1164,
      "step": 338
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2967922596340302,
      "learning_rate": 5.9648330007909104e-05,
      "loss": 2.2833,
      "step": 339
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2281869412560216,
      "learning_rate": 5.964493075756585e-05,
      "loss": 2.098,
      "step": 340
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2335150876300958,
      "learning_rate": 5.9641515255324044e-05,
      "loss": 2.0802,
      "step": 341
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1473837884193767,
      "learning_rate": 5.963808350305615e-05,
      "loss": 2.0916,
      "step": 342
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.15828142065349,
      "learning_rate": 5.963463550264351e-05,
      "loss": 2.0617,
      "step": 343
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.269754207691899,
      "learning_rate": 5.9631171255976386e-05,
      "loss": 2.2763,
      "step": 344
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2162018906141552,
      "learning_rate": 5.962769076495393e-05,
      "loss": 2.0954,
      "step": 345
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2648674921989065,
      "learning_rate": 5.9624194031484216e-05,
      "loss": 2.0857,
      "step": 346
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2483122030302949,
      "learning_rate": 5.962068105748423e-05,
      "loss": 2.0967,
      "step": 347
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.199558376001599,
      "learning_rate": 5.9617151844879825e-05,
      "loss": 2.0053,
      "step": 348
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2007890283781477,
      "learning_rate": 5.961360639560581e-05,
      "loss": 2.063,
      "step": 349
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4331197128636792,
      "learning_rate": 5.961004471160585e-05,
      "loss": 2.131,
      "step": 350
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.326014585936025,
      "learning_rate": 5.960646679483253e-05,
      "loss": 2.0171,
      "step": 351
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.292213699924382,
      "learning_rate": 5.960287264724732e-05,
      "loss": 2.157,
      "step": 352
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1691156840334171,
      "learning_rate": 5.9599262270820604e-05,
      "loss": 1.8816,
      "step": 353
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3061804249827427,
      "learning_rate": 5.959563566753167e-05,
      "loss": 2.2176,
      "step": 354
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2107222007122833,
      "learning_rate": 5.959199283936867e-05,
      "loss": 2.0379,
      "step": 355
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2388075485538426,
      "learning_rate": 5.9588333788328675e-05,
      "loss": 2.1965,
      "step": 356
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2468204832125047,
      "learning_rate": 5.958465851641765e-05,
      "loss": 2.1877,
      "step": 357
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.317768168083762,
      "learning_rate": 5.958096702565044e-05,
      "loss": 2.2308,
      "step": 358
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1925200912731977,
      "learning_rate": 5.957725931805079e-05,
      "loss": 2.1155,
      "step": 359
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1658859207091634,
      "learning_rate": 5.957353539565135e-05,
      "loss": 2.0865,
      "step": 360
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1724023761025306,
      "learning_rate": 5.956979526049362e-05,
      "loss": 2.2708,
      "step": 361
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1513504117494973,
      "learning_rate": 5.956603891462802e-05,
      "loss": 2.0046,
      "step": 362
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2755588782635912,
      "learning_rate": 5.9562266360113846e-05,
      "loss": 2.0305,
      "step": 363
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2053149283299738,
      "learning_rate": 5.955847759901929e-05,
      "loss": 2.0227,
      "step": 364
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1969221990012915,
      "learning_rate": 5.9554672633421413e-05,
      "loss": 2.0558,
      "step": 365
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3277900976314116,
      "learning_rate": 5.955085146540617e-05,
      "loss": 2.0578,
      "step": 366
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3284753384494954,
      "learning_rate": 5.9547014097068386e-05,
      "loss": 2.0067,
      "step": 367
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2694768889790735,
      "learning_rate": 5.95431605305118e-05,
      "loss": 2.1271,
      "step": 368
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3280821526076165,
      "learning_rate": 5.953929076784898e-05,
      "loss": 2.1093,
      "step": 369
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3169386067764008,
      "learning_rate": 5.953540481120142e-05,
      "loss": 1.9497,
      "step": 370
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4435893801272768,
      "learning_rate": 5.953150266269948e-05,
      "loss": 1.9448,
      "step": 371
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3684056221150853,
      "learning_rate": 5.952758432448237e-05,
      "loss": 2.1317,
      "step": 372
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3731328718064308,
      "learning_rate": 5.9523649798698206e-05,
      "loss": 1.9454,
      "step": 373
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3629458393097265,
      "learning_rate": 5.9519699087503975e-05,
      "loss": 2.1552,
      "step": 374
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3056887145404739,
      "learning_rate": 5.9515732193065505e-05,
      "loss": 2.0957,
      "step": 375
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3180818665135607,
      "learning_rate": 5.951174911755755e-05,
      "loss": 2.0321,
      "step": 376
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.316587010964975,
      "learning_rate": 5.9507749863163694e-05,
      "loss": 1.9616,
      "step": 377
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2797287511846984,
      "learning_rate": 5.950373443207639e-05,
      "loss": 1.9237,
      "step": 378
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3642454904048777,
      "learning_rate": 5.949970282649699e-05,
      "loss": 2.0329,
      "step": 379
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.359656892903826,
      "learning_rate": 5.949565504863568e-05,
      "loss": 2.1533,
      "step": 380
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4255738203206072,
      "learning_rate": 5.949159110071153e-05,
      "loss": 2.0418,
      "step": 381
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2868086367910299,
      "learning_rate": 5.9487510984952486e-05,
      "loss": 1.9603,
      "step": 382
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3189678370022466,
      "learning_rate": 5.948341470359533e-05,
      "loss": 2.0091,
      "step": 383
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3841445994669495,
      "learning_rate": 5.947930225888572e-05,
      "loss": 2.0185,
      "step": 384
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4697890549162815,
      "learning_rate": 5.947517365307817e-05,
      "loss": 2.0964,
      "step": 385
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3009999380697674,
      "learning_rate": 5.9471028888436064e-05,
      "loss": 2.0022,
      "step": 386
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3289840267236195,
      "learning_rate": 5.946686796723164e-05,
      "loss": 1.9273,
      "step": 387
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4179541583747663,
      "learning_rate": 5.946269089174598e-05,
      "loss": 2.0726,
      "step": 388
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.361296041013931,
      "learning_rate": 5.945849766426905e-05,
      "loss": 2.0269,
      "step": 389
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3478198857588055,
      "learning_rate": 5.9454288287099654e-05,
      "loss": 2.1589,
      "step": 390
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.499884148175255,
      "learning_rate": 5.9450062762545434e-05,
      "loss": 2.1526,
      "step": 391
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2855493270267841,
      "learning_rate": 5.944582109292292e-05,
      "loss": 2.0081,
      "step": 392
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4386759138032323,
      "learning_rate": 5.9441563280557454e-05,
      "loss": 2.1333,
      "step": 393
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2437232690169742,
      "learning_rate": 5.943728932778327e-05,
      "loss": 2.0795,
      "step": 394
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2311544047266574,
      "learning_rate": 5.94329992369434e-05,
      "loss": 2.0227,
      "step": 395
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.257108775482681,
      "learning_rate": 5.942869301038978e-05,
      "loss": 1.9556,
      "step": 396
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3445334842903762,
      "learning_rate": 5.9424370650483144e-05,
      "loss": 2.0341,
      "step": 397
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3935461821505348,
      "learning_rate": 5.9420032159593094e-05,
      "loss": 2.136,
      "step": 398
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3963936411859232,
      "learning_rate": 5.9415677540098075e-05,
      "loss": 1.9198,
      "step": 399
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3818086129433709,
      "learning_rate": 5.941130679438536e-05,
      "loss": 2.0722,
      "step": 400
    },
    {
      "epoch": 0.46,
      "eval_blimp_filtered_avg": 0.7141791044776119,
      "eval_blimp_filtered_std": 0.0049666821180570295,
      "step": 400
    },
    {
      "epoch": 0.46,
      "eval_blimp_supplement_avg": 0.8232758620689655,
      "eval_blimp_supplement_std": 0.016519166445089422,
      "step": 400
    },
    {
      "epoch": 0.46,
      "eval_vqa_filtered_avg": 0.51,
      "eval_vqa_filtered_std": 0.05024183937956912,
      "step": 400
    },
    {
      "epoch": 0.46,
      "eval_winoground_filtered_avg": 0.66,
      "eval_winoground_filtered_std": 0.04760952285695238,
      "step": 400
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3664326608613617,
      "learning_rate": 5.9406919924851085e-05,
      "loss": 2.0931,
      "step": 401
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6389990863509265,
      "learning_rate": 5.94025169339002e-05,
      "loss": 2.1406,
      "step": 402
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5021952349319885,
      "learning_rate": 5.939809782394652e-05,
      "loss": 2.0638,
      "step": 403
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3487195815944508,
      "learning_rate": 5.9393662597412666e-05,
      "loss": 1.9429,
      "step": 404
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4016116341852694,
      "learning_rate": 5.938921125673013e-05,
      "loss": 2.0495,
      "step": 405
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3159548887852606,
      "learning_rate": 5.938474380433921e-05,
      "loss": 2.009,
      "step": 406
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2851780830573618,
      "learning_rate": 5.938026024268904e-05,
      "loss": 1.9012,
      "step": 407
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.427424856865787,
      "learning_rate": 5.93757605742376e-05,
      "loss": 2.094,
      "step": 408
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2669127438648355,
      "learning_rate": 5.937124480145169e-05,
      "loss": 2.0006,
      "step": 409
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2433808324073934,
      "learning_rate": 5.936671292680694e-05,
      "loss": 1.961,
      "step": 410
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2838708504224463,
      "learning_rate": 5.936216495278781e-05,
      "loss": 2.0752,
      "step": 411
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3889844798020348,
      "learning_rate": 5.935760088188756e-05,
      "loss": 2.142,
      "step": 412
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3479431294154633,
      "learning_rate": 5.935302071660833e-05,
      "loss": 1.9895,
      "step": 413
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3376116311768387,
      "learning_rate": 5.934842445946104e-05,
      "loss": 2.021,
      "step": 414
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4172659003899617,
      "learning_rate": 5.934381211296545e-05,
      "loss": 1.905,
      "step": 415
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3148149065122794,
      "learning_rate": 5.933918367965011e-05,
      "loss": 1.9516,
      "step": 416
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3117842370020287,
      "learning_rate": 5.933453916205243e-05,
      "loss": 1.9978,
      "step": 417
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.408688583907522,
      "learning_rate": 5.932987856271862e-05,
      "loss": 2.0342,
      "step": 418
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3187624699260843,
      "learning_rate": 5.932520188420371e-05,
      "loss": 2.0285,
      "step": 419
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.391764204310888,
      "learning_rate": 5.932050912907153e-05,
      "loss": 1.8797,
      "step": 420
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3639182332949995,
      "learning_rate": 5.9315800299894745e-05,
      "loss": 1.9801,
      "step": 421
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.612189290565815,
      "learning_rate": 5.9311075399254814e-05,
      "loss": 2.1027,
      "step": 422
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3728622426983028,
      "learning_rate": 5.9306334429742015e-05,
      "loss": 2.0547,
      "step": 423
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.319200270153671,
      "learning_rate": 5.9301577393955456e-05,
      "loss": 1.9235,
      "step": 424
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2360352386145184,
      "learning_rate": 5.9296804294503e-05,
      "loss": 1.9851,
      "step": 425
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2758915549865493,
      "learning_rate": 5.929201513400136e-05,
      "loss": 1.9657,
      "step": 426
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4220800031001288,
      "learning_rate": 5.9287209915076054e-05,
      "loss": 2.1046,
      "step": 427
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2664065534729447,
      "learning_rate": 5.928238864036139e-05,
      "loss": 1.8752,
      "step": 428
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.324926207749299,
      "learning_rate": 5.927755131250046e-05,
      "loss": 2.0354,
      "step": 429
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3101246498813304,
      "learning_rate": 5.927269793414519e-05,
      "loss": 1.9993,
      "step": 430
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2371929684955825,
      "learning_rate": 5.926782850795629e-05,
      "loss": 1.8299,
      "step": 431
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3884675897313525,
      "learning_rate": 5.926294303660328e-05,
      "loss": 1.8519,
      "step": 432
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5209720837804508,
      "learning_rate": 5.9258041522764455e-05,
      "loss": 1.9493,
      "step": 433
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6390040959206615,
      "learning_rate": 5.9253123969126905e-05,
      "loss": 2.1591,
      "step": 434
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5683034596292016,
      "learning_rate": 5.924819037838655e-05,
      "loss": 2.1708,
      "step": 435
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5041761191760037,
      "learning_rate": 5.9243240753248055e-05,
      "loss": 1.8142,
      "step": 436
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3554413462068233,
      "learning_rate": 5.92382750964249e-05,
      "loss": 2.031,
      "step": 437
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2642274075558797,
      "learning_rate": 5.923329341063935e-05,
      "loss": 1.9711,
      "step": 438
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2800722591281017,
      "learning_rate": 5.922829569862246e-05,
      "loss": 2.0095,
      "step": 439
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3482057647097374,
      "learning_rate": 5.922328196311407e-05,
      "loss": 1.9507,
      "step": 440
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3798971640903621,
      "learning_rate": 5.9218252206862804e-05,
      "loss": 2.0352,
      "step": 441
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.197611383352257,
      "learning_rate": 5.921320643262605e-05,
      "loss": 1.8756,
      "step": 442
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2779165292464985,
      "learning_rate": 5.920814464317001e-05,
      "loss": 1.9404,
      "step": 443
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2882211176288478,
      "learning_rate": 5.9203066841269657e-05,
      "loss": 1.8659,
      "step": 444
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3500219380753387,
      "learning_rate": 5.9197973029708716e-05,
      "loss": 1.916,
      "step": 445
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4742140519009204,
      "learning_rate": 5.9192863211279717e-05,
      "loss": 1.8983,
      "step": 446
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4340256756080196,
      "learning_rate": 5.918773738878397e-05,
      "loss": 1.8195,
      "step": 447
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7996121199815622,
      "learning_rate": 5.918259556503153e-05,
      "loss": 2.0655,
      "step": 448
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5147225370601338,
      "learning_rate": 5.9177437742841235e-05,
      "loss": 1.9117,
      "step": 449
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4342709151319117,
      "learning_rate": 5.917226392504071e-05,
      "loss": 1.8398,
      "step": 450
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3563654917926247,
      "learning_rate": 5.9167074114466345e-05,
      "loss": 1.8365,
      "step": 451
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2711327439473687,
      "learning_rate": 5.916186831396327e-05,
      "loss": 1.8196,
      "step": 452
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3695990160992515,
      "learning_rate": 5.9156646526385414e-05,
      "loss": 2.097,
      "step": 453
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4871897639069487,
      "learning_rate": 5.915140875459546e-05,
      "loss": 2.1817,
      "step": 454
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3533128335959745,
      "learning_rate": 5.914615500146484e-05,
      "loss": 1.8932,
      "step": 455
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3786970852958933,
      "learning_rate": 5.9140885269873764e-05,
      "loss": 1.9961,
      "step": 456
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2299329994413655,
      "learning_rate": 5.9135599562711194e-05,
      "loss": 1.8255,
      "step": 457
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4617317927107418,
      "learning_rate": 5.9130297882874866e-05,
      "loss": 2.0895,
      "step": 458
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4449019470914999,
      "learning_rate": 5.912498023327123e-05,
      "loss": 1.9115,
      "step": 459
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3357834868075174,
      "learning_rate": 5.9119646616815546e-05,
      "loss": 1.8061,
      "step": 460
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4730246524684811,
      "learning_rate": 5.911429703643179e-05,
      "loss": 2.0277,
      "step": 461
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.538421049731831,
      "learning_rate": 5.9108931495052706e-05,
      "loss": 2.0593,
      "step": 462
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3745664785574534,
      "learning_rate": 5.910354999561977e-05,
      "loss": 1.9005,
      "step": 463
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2976257384929781,
      "learning_rate": 5.909815254108324e-05,
      "loss": 1.8141,
      "step": 464
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5862713082762674,
      "learning_rate": 5.909273913440208e-05,
      "loss": 1.8617,
      "step": 465
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3693871149097732,
      "learning_rate": 5.908730977854404e-05,
      "loss": 1.7841,
      "step": 466
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4762336758351962,
      "learning_rate": 5.908186447648557e-05,
      "loss": 1.8729,
      "step": 467
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.589952231606026,
      "learning_rate": 5.907640323121191e-05,
      "loss": 2.0086,
      "step": 468
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4185918488086209,
      "learning_rate": 5.9070926045717e-05,
      "loss": 1.8511,
      "step": 469
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3434028741036654,
      "learning_rate": 5.906543292300353e-05,
      "loss": 1.9407,
      "step": 470
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4175041578012009,
      "learning_rate": 5.905992386608293e-05,
      "loss": 1.9674,
      "step": 471
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2745484928915343,
      "learning_rate": 5.9054398877975396e-05,
      "loss": 1.7922,
      "step": 472
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.379255662570555,
      "learning_rate": 5.9048857961709796e-05,
      "loss": 1.952,
      "step": 473
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4046762118897531,
      "learning_rate": 5.9043301120323775e-05,
      "loss": 1.8888,
      "step": 474
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.463902247003978,
      "learning_rate": 5.90377283568637e-05,
      "loss": 1.8933,
      "step": 475
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4342756335577491,
      "learning_rate": 5.903213967438465e-05,
      "loss": 1.855,
      "step": 476
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6218841970020996,
      "learning_rate": 5.9026535075950446e-05,
      "loss": 1.957,
      "step": 477
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5622598410878146,
      "learning_rate": 5.9020914564633654e-05,
      "loss": 1.9592,
      "step": 478
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4932097073700579,
      "learning_rate": 5.901527814351551e-05,
      "loss": 1.8888,
      "step": 479
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5451525307641307,
      "learning_rate": 5.900962581568604e-05,
      "loss": 2.0006,
      "step": 480
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5982169313648291,
      "learning_rate": 5.9003957584243925e-05,
      "loss": 2.1902,
      "step": 481
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.469933988078718,
      "learning_rate": 5.89982734522966e-05,
      "loss": 1.9258,
      "step": 482
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3657720917816105,
      "learning_rate": 5.899257342296021e-05,
      "loss": 1.9524,
      "step": 483
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.377547212582918,
      "learning_rate": 5.898685749935964e-05,
      "loss": 1.8658,
      "step": 484
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4119225897635228,
      "learning_rate": 5.898112568462844e-05,
      "loss": 2.0027,
      "step": 485
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4947033217132102,
      "learning_rate": 5.89753779819089e-05,
      "loss": 1.7129,
      "step": 486
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4351046081217016,
      "learning_rate": 5.896961439435202e-05,
      "loss": 1.9381,
      "step": 487
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.370930230748793,
      "learning_rate": 5.8963834925117505e-05,
      "loss": 1.8502,
      "step": 488
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4896212201473686,
      "learning_rate": 5.895803957737377e-05,
      "loss": 1.9041,
      "step": 489
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.478135573090294,
      "learning_rate": 5.895222835429792e-05,
      "loss": 1.937,
      "step": 490
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.473050998004548,
      "learning_rate": 5.89464012590758e-05,
      "loss": 1.8888,
      "step": 491
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5392961747415799,
      "learning_rate": 5.89405582949019e-05,
      "loss": 1.8514,
      "step": 492
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.512876492590634,
      "learning_rate": 5.8934699464979455e-05,
      "loss": 1.9536,
      "step": 493
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4888679785874965,
      "learning_rate": 5.892882477252039e-05,
      "loss": 2.0268,
      "step": 494
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2285419339098338,
      "learning_rate": 5.8922934220745315e-05,
      "loss": 1.7412,
      "step": 495
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3311655152763262,
      "learning_rate": 5.891702781288354e-05,
      "loss": 1.8944,
      "step": 496
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3631941146158584,
      "learning_rate": 5.8911105552173066e-05,
      "loss": 1.9396,
      "step": 497
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3395592998273356,
      "learning_rate": 5.890516744186058e-05,
      "loss": 1.6526,
      "step": 498
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2900116843943075,
      "learning_rate": 5.889921348520147e-05,
      "loss": 1.9024,
      "step": 499
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3589009169576292,
      "learning_rate": 5.88932436854598e-05,
      "loss": 1.8827,
      "step": 500
    },
    {
      "epoch": 0.57,
      "eval_blimp_filtered_avg": 0.7134328358208956,
      "eval_blimp_filtered_std": 0.004977475699452633,
      "step": 500
    },
    {
      "epoch": 0.57,
      "eval_blimp_supplement_avg": 0.8211206896551724,
      "eval_blimp_supplement_std": 0.01672984630252136,
      "step": 500
    },
    {
      "epoch": 0.57,
      "eval_vqa_filtered_avg": 0.54,
      "eval_vqa_filtered_std": 0.05009082659620333,
      "step": 500
    },
    {
      "epoch": 0.57,
      "eval_winoground_filtered_avg": 0.64,
      "eval_winoground_filtered_std": 0.04824181513244218,
      "step": 500
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.379220963164939,
      "learning_rate": 5.888725804590833e-05,
      "loss": 1.9538,
      "step": 501
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6364670332501579,
      "learning_rate": 5.88812565698285e-05,
      "loss": 1.7344,
      "step": 502
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5219396042286801,
      "learning_rate": 5.887523926051043e-05,
      "loss": 1.9693,
      "step": 503
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6871698950668061,
      "learning_rate": 5.886920612125291e-05,
      "loss": 2.0059,
      "step": 504
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.511460121712087,
      "learning_rate": 5.8863157155363425e-05,
      "loss": 1.9434,
      "step": 505
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5056007129435558,
      "learning_rate": 5.885709236615813e-05,
      "loss": 1.9376,
      "step": 506
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4941972044014624,
      "learning_rate": 5.885101175696186e-05,
      "loss": 1.9266,
      "step": 507
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4810081448380263,
      "learning_rate": 5.8844915331108104e-05,
      "loss": 1.9295,
      "step": 508
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4804524824388936,
      "learning_rate": 5.883880309193904e-05,
      "loss": 1.9736,
      "step": 509
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3586852338565012,
      "learning_rate": 5.8832675042805516e-05,
      "loss": 1.9308,
      "step": 510
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3952009628972075,
      "learning_rate": 5.882653118706705e-05,
      "loss": 1.8492,
      "step": 511
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4665988897250795,
      "learning_rate": 5.8820371528091796e-05,
      "loss": 1.9595,
      "step": 512
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3702410618000318,
      "learning_rate": 5.8814196069256605e-05,
      "loss": 1.804,
      "step": 513
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.393405962229494,
      "learning_rate": 5.880800481394697e-05,
      "loss": 1.9046,
      "step": 514
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4405941720497115,
      "learning_rate": 5.880179776555706e-05,
      "loss": 1.8994,
      "step": 515
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.606668375070894,
      "learning_rate": 5.879557492748969e-05,
      "loss": 1.9747,
      "step": 516
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3219812532840705,
      "learning_rate": 5.878933630315633e-05,
      "loss": 1.7787,
      "step": 517
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.514461180123167,
      "learning_rate": 5.878308189597713e-05,
      "loss": 2.0403,
      "step": 518
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6087834966442474,
      "learning_rate": 5.877681170938084e-05,
      "loss": 1.838,
      "step": 519
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7183531189149595,
      "learning_rate": 5.877052574680491e-05,
      "loss": 1.9835,
      "step": 520
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4772559896235713,
      "learning_rate": 5.8764224011695426e-05,
      "loss": 1.8148,
      "step": 521
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4721227392500194,
      "learning_rate": 5.8757906507507114e-05,
      "loss": 1.6638,
      "step": 522
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5191760985419345,
      "learning_rate": 5.8751573237703325e-05,
      "loss": 1.8198,
      "step": 523
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.618920344070238,
      "learning_rate": 5.8745224205756094e-05,
      "loss": 1.8783,
      "step": 524
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4463581828468874,
      "learning_rate": 5.873885941514608e-05,
      "loss": 1.6541,
      "step": 525
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5607246354772613,
      "learning_rate": 5.873247886936257e-05,
      "loss": 1.911,
      "step": 526
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6935421767410965,
      "learning_rate": 5.87260825719035e-05,
      "loss": 1.8775,
      "step": 527
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7743862949349944,
      "learning_rate": 5.8719670526275425e-05,
      "loss": 1.9755,
      "step": 528
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5452372520337663,
      "learning_rate": 5.8713242735993576e-05,
      "loss": 1.9264,
      "step": 529
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.553267096578831,
      "learning_rate": 5.870679920458176e-05,
      "loss": 1.819,
      "step": 530
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.385543992104837,
      "learning_rate": 5.8700339935572465e-05,
      "loss": 1.7889,
      "step": 531
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4265658025609251,
      "learning_rate": 5.869386493250675e-05,
      "loss": 1.7971,
      "step": 532
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3306100965848433,
      "learning_rate": 5.868737419893435e-05,
      "loss": 1.7378,
      "step": 533
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3196778042200032,
      "learning_rate": 5.868086773841361e-05,
      "loss": 1.7678,
      "step": 534
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4506652706783074,
      "learning_rate": 5.8674345554511486e-05,
      "loss": 1.7899,
      "step": 535
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4360437998685438,
      "learning_rate": 5.866780765080355e-05,
      "loss": 1.8561,
      "step": 536
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5468540499214871,
      "learning_rate": 5.866125403087401e-05,
      "loss": 1.7967,
      "step": 537
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5188092887607592,
      "learning_rate": 5.865468469831569e-05,
      "loss": 1.7054,
      "step": 538
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.560627719293142,
      "learning_rate": 5.864809965673e-05,
      "loss": 1.7627,
      "step": 539
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5667344019171625,
      "learning_rate": 5.8641498909727004e-05,
      "loss": 1.816,
      "step": 540
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6440585221577522,
      "learning_rate": 5.863488246092534e-05,
      "loss": 1.7869,
      "step": 541
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5628048190250383,
      "learning_rate": 5.862825031395226e-05,
      "loss": 1.817,
      "step": 542
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.576381370082841,
      "learning_rate": 5.862160247244364e-05,
      "loss": 1.7963,
      "step": 543
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5914240244902882,
      "learning_rate": 5.861493894004395e-05,
      "loss": 1.8463,
      "step": 544
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6349847233148036,
      "learning_rate": 5.8608259720406265e-05,
      "loss": 1.8756,
      "step": 545
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6414205175833703,
      "learning_rate": 5.860156481719225e-05,
      "loss": 1.7641,
      "step": 546
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5308504995235674,
      "learning_rate": 5.8594854234072166e-05,
      "loss": 1.7998,
      "step": 547
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6019313511267619,
      "learning_rate": 5.8588127974724895e-05,
      "loss": 1.8833,
      "step": 548
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.598795433927205,
      "learning_rate": 5.858138604283789e-05,
      "loss": 1.9328,
      "step": 549
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.481241913391856,
      "learning_rate": 5.8574628442107203e-05,
      "loss": 1.7282,
      "step": 550
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4827933105733793,
      "learning_rate": 5.856785517623747e-05,
      "loss": 1.7229,
      "step": 551
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5530975558961422,
      "learning_rate": 5.856106624894194e-05,
      "loss": 1.7908,
      "step": 552
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4503807383469403,
      "learning_rate": 5.8554261663942394e-05,
      "loss": 1.8634,
      "step": 553
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5518358690160396,
      "learning_rate": 5.8547441424969264e-05,
      "loss": 1.8909,
      "step": 554
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5212957460001622,
      "learning_rate": 5.8540605535761506e-05,
      "loss": 1.8491,
      "step": 555
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4985582185512958,
      "learning_rate": 5.853375400006671e-05,
      "loss": 1.7447,
      "step": 556
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4868473598750502,
      "learning_rate": 5.8526886821640985e-05,
      "loss": 1.7318,
      "step": 557
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.528492874885668,
      "learning_rate": 5.8520004004249056e-05,
      "loss": 1.6876,
      "step": 558
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6853714965880087,
      "learning_rate": 5.851310555166422e-05,
      "loss": 1.8257,
      "step": 559
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6845398956102764,
      "learning_rate": 5.8506191467668325e-05,
      "loss": 1.704,
      "step": 560
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.717000621939236,
      "learning_rate": 5.8499261756051794e-05,
      "loss": 1.893,
      "step": 561
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8287882863677245,
      "learning_rate": 5.849231642061364e-05,
      "loss": 2.0222,
      "step": 562
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5263299683963492,
      "learning_rate": 5.8485355465161415e-05,
      "loss": 1.8012,
      "step": 563
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4711168331028974,
      "learning_rate": 5.847837889351124e-05,
      "loss": 1.8187,
      "step": 564
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4144258020009974,
      "learning_rate": 5.8471386709487806e-05,
      "loss": 1.676,
      "step": 565
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3995946296850141,
      "learning_rate": 5.8464378916924355e-05,
      "loss": 1.8697,
      "step": 566
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4508229955130203,
      "learning_rate": 5.8457355519662686e-05,
      "loss": 1.7251,
      "step": 567
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4460340864253767,
      "learning_rate": 5.845031652155316e-05,
      "loss": 1.9053,
      "step": 568
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4567958717718683,
      "learning_rate": 5.844326192645468e-05,
      "loss": 1.7029,
      "step": 569
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6652460958939097,
      "learning_rate": 5.843619173823472e-05,
      "loss": 1.7844,
      "step": 570
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5791193440825284,
      "learning_rate": 5.8429105960769265e-05,
      "loss": 1.8776,
      "step": 571
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.775479429957605,
      "learning_rate": 5.842200459794289e-05,
      "loss": 1.8979,
      "step": 572
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.615240158044289,
      "learning_rate": 5.841488765364868e-05,
      "loss": 1.6759,
      "step": 573
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5662076032860288,
      "learning_rate": 5.8407755131788274e-05,
      "loss": 1.7542,
      "step": 574
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.641247420293409,
      "learning_rate": 5.8400607036271865e-05,
      "loss": 1.7681,
      "step": 575
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5864634681787138,
      "learning_rate": 5.8393443371018164e-05,
      "loss": 1.723,
      "step": 576
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.522625085001753,
      "learning_rate": 5.838626413995442e-05,
      "loss": 1.6593,
      "step": 577
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.494925179754713,
      "learning_rate": 5.837906934701642e-05,
      "loss": 1.6621,
      "step": 578
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7582175906619573,
      "learning_rate": 5.837185899614849e-05,
      "loss": 1.8487,
      "step": 579
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4946490679328914,
      "learning_rate": 5.836463309130348e-05,
      "loss": 1.7266,
      "step": 580
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4574843830607183,
      "learning_rate": 5.8357391636442755e-05,
      "loss": 1.7659,
      "step": 581
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5016478398629522,
      "learning_rate": 5.835013463553622e-05,
      "loss": 1.6866,
      "step": 582
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.515339077377514,
      "learning_rate": 5.834286209256229e-05,
      "loss": 1.8345,
      "step": 583
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6322075903985516,
      "learning_rate": 5.833557401150792e-05,
      "loss": 1.8147,
      "step": 584
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4575164415734274,
      "learning_rate": 5.832827039636855e-05,
      "loss": 1.6829,
      "step": 585
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6242985309701936,
      "learning_rate": 5.832095125114818e-05,
      "loss": 1.7009,
      "step": 586
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8438585295985415,
      "learning_rate": 5.831361657985928e-05,
      "loss": 1.7477,
      "step": 587
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7952044172563602,
      "learning_rate": 5.8306266386522874e-05,
      "loss": 1.8573,
      "step": 588
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6275449606153793,
      "learning_rate": 5.829890067516846e-05,
      "loss": 1.8968,
      "step": 589
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6759355957905702,
      "learning_rate": 5.829151944983406e-05,
      "loss": 1.8283,
      "step": 590
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5962745510647283,
      "learning_rate": 5.828412271456619e-05,
      "loss": 1.8305,
      "step": 591
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5002393982482791,
      "learning_rate": 5.8276710473419885e-05,
      "loss": 1.7641,
      "step": 592
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4354615434553755,
      "learning_rate": 5.826928273045867e-05,
      "loss": 1.5515,
      "step": 593
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4317207441261877,
      "learning_rate": 5.826183948975458e-05,
      "loss": 1.7513,
      "step": 594
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.530608285497835,
      "learning_rate": 5.8254380755388124e-05,
      "loss": 1.8048,
      "step": 595
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.490453148321634,
      "learning_rate": 5.824690653144833e-05,
      "loss": 1.6398,
      "step": 596
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5989651281982704,
      "learning_rate": 5.823941682203269e-05,
      "loss": 1.8337,
      "step": 597
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7374785429243094,
      "learning_rate": 5.8231911631247215e-05,
      "loss": 1.8362,
      "step": 598
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.836490682916201,
      "learning_rate": 5.822439096320638e-05,
      "loss": 1.7991,
      "step": 599
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.863483992226924,
      "learning_rate": 5.8216854822033165e-05,
      "loss": 1.9082,
      "step": 600
    },
    {
      "epoch": 0.69,
      "eval_blimp_filtered_avg": 0.7102985074626865,
      "eval_blimp_filtered_std": 0.005028054039077976,
      "step": 600
    },
    {
      "epoch": 0.69,
      "eval_blimp_supplement_avg": 0.8362068965517241,
      "eval_blimp_supplement_std": 0.016284784318052175,
      "step": 600
    },
    {
      "epoch": 0.69,
      "eval_vqa_filtered_avg": 0.52,
      "eval_vqa_filtered_std": 0.05021167315686779,
      "step": 600
    },
    {
      "epoch": 0.69,
      "eval_winoground_filtered_avg": 0.64,
      "eval_winoground_filtered_std": 0.04824181513244218,
      "step": 600
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7430013794191175,
      "learning_rate": 5.8209303211859013e-05,
      "loss": 1.8409,
      "step": 601
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5744116607392427,
      "learning_rate": 5.820173613682385e-05,
      "loss": 1.7111,
      "step": 602
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5082416664416878,
      "learning_rate": 5.8194153601076105e-05,
      "loss": 1.7659,
      "step": 603
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4543650351127433,
      "learning_rate": 5.818655560877264e-05,
      "loss": 1.6711,
      "step": 604
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4131643401493845,
      "learning_rate": 5.817894216407885e-05,
      "loss": 1.6645,
      "step": 605
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4438371973477644,
      "learning_rate": 5.817131327116853e-05,
      "loss": 1.7453,
      "step": 606
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5104697041439188,
      "learning_rate": 5.8163668934224e-05,
      "loss": 1.707,
      "step": 607
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5218371834996736,
      "learning_rate": 5.8156009157436016e-05,
      "loss": 1.7793,
      "step": 608
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5153907013889139,
      "learning_rate": 5.814833394500381e-05,
      "loss": 1.5942,
      "step": 609
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.612049179771305,
      "learning_rate": 5.814064330113509e-05,
      "loss": 1.6333,
      "step": 610
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.714764261734268,
      "learning_rate": 5.8132937230045985e-05,
      "loss": 1.7595,
      "step": 611
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.743925227431969,
      "learning_rate": 5.812521573596113e-05,
      "loss": 1.6356,
      "step": 612
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.580374296603772,
      "learning_rate": 5.811747882311358e-05,
      "loss": 1.6131,
      "step": 613
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6329109799187322,
      "learning_rate": 5.8109726495744845e-05,
      "loss": 1.7222,
      "step": 614
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5392780831493122,
      "learning_rate": 5.81019587581049e-05,
      "loss": 1.5198,
      "step": 615
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5364521927705237,
      "learning_rate": 5.8094175614452176e-05,
      "loss": 1.6532,
      "step": 616
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.593257320215041,
      "learning_rate": 5.808637706905351e-05,
      "loss": 1.6483,
      "step": 617
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6178852360497715,
      "learning_rate": 5.8078563126184235e-05,
      "loss": 1.7794,
      "step": 618
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5119030123954682,
      "learning_rate": 5.8070733790128084e-05,
      "loss": 1.6192,
      "step": 619
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7212383163712792,
      "learning_rate": 5.8062889065177247e-05,
      "loss": 1.7382,
      "step": 620
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.650673489680751,
      "learning_rate": 5.805502895563234e-05,
      "loss": 1.7961,
      "step": 621
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9182761900223195,
      "learning_rate": 5.804715346580244e-05,
      "loss": 1.6854,
      "step": 622
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5673829194478335,
      "learning_rate": 5.8039262600005005e-05,
      "loss": 1.8138,
      "step": 623
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7408023005687794,
      "learning_rate": 5.803135636256599e-05,
      "loss": 1.7906,
      "step": 624
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6631512895918787,
      "learning_rate": 5.802343475781971e-05,
      "loss": 1.7714,
      "step": 625
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.458740877319592,
      "learning_rate": 5.8015497790108944e-05,
      "loss": 1.6118,
      "step": 626
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5226711072440202,
      "learning_rate": 5.800754546378489e-05,
      "loss": 1.7187,
      "step": 627
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.717290327580548,
      "learning_rate": 5.799957778320716e-05,
      "loss": 1.733,
      "step": 628
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6969428566423925,
      "learning_rate": 5.799159475274378e-05,
      "loss": 1.6879,
      "step": 629
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7963013411587778,
      "learning_rate": 5.7983596376771186e-05,
      "loss": 1.6802,
      "step": 630
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6412751354214858,
      "learning_rate": 5.797558265967425e-05,
      "loss": 1.7439,
      "step": 631
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.497637607165799,
      "learning_rate": 5.796755360584624e-05,
      "loss": 1.5714,
      "step": 632
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7385500275249575,
      "learning_rate": 5.795950921968882e-05,
      "loss": 1.737,
      "step": 633
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.559337607646749,
      "learning_rate": 5.7951449505612075e-05,
      "loss": 1.5983,
      "step": 634
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4727531709525619,
      "learning_rate": 5.794337446803449e-05,
      "loss": 1.5709,
      "step": 635
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6326188065511105,
      "learning_rate": 5.793528411138296e-05,
      "loss": 1.7166,
      "step": 636
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.425142637233881,
      "learning_rate": 5.7927178440092756e-05,
      "loss": 1.4298,
      "step": 637
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.505911439630627,
      "learning_rate": 5.791905745860756e-05,
      "loss": 1.5828,
      "step": 638
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5749915571283277,
      "learning_rate": 5.791092117137944e-05,
      "loss": 1.6709,
      "step": 639
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.794878482173177,
      "learning_rate": 5.790276958286886e-05,
      "loss": 1.7698,
      "step": 640
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6532117064624086,
      "learning_rate": 5.789460269754468e-05,
      "loss": 1.6345,
      "step": 641
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6518182209441903,
      "learning_rate": 5.7886420519884134e-05,
      "loss": 1.6829,
      "step": 642
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6976627105095925,
      "learning_rate": 5.7878223054372834e-05,
      "loss": 1.7142,
      "step": 643
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7373185453870326,
      "learning_rate": 5.787001030550479e-05,
      "loss": 1.8046,
      "step": 644
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5998818685940552,
      "learning_rate": 5.7861782277782374e-05,
      "loss": 1.5406,
      "step": 645
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.716506575171776,
      "learning_rate": 5.7853538975716356e-05,
      "loss": 1.5766,
      "step": 646
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5119783401317384,
      "learning_rate": 5.7845280403825856e-05,
      "loss": 1.4868,
      "step": 647
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.497912011097032,
      "learning_rate": 5.783700656663839e-05,
      "loss": 1.4898,
      "step": 648
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8207240811654104,
      "learning_rate": 5.782871746868981e-05,
      "loss": 1.6679,
      "step": 649
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6024105602386376,
      "learning_rate": 5.782041311452436e-05,
      "loss": 1.6401,
      "step": 650
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7487095712193634,
      "learning_rate": 5.781209350869464e-05,
      "loss": 1.601,
      "step": 651
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5976865880511528,
      "learning_rate": 5.780375865576161e-05,
      "loss": 1.6271,
      "step": 652
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8063776507617555,
      "learning_rate": 5.7795408560294596e-05,
      "loss": 1.7305,
      "step": 653
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.26946164186439,
      "learning_rate": 5.778704322687127e-05,
      "loss": 1.7292,
      "step": 654
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5797222119353855,
      "learning_rate": 5.777866266007767e-05,
      "loss": 1.5256,
      "step": 655
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5921850043270243,
      "learning_rate": 5.7770266864508164e-05,
      "loss": 1.6825,
      "step": 656
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5868714998473963,
      "learning_rate": 5.7761855844765484e-05,
      "loss": 1.7949,
      "step": 657
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.456389686982743,
      "learning_rate": 5.7753429605460724e-05,
      "loss": 1.6171,
      "step": 658
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4415666087406913,
      "learning_rate": 5.774498815121328e-05,
      "loss": 1.4847,
      "step": 659
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6152171741939656,
      "learning_rate": 5.773653148665093e-05,
      "loss": 1.6331,
      "step": 660
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5775013880187458,
      "learning_rate": 5.772805961640977e-05,
      "loss": 1.637,
      "step": 661
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5518600739163702,
      "learning_rate": 5.771957254513423e-05,
      "loss": 1.609,
      "step": 662
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7947592284839593,
      "learning_rate": 5.7711070277477085e-05,
      "loss": 1.6771,
      "step": 663
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.660992267626647,
      "learning_rate": 5.770255281809944e-05,
      "loss": 1.6441,
      "step": 664
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7619666701343981,
      "learning_rate": 5.769402017167071e-05,
      "loss": 1.6643,
      "step": 665
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7132020498947365,
      "learning_rate": 5.7685472342868656e-05,
      "loss": 1.6235,
      "step": 666
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7223802633544651,
      "learning_rate": 5.767690933637936e-05,
      "loss": 1.7005,
      "step": 667
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5803151234840345,
      "learning_rate": 5.766833115689722e-05,
      "loss": 1.5253,
      "step": 668
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.41228116841019,
      "learning_rate": 5.765973780912494e-05,
      "loss": 1.5039,
      "step": 669
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7029229700509836,
      "learning_rate": 5.765112929777355e-05,
      "loss": 1.8184,
      "step": 670
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.681443075361966,
      "learning_rate": 5.7642505627562424e-05,
      "loss": 1.6926,
      "step": 671
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.823429810014774,
      "learning_rate": 5.763386680321919e-05,
      "loss": 1.7263,
      "step": 672
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5504317342764522,
      "learning_rate": 5.762521282947983e-05,
      "loss": 1.6658,
      "step": 673
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.521913289184347,
      "learning_rate": 5.761654371108859e-05,
      "loss": 1.4719,
      "step": 674
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.578479594016252,
      "learning_rate": 5.760785945279805e-05,
      "loss": 1.531,
      "step": 675
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6311102747152653,
      "learning_rate": 5.75991600593691e-05,
      "loss": 1.5219,
      "step": 676
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6054884597081973,
      "learning_rate": 5.7590445535570875e-05,
      "loss": 1.4956,
      "step": 677
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6551405922780735,
      "learning_rate": 5.758171588618087e-05,
      "loss": 1.5852,
      "step": 678
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.798003373607632,
      "learning_rate": 5.757297111598482e-05,
      "loss": 1.6767,
      "step": 679
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8148144685629806,
      "learning_rate": 5.7564211229776766e-05,
      "loss": 1.6847,
      "step": 680
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8499652276375556,
      "learning_rate": 5.755543623235906e-05,
      "loss": 1.6875,
      "step": 681
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.747238532251136,
      "learning_rate": 5.754664612854229e-05,
      "loss": 1.6937,
      "step": 682
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5585925483783356,
      "learning_rate": 5.753784092314538e-05,
      "loss": 1.6086,
      "step": 683
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6195789740957605,
      "learning_rate": 5.752902062099549e-05,
      "loss": 1.7246,
      "step": 684
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.830967218665342,
      "learning_rate": 5.7520185226928086e-05,
      "loss": 1.8162,
      "step": 685
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4361537539799465,
      "learning_rate": 5.7511334745786875e-05,
      "loss": 1.565,
      "step": 686
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5913102410449749,
      "learning_rate": 5.750246918242387e-05,
      "loss": 1.6448,
      "step": 687
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4309582669910907,
      "learning_rate": 5.749358854169933e-05,
      "loss": 1.3467,
      "step": 688
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.749872440914473,
      "learning_rate": 5.748469282848179e-05,
      "loss": 1.6371,
      "step": 689
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.939417844380686,
      "learning_rate": 5.7475782047648036e-05,
      "loss": 1.6298,
      "step": 690
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8475441908714492,
      "learning_rate": 5.7466856204083126e-05,
      "loss": 1.6723,
      "step": 691
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9103430026677206,
      "learning_rate": 5.745791530268038e-05,
      "loss": 1.5915,
      "step": 692
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7739075845970587,
      "learning_rate": 5.744895934834136e-05,
      "loss": 1.5716,
      "step": 693
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7561490703722453,
      "learning_rate": 5.7439988345975885e-05,
      "loss": 1.6033,
      "step": 694
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8014912024461602,
      "learning_rate": 5.743100230050202e-05,
      "loss": 1.584,
      "step": 695
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7185177018550932,
      "learning_rate": 5.7422001216846096e-05,
      "loss": 1.6293,
      "step": 696
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6251365169995042,
      "learning_rate": 5.741298509994265e-05,
      "loss": 1.6224,
      "step": 697
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.546385087946191,
      "learning_rate": 5.740395395473451e-05,
      "loss": 1.5058,
      "step": 698
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7338116314770655,
      "learning_rate": 5.7394907786172696e-05,
      "loss": 1.5976,
      "step": 699
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6435426965106716,
      "learning_rate": 5.738584659921649e-05,
      "loss": 1.5081,
      "step": 700
    },
    {
      "epoch": 0.8,
      "eval_blimp_filtered_avg": 0.7052238805970149,
      "eval_blimp_filtered_std": 0.005051717071290114,
      "step": 700
    },
    {
      "epoch": 0.8,
      "eval_blimp_supplement_avg": 0.8275862068965517,
      "eval_blimp_supplement_std": 0.01655075764713848,
      "step": 700
    },
    {
      "epoch": 0.8,
      "eval_vqa_filtered_avg": 0.53,
      "eval_vqa_filtered_std": 0.0501613558046592,
      "step": 700
    },
    {
      "epoch": 0.8,
      "eval_winoground_filtered_avg": 0.62,
      "eval_winoground_filtered_std": 0.04878317312145633,
      "step": 700
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9110105630815188,
      "learning_rate": 5.737677039883341e-05,
      "loss": 1.6605,
      "step": 701
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.565261478096994,
      "learning_rate": 5.736767918999919e-05,
      "loss": 1.3805,
      "step": 702
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7969982124845474,
      "learning_rate": 5.7358572977697795e-05,
      "loss": 1.6652,
      "step": 703
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.928873764663935,
      "learning_rate": 5.7349451766921435e-05,
      "loss": 1.7006,
      "step": 704
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.723039588097525,
      "learning_rate": 5.73403155626705e-05,
      "loss": 1.5689,
      "step": 705
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9077719361845202,
      "learning_rate": 5.7331164369953646e-05,
      "loss": 1.4836,
      "step": 706
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4916824240219548,
      "learning_rate": 5.732199819378772e-05,
      "loss": 1.4547,
      "step": 707
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4991371056944331,
      "learning_rate": 5.7312817039197796e-05,
      "loss": 1.5963,
      "step": 708
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6596819609888092,
      "learning_rate": 5.730362091121713e-05,
      "loss": 1.5225,
      "step": 709
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7131121772456006,
      "learning_rate": 5.7294409814887237e-05,
      "loss": 1.6287,
      "step": 710
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.416510789397455,
      "learning_rate": 5.72851837552578e-05,
      "loss": 1.4194,
      "step": 711
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7063426321705,
      "learning_rate": 5.727594273738671e-05,
      "loss": 1.6434,
      "step": 712
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7699655394985176,
      "learning_rate": 5.726668676634006e-05,
      "loss": 1.6316,
      "step": 713
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7648128693820695,
      "learning_rate": 5.725741584719216e-05,
      "loss": 1.6429,
      "step": 714
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7005402154778722,
      "learning_rate": 5.724812998502549e-05,
      "loss": 1.5661,
      "step": 715
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7714022777907252,
      "learning_rate": 5.723882918493073e-05,
      "loss": 1.5859,
      "step": 716
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9492648118743179,
      "learning_rate": 5.7229513452006746e-05,
      "loss": 1.7746,
      "step": 717
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8535964171854078,
      "learning_rate": 5.722018279136061e-05,
      "loss": 1.5977,
      "step": 718
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6886137088939162,
      "learning_rate": 5.721083720810755e-05,
      "loss": 1.4162,
      "step": 719
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.571900103279235,
      "learning_rate": 5.720147670737099e-05,
      "loss": 1.4855,
      "step": 720
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7306377649479803,
      "learning_rate": 5.719210129428253e-05,
      "loss": 1.6876,
      "step": 721
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6235723079942894,
      "learning_rate": 5.718271097398194e-05,
      "loss": 1.524,
      "step": 722
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4738867484527274,
      "learning_rate": 5.717330575161717e-05,
      "loss": 1.4971,
      "step": 723
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5472601800194286,
      "learning_rate": 5.716388563234435e-05,
      "loss": 1.5674,
      "step": 724
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5627280861295387,
      "learning_rate": 5.715445062132774e-05,
      "loss": 1.4324,
      "step": 725
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6459157280214165,
      "learning_rate": 5.71450007237398e-05,
      "loss": 1.6401,
      "step": 726
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6622956726763922,
      "learning_rate": 5.713553594476115e-05,
      "loss": 1.5517,
      "step": 727
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6197802555714365,
      "learning_rate": 5.712605628958053e-05,
      "loss": 1.4761,
      "step": 728
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8832742380599001,
      "learning_rate": 5.711656176339488e-05,
      "loss": 1.622,
      "step": 729
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9349207100875518,
      "learning_rate": 5.710705237140927e-05,
      "loss": 1.6419,
      "step": 730
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.697782032029928,
      "learning_rate": 5.709752811883693e-05,
      "loss": 1.5699,
      "step": 731
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8879021226102382,
      "learning_rate": 5.708798901089922e-05,
      "loss": 1.5787,
      "step": 732
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.740302944013147,
      "learning_rate": 5.707843505282566e-05,
      "loss": 1.4894,
      "step": 733
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7568808189047107,
      "learning_rate": 5.706886624985392e-05,
      "loss": 1.505,
      "step": 734
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6212350911434437,
      "learning_rate": 5.705928260722976e-05,
      "loss": 1.4856,
      "step": 735
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7300002060775852,
      "learning_rate": 5.704968413020715e-05,
      "loss": 1.5787,
      "step": 736
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.763356638847569,
      "learning_rate": 5.7040070824048126e-05,
      "loss": 1.5209,
      "step": 737
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7748074000072023,
      "learning_rate": 5.7030442694022886e-05,
      "loss": 1.6342,
      "step": 738
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7330199947137932,
      "learning_rate": 5.702079974540976e-05,
      "loss": 1.6198,
      "step": 739
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.693008896718195,
      "learning_rate": 5.701114198349518e-05,
      "loss": 1.386,
      "step": 740
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7596256110402329,
      "learning_rate": 5.7001469413573704e-05,
      "loss": 1.4854,
      "step": 741
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.367985481151511,
      "learning_rate": 5.699178204094802e-05,
      "loss": 1.6412,
      "step": 742
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7695093785580216,
      "learning_rate": 5.698207987092893e-05,
      "loss": 1.5202,
      "step": 743
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.694496942974812,
      "learning_rate": 5.697236290883534e-05,
      "loss": 1.5053,
      "step": 744
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5622974180626237,
      "learning_rate": 5.696263115999426e-05,
      "loss": 1.4753,
      "step": 745
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5362810234236015,
      "learning_rate": 5.695288462974082e-05,
      "loss": 1.4492,
      "step": 746
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.770424581927279,
      "learning_rate": 5.694312332341825e-05,
      "loss": 1.6572,
      "step": 747
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7963529728131122,
      "learning_rate": 5.693334724637788e-05,
      "loss": 1.425,
      "step": 748
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.63558512766653,
      "learning_rate": 5.692355640397912e-05,
      "loss": 1.4549,
      "step": 749
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9022600824145552,
      "learning_rate": 5.6913750801589515e-05,
      "loss": 1.6286,
      "step": 750
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8763782106540088,
      "learning_rate": 5.690393044458466e-05,
      "loss": 1.5855,
      "step": 751
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7860949578646186,
      "learning_rate": 5.689409533834826e-05,
      "loss": 1.5874,
      "step": 752
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5033113639117068,
      "learning_rate": 5.688424548827211e-05,
      "loss": 1.371,
      "step": 753
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6764126985711763,
      "learning_rate": 5.687438089975607e-05,
      "loss": 1.5559,
      "step": 754
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.644687079740996,
      "learning_rate": 5.686450157820809e-05,
      "loss": 1.4085,
      "step": 755
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.721947600741603,
      "learning_rate": 5.68546075290442e-05,
      "loss": 1.6028,
      "step": 756
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7256661817965333,
      "learning_rate": 5.68446987576885e-05,
      "loss": 1.5347,
      "step": 757
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7402371584452994,
      "learning_rate": 5.683477526957315e-05,
      "loss": 1.5322,
      "step": 758
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5700064960949818,
      "learning_rate": 5.682483707013842e-05,
      "loss": 1.3175,
      "step": 759
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7159446382243808,
      "learning_rate": 5.681488416483259e-05,
      "loss": 1.5889,
      "step": 760
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7825651535746163,
      "learning_rate": 5.680491655911203e-05,
      "loss": 1.4108,
      "step": 761
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6353168631902557,
      "learning_rate": 5.6794934258441164e-05,
      "loss": 1.486,
      "step": 762
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5107975551230057,
      "learning_rate": 5.678493726829248e-05,
      "loss": 1.2792,
      "step": 763
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8369147357978597,
      "learning_rate": 5.67749255941465e-05,
      "loss": 1.4489,
      "step": 764
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.526571223772283,
      "learning_rate": 5.676489924149183e-05,
      "loss": 1.252,
      "step": 765
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5992127492029335,
      "learning_rate": 5.6754858215825085e-05,
      "loss": 1.3316,
      "step": 766
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8299528198041322,
      "learning_rate": 5.6744802522650944e-05,
      "loss": 1.5477,
      "step": 767
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6156262424229764,
      "learning_rate": 5.673473216748212e-05,
      "loss": 1.4358,
      "step": 768
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5921057598282573,
      "learning_rate": 5.672464715583939e-05,
      "loss": 1.535,
      "step": 769
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5800681632402584,
      "learning_rate": 5.671454749325151e-05,
      "loss": 1.5051,
      "step": 770
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.522542692076913,
      "learning_rate": 5.6704433185255324e-05,
      "loss": 1.4143,
      "step": 771
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4122841046653178,
      "learning_rate": 5.669430423739568e-05,
      "loss": 1.3589,
      "step": 772
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6029727039834765,
      "learning_rate": 5.6684160655225446e-05,
      "loss": 1.4564,
      "step": 773
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7385226159981497,
      "learning_rate": 5.667400244430553e-05,
      "loss": 1.4022,
      "step": 774
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.637808276996643,
      "learning_rate": 5.6663829610204836e-05,
      "loss": 1.4487,
      "step": 775
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6809551586200526,
      "learning_rate": 5.665364215850032e-05,
      "loss": 1.4936,
      "step": 776
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6773234327163729,
      "learning_rate": 5.664344009477692e-05,
      "loss": 1.4115,
      "step": 777
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6871481040570757,
      "learning_rate": 5.663322342462759e-05,
      "loss": 1.519,
      "step": 778
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7775139284868298,
      "learning_rate": 5.6622992153653316e-05,
      "loss": 1.522,
      "step": 779
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.0401190913530463,
      "learning_rate": 5.661274628746304e-05,
      "loss": 1.7242,
      "step": 780
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.394648919354293,
      "learning_rate": 5.660248583167377e-05,
      "loss": 1.229,
      "step": 781
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.75004198831812,
      "learning_rate": 5.659221079191045e-05,
      "loss": 1.4406,
      "step": 782
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7872628324244746,
      "learning_rate": 5.6581921173806064e-05,
      "loss": 1.5787,
      "step": 783
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5313600542656387,
      "learning_rate": 5.6571616983001565e-05,
      "loss": 1.3778,
      "step": 784
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.847054345427227,
      "learning_rate": 5.65612982251459e-05,
      "loss": 1.5542,
      "step": 785
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6946716426649446,
      "learning_rate": 5.6550964905895994e-05,
      "loss": 1.475,
      "step": 786
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.627514291271781,
      "learning_rate": 5.654061703091677e-05,
      "loss": 1.3981,
      "step": 787
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6858299267923351,
      "learning_rate": 5.653025460588113e-05,
      "loss": 1.5196,
      "step": 788
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6592028486703365,
      "learning_rate": 5.6519877636469946e-05,
      "loss": 1.4756,
      "step": 789
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.460351222519514,
      "learning_rate": 5.650948612837206e-05,
      "loss": 1.2628,
      "step": 790
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6333832318445263,
      "learning_rate": 5.649908008728429e-05,
      "loss": 1.5509,
      "step": 791
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7202486213781838,
      "learning_rate": 5.648865951891142e-05,
      "loss": 1.5534,
      "step": 792
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.760820750060386,
      "learning_rate": 5.6478224428966185e-05,
      "loss": 1.6351,
      "step": 793
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7045348327043495,
      "learning_rate": 5.6467774823169316e-05,
      "loss": 1.4872,
      "step": 794
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8581844149651334,
      "learning_rate": 5.645731070724947e-05,
      "loss": 1.5207,
      "step": 795
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5298169629467335,
      "learning_rate": 5.6446832086943274e-05,
      "loss": 1.2457,
      "step": 796
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.789030133176417,
      "learning_rate": 5.643633896799529e-05,
      "loss": 1.5014,
      "step": 797
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7968362497401356,
      "learning_rate": 5.642583135615804e-05,
      "loss": 1.4271,
      "step": 798
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7647729625037114,
      "learning_rate": 5.6415309257192006e-05,
      "loss": 1.4882,
      "step": 799
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.856323828143797,
      "learning_rate": 5.640477267686558e-05,
      "loss": 1.6109,
      "step": 800
    },
    {
      "epoch": 0.91,
      "eval_blimp_filtered_avg": 0.7085074626865672,
      "eval_blimp_filtered_std": 0.005060439902410999,
      "step": 800
    },
    {
      "epoch": 0.91,
      "eval_blimp_supplement_avg": 0.8297413793103449,
      "eval_blimp_supplement_std": 0.016338037986177327,
      "step": 800
    },
    {
      "epoch": 0.91,
      "eval_vqa_filtered_avg": 0.53,
      "eval_vqa_filtered_std": 0.0501613558046592,
      "step": 800
    },
    {
      "epoch": 0.91,
      "eval_winoground_filtered_avg": 0.62,
      "eval_winoground_filtered_std": 0.04878317312145633,
      "step": 800
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8790589190121505,
      "learning_rate": 5.639422162095512e-05,
      "loss": 1.5526,
      "step": 801
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.121296894218886,
      "learning_rate": 5.638365609524491e-05,
      "loss": 1.5714,
      "step": 802
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6775928185789761,
      "learning_rate": 5.637307610552716e-05,
      "loss": 1.3583,
      "step": 803
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5514062034798604,
      "learning_rate": 5.636248165760202e-05,
      "loss": 1.3405,
      "step": 804
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7384456466832534,
      "learning_rate": 5.6351872757277564e-05,
      "loss": 1.4176,
      "step": 805
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5919672191007601,
      "learning_rate": 5.634124941036978e-05,
      "loss": 1.4862,
      "step": 806
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7470197323663734,
      "learning_rate": 5.633061162270258e-05,
      "loss": 1.5187,
      "step": 807
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8652758089896153,
      "learning_rate": 5.631995940010782e-05,
      "loss": 1.5705,
      "step": 808
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8037763592580567,
      "learning_rate": 5.6309292748425225e-05,
      "loss": 1.5384,
      "step": 809
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7379528978328818,
      "learning_rate": 5.6298611673502446e-05,
      "loss": 1.3855,
      "step": 810
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9581367572833903,
      "learning_rate": 5.6287916181195056e-05,
      "loss": 1.5563,
      "step": 811
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6184991515766893,
      "learning_rate": 5.6277206277366516e-05,
      "loss": 1.4172,
      "step": 812
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6967491988797159,
      "learning_rate": 5.62664819678882e-05,
      "loss": 1.27,
      "step": 813
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8721016139206823,
      "learning_rate": 5.6255743258639364e-05,
      "loss": 1.377,
      "step": 814
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8196223785199048,
      "learning_rate": 5.6244990155507165e-05,
      "loss": 1.493,
      "step": 815
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.903548237563561,
      "learning_rate": 5.623422266438667e-05,
      "loss": 1.5379,
      "step": 816
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6181487352912887,
      "learning_rate": 5.62234407911808e-05,
      "loss": 1.2756,
      "step": 817
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4248673353146732,
      "learning_rate": 5.621264454180038e-05,
      "loss": 1.2903,
      "step": 818
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7797501707108152,
      "learning_rate": 5.6201833922164124e-05,
      "loss": 1.5614,
      "step": 819
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.497585824534769,
      "learning_rate": 5.6191008938198595e-05,
      "loss": 1.2819,
      "step": 820
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.71091108234463,
      "learning_rate": 5.618016959583826e-05,
      "loss": 1.3298,
      "step": 821
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6721179643596733,
      "learning_rate": 5.616931590102545e-05,
      "loss": 1.3039,
      "step": 822
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8436012048375339,
      "learning_rate": 5.615844785971036e-05,
      "loss": 1.4018,
      "step": 823
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.150855453253326,
      "learning_rate": 5.614756547785103e-05,
      "loss": 1.4952,
      "step": 824
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6908096801177752,
      "learning_rate": 5.6136668761413413e-05,
      "loss": 1.2689,
      "step": 825
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9919240142709989,
      "learning_rate": 5.612575771637128e-05,
      "loss": 1.6159,
      "step": 826
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6160594010231644,
      "learning_rate": 5.6114832348706254e-05,
      "loss": 1.4677,
      "step": 827
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6933276291028838,
      "learning_rate": 5.610389266440784e-05,
      "loss": 1.5009,
      "step": 828
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5835109673189331,
      "learning_rate": 5.609293866947337e-05,
      "loss": 1.4373,
      "step": 829
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3641103552092122,
      "learning_rate": 5.608197036990802e-05,
      "loss": 1.2046,
      "step": 830
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7500003174599401,
      "learning_rate": 5.607098777172483e-05,
      "loss": 1.4292,
      "step": 831
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8342377617461143,
      "learning_rate": 5.6059990880944644e-05,
      "loss": 1.3804,
      "step": 832
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5803000338559456,
      "learning_rate": 5.604897970359618e-05,
      "loss": 1.3333,
      "step": 833
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7657431513767947,
      "learning_rate": 5.6037954245715946e-05,
      "loss": 1.4091,
      "step": 834
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.740920083210782,
      "learning_rate": 5.602691451334832e-05,
      "loss": 1.3477,
      "step": 835
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8037154606205328,
      "learning_rate": 5.601586051254548e-05,
      "loss": 1.5045,
      "step": 836
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7534690156053325,
      "learning_rate": 5.600479224936743e-05,
      "loss": 1.383,
      "step": 837
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.06718393118272,
      "learning_rate": 5.5993709729882006e-05,
      "loss": 1.509,
      "step": 838
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.878456814037823,
      "learning_rate": 5.5982612960164844e-05,
      "loss": 1.3172,
      "step": 839
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6505307629434443,
      "learning_rate": 5.5971501946299395e-05,
      "loss": 1.3661,
      "step": 840
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7104406512838177,
      "learning_rate": 5.5960376694376924e-05,
      "loss": 1.4109,
      "step": 841
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4866225842961254,
      "learning_rate": 5.5949237210496514e-05,
      "loss": 1.3352,
      "step": 842
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.755353877176802,
      "learning_rate": 5.593808350076501e-05,
      "loss": 1.4586,
      "step": 843
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.599089986509237,
      "learning_rate": 5.59269155712971e-05,
      "loss": 1.3165,
      "step": 844
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6318907587122635,
      "learning_rate": 5.5915733428215244e-05,
      "loss": 1.2954,
      "step": 845
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7234337109552504,
      "learning_rate": 5.59045370776497e-05,
      "loss": 1.4602,
      "step": 846
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7278513257360841,
      "learning_rate": 5.589332652573851e-05,
      "loss": 1.3357,
      "step": 847
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.0195180437762614,
      "learning_rate": 5.5882101778627516e-05,
      "loss": 1.5811,
      "step": 848
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6792479848974846,
      "learning_rate": 5.5870862842470316e-05,
      "loss": 1.2462,
      "step": 849
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7469986323216977,
      "learning_rate": 5.585960972342832e-05,
      "loss": 1.3652,
      "step": 850
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8682783298024799,
      "learning_rate": 5.5848342427670676e-05,
      "loss": 1.4655,
      "step": 851
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.899875084853686,
      "learning_rate": 5.5837060961374336e-05,
      "loss": 1.4152,
      "step": 852
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.827831224678235,
      "learning_rate": 5.582576533072401e-05,
      "loss": 1.3493,
      "step": 853
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.936621775530969,
      "learning_rate": 5.5814455541912173e-05,
      "loss": 1.4986,
      "step": 854
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6372378614121215,
      "learning_rate": 5.580313160113905e-05,
      "loss": 1.4051,
      "step": 855
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7129504626858272,
      "learning_rate": 5.5791793514612635e-05,
      "loss": 1.4889,
      "step": 856
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.58069726186269,
      "learning_rate": 5.578044128854869e-05,
      "loss": 1.347,
      "step": 857
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7578940003634131,
      "learning_rate": 5.5769074929170704e-05,
      "loss": 1.3574,
      "step": 858
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7037191148719204,
      "learning_rate": 5.575769444270992e-05,
      "loss": 1.4026,
      "step": 859
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.428697297781483,
      "learning_rate": 5.574629983540535e-05,
      "loss": 1.0523,
      "step": 860
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.638292286410352,
      "learning_rate": 5.57348911135037e-05,
      "loss": 1.2439,
      "step": 861
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9671074906507127,
      "learning_rate": 5.572346828325947e-05,
      "loss": 1.4241,
      "step": 862
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8920256434706866,
      "learning_rate": 5.5712031350934847e-05,
      "loss": 1.4209,
      "step": 863
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.722381481788187,
      "learning_rate": 5.570058032279978e-05,
      "loss": 1.3302,
      "step": 864
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7052401816128466,
      "learning_rate": 5.568911520513193e-05,
      "loss": 1.3694,
      "step": 865
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6173133367941854,
      "learning_rate": 5.567763600421668e-05,
      "loss": 1.1603,
      "step": 866
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7222693813355923,
      "learning_rate": 5.566614272634716e-05,
      "loss": 1.3202,
      "step": 867
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7271120774317974,
      "learning_rate": 5.565463537782417e-05,
      "loss": 1.3453,
      "step": 868
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7111293618188352,
      "learning_rate": 5.564311396495628e-05,
      "loss": 1.3322,
      "step": 869
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7007313902870944,
      "learning_rate": 5.563157849405972e-05,
      "loss": 1.3268,
      "step": 870
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6431543280475667,
      "learning_rate": 5.5620028971458444e-05,
      "loss": 1.1841,
      "step": 871
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8919876042119086,
      "learning_rate": 5.560846540348413e-05,
      "loss": 1.3531,
      "step": 872
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.0046110552155443,
      "learning_rate": 5.559688779647613e-05,
      "loss": 1.3658,
      "step": 873
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8623046649624033,
      "learning_rate": 5.5585296156781516e-05,
      "loss": 1.4509,
      "step": 874
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7354970508087575,
      "learning_rate": 5.557369049075502e-05,
      "loss": 1.2686,
      "step": 875
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3138474881892819,
      "learning_rate": 5.5562070804759084e-05,
      "loss": 0.9926,
      "step": 876
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5534278872604412,
      "learning_rate": 5.555043710516385e-05,
      "loss": 1.1746,
      "step": 877
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4237316596311138,
      "learning_rate": 5.5538789398347114e-05,
      "loss": 0.9525,
      "step": 878
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5787668353119084,
      "learning_rate": 5.5527127690694375e-05,
      "loss": 1.1663,
      "step": 879
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.5115540924493072,
      "learning_rate": 5.5515451988598783e-05,
      "loss": 1.0825,
      "step": 880
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.6484470445419066,
      "learning_rate": 5.5503762298461185e-05,
      "loss": 1.1033,
      "step": 881
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.764900907333099,
      "learning_rate": 5.5492058626690074e-05,
      "loss": 1.1477,
      "step": 882
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.6766507463121467,
      "learning_rate": 5.5480340979701625e-05,
      "loss": 1.1407,
      "step": 883
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.8949985740075348,
      "learning_rate": 5.546860936391968e-05,
      "loss": 1.2781,
      "step": 884
    },
    {
      "epoch": 1.01,
      "grad_norm": 2.014030869959555,
      "learning_rate": 5.54568637857757e-05,
      "loss": 1.2259,
      "step": 885
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.9194144379328961,
      "learning_rate": 5.544510425170886e-05,
      "loss": 1.296,
      "step": 886
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.7725077198307666,
      "learning_rate": 5.543333076816593e-05,
      "loss": 1.1042,
      "step": 887
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.6473085202798547,
      "learning_rate": 5.542154334160135e-05,
      "loss": 1.1338,
      "step": 888
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.8429418590418059,
      "learning_rate": 5.5409741978477225e-05,
      "loss": 1.2366,
      "step": 889
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.5709096223025258,
      "learning_rate": 5.539792668526326e-05,
      "loss": 1.095,
      "step": 890
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.509383064121676,
      "learning_rate": 5.538609746843683e-05,
      "loss": 1.1844,
      "step": 891
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4784787986967862,
      "learning_rate": 5.537425433448293e-05,
      "loss": 0.9678,
      "step": 892
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.7204751221832322,
      "learning_rate": 5.5362397289894176e-05,
      "loss": 1.223,
      "step": 893
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.7702079437459524,
      "learning_rate": 5.5350526341170826e-05,
      "loss": 1.1424,
      "step": 894
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.6755629664709322,
      "learning_rate": 5.5338641494820755e-05,
      "loss": 1.1368,
      "step": 895
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.617640194174331,
      "learning_rate": 5.532674275735944e-05,
      "loss": 1.0904,
      "step": 896
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.6708626431243085,
      "learning_rate": 5.531483013531001e-05,
      "loss": 1.0367,
      "step": 897
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.674642529155393,
      "learning_rate": 5.530290363520317e-05,
      "loss": 1.0291,
      "step": 898
    },
    {
      "epoch": 1.03,
      "grad_norm": 2.0059406506400603,
      "learning_rate": 5.529096326357725e-05,
      "loss": 1.3624,
      "step": 899
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.6501950750080034,
      "learning_rate": 5.5279009026978184e-05,
      "loss": 1.1796,
      "step": 900
    },
    {
      "epoch": 1.03,
      "eval_blimp_filtered_avg": 0.7044776119402985,
      "eval_blimp_filtered_std": 0.005064151442605659,
      "step": 900
    },
    {
      "epoch": 1.03,
      "eval_blimp_supplement_avg": 0.8254310344827587,
      "eval_blimp_supplement_std": 0.017088747032722257,
      "step": 900
    },
    {
      "epoch": 1.03,
      "eval_vqa_filtered_avg": 0.54,
      "eval_vqa_filtered_std": 0.05009082659620333,
      "step": 900
    },
    {
      "epoch": 1.03,
      "eval_winoground_filtered_avg": 0.61,
      "eval_winoground_filtered_std": 0.04902071300001975,
      "step": 900
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.7579039736666338,
      "learning_rate": 5.5267040931959514e-05,
      "loss": 1.1143,
      "step": 901
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.6554733239759127,
      "learning_rate": 5.525505898508235e-05,
      "loss": 1.159,
      "step": 902
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.6757364768599994,
      "learning_rate": 5.524306319291543e-05,
      "loss": 1.1864,
      "step": 903
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.4172735747824476,
      "learning_rate": 5.523105356203508e-05,
      "loss": 1.0644,
      "step": 904
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.3527612720357602,
      "learning_rate": 5.5219030099025175e-05,
      "loss": 1.0255,
      "step": 905
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5670271953953838,
      "learning_rate": 5.520699281047722e-05,
      "loss": 1.1365,
      "step": 906
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.533901837650238,
      "learning_rate": 5.519494170299027e-05,
      "loss": 1.159,
      "step": 907
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.718636421690865,
      "learning_rate": 5.5182876783170963e-05,
      "loss": 1.2194,
      "step": 908
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5203406360271439,
      "learning_rate": 5.517079805763352e-05,
      "loss": 1.0914,
      "step": 909
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.850345486149302,
      "learning_rate": 5.5158705532999715e-05,
      "loss": 1.172,
      "step": 910
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.8912234318595105,
      "learning_rate": 5.51465992158989e-05,
      "loss": 1.0514,
      "step": 911
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.891501086577689,
      "learning_rate": 5.513447911296797e-05,
      "loss": 1.25,
      "step": 912
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.891199235142435,
      "learning_rate": 5.512234523085139e-05,
      "loss": 1.1311,
      "step": 913
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.7283546175903768,
      "learning_rate": 5.5110197576201194e-05,
      "loss": 1.0977,
      "step": 914
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5783018821471697,
      "learning_rate": 5.509803615567694e-05,
      "loss": 1.1124,
      "step": 915
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5829142873340472,
      "learning_rate": 5.508586097594574e-05,
      "loss": 1.0481,
      "step": 916
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5932315581320509,
      "learning_rate": 5.507367204368225e-05,
      "loss": 1.2168,
      "step": 917
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.7176954227936598,
      "learning_rate": 5.506146936556868e-05,
      "loss": 1.1705,
      "step": 918
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6977171771177324,
      "learning_rate": 5.5049252948294764e-05,
      "loss": 1.2003,
      "step": 919
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.740234486574763,
      "learning_rate": 5.5037022798557755e-05,
      "loss": 1.1791,
      "step": 920
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.7482636514050887,
      "learning_rate": 5.502477892306245e-05,
      "loss": 1.1658,
      "step": 921
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.646171629864516,
      "learning_rate": 5.501252132852118e-05,
      "loss": 1.1725,
      "step": 922
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6951993193272727,
      "learning_rate": 5.500025002165376e-05,
      "loss": 1.0847,
      "step": 923
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.5848671307837516,
      "learning_rate": 5.498796500918757e-05,
      "loss": 1.151,
      "step": 924
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.85152009849042,
      "learning_rate": 5.497566629785748e-05,
      "loss": 1.2754,
      "step": 925
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.7162738720512114,
      "learning_rate": 5.496335389440585e-05,
      "loss": 1.1225,
      "step": 926
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.6518867319069868,
      "learning_rate": 5.495102780558259e-05,
      "loss": 1.2111,
      "step": 927
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.6878116801508698,
      "learning_rate": 5.4938688038145075e-05,
      "loss": 1.1914,
      "step": 928
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.5817962510224797,
      "learning_rate": 5.4926334598858205e-05,
      "loss": 1.1052,
      "step": 929
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.4317878380145663,
      "learning_rate": 5.491396749449436e-05,
      "loss": 1.0167,
      "step": 930
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.646762006642982,
      "learning_rate": 5.4901586731833395e-05,
      "loss": 1.1597,
      "step": 931
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.612981725149576,
      "learning_rate": 5.48891923176627e-05,
      "loss": 1.1167,
      "step": 932
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.4917982444623474,
      "learning_rate": 5.4876784258777116e-05,
      "loss": 1.0896,
      "step": 933
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.650474557284239,
      "learning_rate": 5.486436256197897e-05,
      "loss": 1.1622,
      "step": 934
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5556299955796078,
      "learning_rate": 5.485192723407804e-05,
      "loss": 1.0476,
      "step": 935
    },
    {
      "epoch": 1.07,
      "grad_norm": 2.016498825442656,
      "learning_rate": 5.483947828189164e-05,
      "loss": 1.3163,
      "step": 936
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.8838398691522884,
      "learning_rate": 5.4827015712244496e-05,
      "loss": 1.123,
      "step": 937
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.8028656281925466,
      "learning_rate": 5.481453953196882e-05,
      "loss": 1.2448,
      "step": 938
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.7055450305490294,
      "learning_rate": 5.480204974790429e-05,
      "loss": 1.1388,
      "step": 939
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5101133511096079,
      "learning_rate": 5.4789546366898034e-05,
      "loss": 1.0473,
      "step": 940
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.68359628386981,
      "learning_rate": 5.4777029395804625e-05,
      "loss": 1.0457,
      "step": 941
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.6882323953061034,
      "learning_rate": 5.476449884148612e-05,
      "loss": 1.048,
      "step": 942
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.9229506252505228,
      "learning_rate": 5.4751954710811985e-05,
      "loss": 1.3584,
      "step": 943
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.8377692450268086,
      "learning_rate": 5.473939701065915e-05,
      "loss": 1.1521,
      "step": 944
    },
    {
      "epoch": 1.08,
      "grad_norm": 2.1668363925389773,
      "learning_rate": 5.4726825747911974e-05,
      "loss": 1.2132,
      "step": 945
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.6794295088621158,
      "learning_rate": 5.471424092946225e-05,
      "loss": 1.0967,
      "step": 946
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.65986994241679,
      "learning_rate": 5.470164256220922e-05,
      "loss": 1.2218,
      "step": 947
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.6260234512854137,
      "learning_rate": 5.4689030653059536e-05,
      "loss": 1.1262,
      "step": 948
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5773322684362765,
      "learning_rate": 5.467640520892728e-05,
      "loss": 1.1962,
      "step": 949
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5122110569932246,
      "learning_rate": 5.466376623673395e-05,
      "loss": 1.086,
      "step": 950
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.8085713426573027,
      "learning_rate": 5.465111374340848e-05,
      "loss": 1.213,
      "step": 951
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5727126622756444,
      "learning_rate": 5.463844773588717e-05,
      "loss": 1.1247,
      "step": 952
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.7072911133381254,
      "learning_rate": 5.462576822111379e-05,
      "loss": 1.1894,
      "step": 953
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.6901064528252254,
      "learning_rate": 5.461307520603946e-05,
      "loss": 1.226,
      "step": 954
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.6402987907935895,
      "learning_rate": 5.460036869762275e-05,
      "loss": 1.0601,
      "step": 955
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.7480113717216843,
      "learning_rate": 5.4587648702829576e-05,
      "loss": 1.1587,
      "step": 956
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.784621673070492,
      "learning_rate": 5.457491522863331e-05,
      "loss": 1.2676,
      "step": 957
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.745491347799377,
      "learning_rate": 5.456216828201463e-05,
      "loss": 1.0565,
      "step": 958
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.603916742898186,
      "learning_rate": 5.454940786996168e-05,
      "loss": 1.0461,
      "step": 959
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6028520089829588,
      "learning_rate": 5.4536633999469956e-05,
      "loss": 1.0577,
      "step": 960
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6607306081282303,
      "learning_rate": 5.4523846677542306e-05,
      "loss": 0.9817,
      "step": 961
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.7426113383147426,
      "learning_rate": 5.4511045911188995e-05,
      "loss": 1.1279,
      "step": 962
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.486565769543867,
      "learning_rate": 5.449823170742764e-05,
      "loss": 1.0643,
      "step": 963
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.7673551889943475,
      "learning_rate": 5.4485404073283206e-05,
      "loss": 1.0907,
      "step": 964
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.564699652960806,
      "learning_rate": 5.447256301578807e-05,
      "loss": 1.1288,
      "step": 965
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6901954705791966,
      "learning_rate": 5.4459708541981904e-05,
      "loss": 1.1171,
      "step": 966
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.677001631002564,
      "learning_rate": 5.4446840658911786e-05,
      "loss": 1.1619,
      "step": 967
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.9186226987536292,
      "learning_rate": 5.443395937363213e-05,
      "loss": 1.1768,
      "step": 968
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.9223018226846544,
      "learning_rate": 5.4421064693204684e-05,
      "loss": 1.2174,
      "step": 969
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.592226847813751,
      "learning_rate": 5.440815662469855e-05,
      "loss": 1.047,
      "step": 970
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.678476034716879,
      "learning_rate": 5.439523517519018e-05,
      "loss": 1.1233,
      "step": 971
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.7236861990783727,
      "learning_rate": 5.438230035176333e-05,
      "loss": 1.1277,
      "step": 972
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.834035388234123,
      "learning_rate": 5.436935216150914e-05,
      "loss": 1.2324,
      "step": 973
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.5465489313891918,
      "learning_rate": 5.435639061152603e-05,
      "loss": 1.1061,
      "step": 974
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.578082553544806,
      "learning_rate": 5.434341570891976e-05,
      "loss": 1.1045,
      "step": 975
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.7813262202166595,
      "learning_rate": 5.4330427460803404e-05,
      "loss": 1.1791,
      "step": 976
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.7824119238154674,
      "learning_rate": 5.431742587429738e-05,
      "loss": 1.1478,
      "step": 977
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6913317071628953,
      "learning_rate": 5.430441095652938e-05,
      "loss": 1.1413,
      "step": 978
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.7748357029755866,
      "learning_rate": 5.429138271463444e-05,
      "loss": 1.1377,
      "step": 979
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.9531419419080451,
      "learning_rate": 5.427834115575485e-05,
      "loss": 1.3046,
      "step": 980
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6813003436591678,
      "learning_rate": 5.4265286287040274e-05,
      "loss": 1.1882,
      "step": 981
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.548979271408274,
      "learning_rate": 5.42522181156476e-05,
      "loss": 1.1499,
      "step": 982
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6967458598685092,
      "learning_rate": 5.423913664874106e-05,
      "loss": 1.1995,
      "step": 983
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5587638273234685,
      "learning_rate": 5.422604189349215e-05,
      "loss": 1.1293,
      "step": 984
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.6593444890369444,
      "learning_rate": 5.421293385707967e-05,
      "loss": 1.2418,
      "step": 985
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.924991438119862,
      "learning_rate": 5.419981254668966e-05,
      "loss": 1.1815,
      "step": 986
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.9477075715619698,
      "learning_rate": 5.4186677969515475e-05,
      "loss": 1.2513,
      "step": 987
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.6353173540184325,
      "learning_rate": 5.4173530132757745e-05,
      "loss": 0.9925,
      "step": 988
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.7643889322461666,
      "learning_rate": 5.4160369043624344e-05,
      "loss": 1.1897,
      "step": 989
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.8134144929637626,
      "learning_rate": 5.4147194709330434e-05,
      "loss": 1.1083,
      "step": 990
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.8624626376899542,
      "learning_rate": 5.4134007137098415e-05,
      "loss": 1.1729,
      "step": 991
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.7118188829118786,
      "learning_rate": 5.4120806334157976e-05,
      "loss": 1.1064,
      "step": 992
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.6003118179064162,
      "learning_rate": 5.410759230774602e-05,
      "loss": 1.0625,
      "step": 993
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.7833187933324484,
      "learning_rate": 5.409436506510674e-05,
      "loss": 1.1737,
      "step": 994
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.749874896864711,
      "learning_rate": 5.408112461349154e-05,
      "loss": 1.1971,
      "step": 995
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.668026314901681,
      "learning_rate": 5.406787096015909e-05,
      "loss": 1.1074,
      "step": 996
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.5920711912533734,
      "learning_rate": 5.405460411237527e-05,
      "loss": 1.0797,
      "step": 997
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.5629887080669658,
      "learning_rate": 5.404132407741325e-05,
      "loss": 1.0587,
      "step": 998
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.5522004933016313,
      "learning_rate": 5.402803086255335e-05,
      "loss": 1.0546,
      "step": 999
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.5970291021086827,
      "learning_rate": 5.401472447508317e-05,
      "loss": 1.2211,
      "step": 1000
    },
    {
      "epoch": 1.14,
      "eval_blimp_filtered_avg": 0.698955223880597,
      "eval_blimp_filtered_std": 0.005111275859983884,
      "step": 1000
    },
    {
      "epoch": 1.14,
      "eval_blimp_supplement_avg": 0.8232758620689655,
      "eval_blimp_supplement_std": 0.017380679753624985,
      "step": 1000
    },
    {
      "epoch": 1.14,
      "eval_vqa_filtered_avg": 0.53,
      "eval_vqa_filtered_std": 0.0501613558046592,
      "step": 1000
    },
    {
      "epoch": 1.14,
      "eval_winoground_filtered_avg": 0.63,
      "eval_winoground_filtered_std": 0.048523658709391,
      "step": 1000
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.819915789246139,
      "learning_rate": 5.4001404922297515e-05,
      "loss": 1.1056,
      "step": 1001
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.580636925218752,
      "learning_rate": 5.398807221149842e-05,
      "loss": 1.013,
      "step": 1002
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.6261518116219325,
      "learning_rate": 5.3974726349995106e-05,
      "loss": 1.1264,
      "step": 1003
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5972909505420851,
      "learning_rate": 5.396136734510405e-05,
      "loss": 1.0761,
      "step": 1004
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5888042059749012,
      "learning_rate": 5.394799520414887e-05,
      "loss": 1.1258,
      "step": 1005
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.8747647919047106,
      "learning_rate": 5.3934609934460444e-05,
      "loss": 1.1713,
      "step": 1006
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5150757247258078,
      "learning_rate": 5.392121154337681e-05,
      "loss": 0.9526,
      "step": 1007
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.8576298263649074,
      "learning_rate": 5.3907800038243225e-05,
      "loss": 1.1889,
      "step": 1008
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.787894366806377,
      "learning_rate": 5.3894375426412115e-05,
      "loss": 1.1188,
      "step": 1009
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.7732379140996226,
      "learning_rate": 5.3880937715243114e-05,
      "loss": 1.1043,
      "step": 1010
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.817881637405633,
      "learning_rate": 5.3867486912103e-05,
      "loss": 1.1021,
      "step": 1011
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6363124114808147,
      "learning_rate": 5.385402302436577e-05,
      "loss": 1.02,
      "step": 1012
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.8703963269643766,
      "learning_rate": 5.384054605941258e-05,
      "loss": 1.2035,
      "step": 1013
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.642606672824856,
      "learning_rate": 5.382705602463173e-05,
      "loss": 1.0971,
      "step": 1014
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6940397636996523,
      "learning_rate": 5.381355292741873e-05,
      "loss": 1.1721,
      "step": 1015
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.8123235648157374,
      "learning_rate": 5.3800036775176207e-05,
      "loss": 1.2266,
      "step": 1016
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6178277916842527,
      "learning_rate": 5.3786507575313976e-05,
      "loss": 1.1208,
      "step": 1017
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.550090894387767,
      "learning_rate": 5.3772965335249e-05,
      "loss": 1.0452,
      "step": 1018
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.3779842400856477,
      "learning_rate": 5.375941006240537e-05,
      "loss": 0.99,
      "step": 1019
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.7596020772975824,
      "learning_rate": 5.374584176421434e-05,
      "loss": 1.1274,
      "step": 1020
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.929745653767573,
      "learning_rate": 5.3732260448114315e-05,
      "loss": 1.2141,
      "step": 1021
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.8611991198800302,
      "learning_rate": 5.371866612155081e-05,
      "loss": 1.2536,
      "step": 1022
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5760130569038542,
      "learning_rate": 5.370505879197649e-05,
      "loss": 1.1188,
      "step": 1023
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5827768024208306,
      "learning_rate": 5.369143846685114e-05,
      "loss": 1.0532,
      "step": 1024
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.500959813514371,
      "learning_rate": 5.367780515364169e-05,
      "loss": 0.9844,
      "step": 1025
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5966239809586162,
      "learning_rate": 5.3664158859822154e-05,
      "loss": 1.0161,
      "step": 1026
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5930072875217756,
      "learning_rate": 5.365049959287369e-05,
      "loss": 1.0222,
      "step": 1027
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.7125894954372352,
      "learning_rate": 5.3636827360284565e-05,
      "loss": 1.1361,
      "step": 1028
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.6602043046949635,
      "learning_rate": 5.362314216955015e-05,
      "loss": 0.9786,
      "step": 1029
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.7434546421924721,
      "learning_rate": 5.36094440281729e-05,
      "loss": 1.0392,
      "step": 1030
    },
    {
      "epoch": 1.18,
      "grad_norm": 2.1017262882490675,
      "learning_rate": 5.359573294366242e-05,
      "loss": 1.1242,
      "step": 1031
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.596046616227655,
      "learning_rate": 5.3582008923535355e-05,
      "loss": 0.9845,
      "step": 1032
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.9328842398430688,
      "learning_rate": 5.356827197531547e-05,
      "loss": 1.0444,
      "step": 1033
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.7912369613876944,
      "learning_rate": 5.3554522106533635e-05,
      "loss": 1.1381,
      "step": 1034
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.7726811513267489,
      "learning_rate": 5.3540759324727755e-05,
      "loss": 1.0894,
      "step": 1035
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.966233375679466,
      "learning_rate": 5.3526983637442855e-05,
      "loss": 1.2745,
      "step": 1036
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.6308415983828826,
      "learning_rate": 5.351319505223101e-05,
      "loss": 1.0883,
      "step": 1037
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.6668451452086293,
      "learning_rate": 5.3499393576651396e-05,
      "loss": 1.1259,
      "step": 1038
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.7571272147508983,
      "learning_rate": 5.348557921827022e-05,
      "loss": 1.1862,
      "step": 1039
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.4246589976191855,
      "learning_rate": 5.3471751984660776e-05,
      "loss": 1.0114,
      "step": 1040
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.5942451416513748,
      "learning_rate": 5.345791188340341e-05,
      "loss": 1.0137,
      "step": 1041
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.513479755403777,
      "learning_rate": 5.3444058922085505e-05,
      "loss": 1.0392,
      "step": 1042
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.7915005648037152,
      "learning_rate": 5.343019310830154e-05,
      "loss": 1.1425,
      "step": 1043
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.8570785789575273,
      "learning_rate": 5.341631444965299e-05,
      "loss": 1.1738,
      "step": 1044
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.7301342334523904,
      "learning_rate": 5.3402422953748395e-05,
      "loss": 1.098,
      "step": 1045
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.8588931922649303,
      "learning_rate": 5.338851862820333e-05,
      "loss": 1.0626,
      "step": 1046
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.6264601430968642,
      "learning_rate": 5.3374601480640394e-05,
      "loss": 0.9818,
      "step": 1047
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.6506862670067801,
      "learning_rate": 5.3360671518689244e-05,
      "loss": 1.1456,
      "step": 1048
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.612981990574752,
      "learning_rate": 5.334672874998653e-05,
      "loss": 1.0095,
      "step": 1049
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.663598402940263,
      "learning_rate": 5.3332773182175935e-05,
      "loss": 1.0619,
      "step": 1050
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.8893593416285273,
      "learning_rate": 5.331880482290816e-05,
      "loss": 1.1506,
      "step": 1051
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.714777490231793,
      "learning_rate": 5.330482367984092e-05,
      "loss": 1.0455,
      "step": 1052
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.3240848084315144,
      "learning_rate": 5.329082976063894e-05,
      "loss": 0.8957,
      "step": 1053
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.842580347433187,
      "learning_rate": 5.3276823072973936e-05,
      "loss": 1.118,
      "step": 1054
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.6240745685779476,
      "learning_rate": 5.326280362452464e-05,
      "loss": 1.0967,
      "step": 1055
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.7469740867223484,
      "learning_rate": 5.324877142297676e-05,
      "loss": 1.1894,
      "step": 1056
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.4768192243820932,
      "learning_rate": 5.323472647602303e-05,
      "loss": 1.0052,
      "step": 1057
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.7212541785013353,
      "learning_rate": 5.3220668791363126e-05,
      "loss": 1.045,
      "step": 1058
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.6401288863367596,
      "learning_rate": 5.320659837670375e-05,
      "loss": 0.9986,
      "step": 1059
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.850649128874655,
      "learning_rate": 5.319251523975854e-05,
      "loss": 0.9583,
      "step": 1060
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.553972107166094,
      "learning_rate": 5.3178419388248154e-05,
      "loss": 0.9371,
      "step": 1061
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.5831187435451846,
      "learning_rate": 5.316431082990019e-05,
      "loss": 1.0693,
      "step": 1062
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.8538102586817027,
      "learning_rate": 5.315018957244923e-05,
      "loss": 1.1254,
      "step": 1063
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.7298849952950737,
      "learning_rate": 5.313605562363679e-05,
      "loss": 1.0943,
      "step": 1064
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.740347318096972,
      "learning_rate": 5.3121908991211364e-05,
      "loss": 1.0713,
      "step": 1065
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.5995957491917872,
      "learning_rate": 5.3107749682928405e-05,
      "loss": 1.0653,
      "step": 1066
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.6910855345941953,
      "learning_rate": 5.309357770655032e-05,
      "loss": 1.0563,
      "step": 1067
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.6628519747689388,
      "learning_rate": 5.307939306984641e-05,
      "loss": 1.2135,
      "step": 1068
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.662643998121514,
      "learning_rate": 5.306519578059299e-05,
      "loss": 1.1479,
      "step": 1069
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.7344170602102815,
      "learning_rate": 5.305098584657326e-05,
      "loss": 1.1381,
      "step": 1070
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.5445880033980561,
      "learning_rate": 5.3036763275577384e-05,
      "loss": 1.0871,
      "step": 1071
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.656281052056438,
      "learning_rate": 5.3022528075402425e-05,
      "loss": 1.1171,
      "step": 1072
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5581464229089637,
      "learning_rate": 5.300828025385238e-05,
      "loss": 0.9813,
      "step": 1073
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.704895389102934,
      "learning_rate": 5.299401981873817e-05,
      "loss": 1.0612,
      "step": 1074
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.6318669092488645,
      "learning_rate": 5.2979746777877636e-05,
      "loss": 1.0346,
      "step": 1075
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.7889295178362676,
      "learning_rate": 5.296546113909552e-05,
      "loss": 1.0499,
      "step": 1076
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.8084453464558534,
      "learning_rate": 5.295116291022347e-05,
      "loss": 1.0878,
      "step": 1077
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.9831115230227918,
      "learning_rate": 5.2936852099100036e-05,
      "loss": 1.0845,
      "step": 1078
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.866596183532724,
      "learning_rate": 5.292252871357067e-05,
      "loss": 1.1194,
      "step": 1079
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.477178157715731,
      "learning_rate": 5.290819276148773e-05,
      "loss": 1.0009,
      "step": 1080
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.606356635784749,
      "learning_rate": 5.289384425071044e-05,
      "loss": 1.1185,
      "step": 1081
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.8843900426314368,
      "learning_rate": 5.2879483189104904e-05,
      "loss": 1.1546,
      "step": 1082
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.7124952857671125,
      "learning_rate": 5.286510958454414e-05,
      "loss": 1.0521,
      "step": 1083
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5601662459647576,
      "learning_rate": 5.285072344490803e-05,
      "loss": 1.0765,
      "step": 1084
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.797902741299415,
      "learning_rate": 5.28363247780833e-05,
      "loss": 1.133,
      "step": 1085
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.6968860991511139,
      "learning_rate": 5.282191359196358e-05,
      "loss": 1.0629,
      "step": 1086
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5482313012793754,
      "learning_rate": 5.280748989444934e-05,
      "loss": 0.9879,
      "step": 1087
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.8081104028361141,
      "learning_rate": 5.279305369344792e-05,
      "loss": 1.0962,
      "step": 1088
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.8401438709176026,
      "learning_rate": 5.277860499687352e-05,
      "loss": 0.9601,
      "step": 1089
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.7996850601573224,
      "learning_rate": 5.2764143812647167e-05,
      "loss": 0.9779,
      "step": 1090
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.8123891313955476,
      "learning_rate": 5.2749670148696755e-05,
      "loss": 1.07,
      "step": 1091
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.6869413208074666,
      "learning_rate": 5.273518401295703e-05,
      "loss": 1.0441,
      "step": 1092
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.8453659929231079,
      "learning_rate": 5.272068541336953e-05,
      "loss": 1.1031,
      "step": 1093
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.7263325779452556,
      "learning_rate": 5.2706174357882685e-05,
      "loss": 1.0784,
      "step": 1094
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.6142748246614407,
      "learning_rate": 5.26916508544517e-05,
      "loss": 0.9503,
      "step": 1095
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.9335079183541168,
      "learning_rate": 5.267711491103864e-05,
      "loss": 1.1268,
      "step": 1096
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.6422791117440658,
      "learning_rate": 5.266256653561237e-05,
      "loss": 0.9563,
      "step": 1097
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.5631519097305542,
      "learning_rate": 5.26480057361486e-05,
      "loss": 1.0132,
      "step": 1098
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.6261338536999068,
      "learning_rate": 5.263343252062979e-05,
      "loss": 0.9155,
      "step": 1099
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.5988154483722148,
      "learning_rate": 5.261884689704527e-05,
      "loss": 1.0728,
      "step": 1100
    },
    {
      "epoch": 1.26,
      "eval_blimp_filtered_avg": 0.705820895522388,
      "eval_blimp_filtered_std": 0.005071411082783605,
      "step": 1100
    },
    {
      "epoch": 1.26,
      "eval_blimp_supplement_avg": 0.8275862068965517,
      "eval_blimp_supplement_std": 0.016977103101302883,
      "step": 1100
    },
    {
      "epoch": 1.26,
      "eval_vqa_filtered_avg": 0.54,
      "eval_vqa_filtered_std": 0.05009082659620333,
      "step": 1100
    },
    {
      "epoch": 1.26,
      "eval_winoground_filtered_avg": 0.6,
      "eval_winoground_filtered_std": 0.04923659639173309,
      "step": 1100
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.5968080534283327,
      "learning_rate": 5.2604248873391154e-05,
      "loss": 0.9801,
      "step": 1101
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.9870386142915577,
      "learning_rate": 5.2589638457670346e-05,
      "loss": 1.2013,
      "step": 1102
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.7518360710073269,
      "learning_rate": 5.257501565789253e-05,
      "loss": 1.1717,
      "step": 1103
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.6994207674161943,
      "learning_rate": 5.2560380482074204e-05,
      "loss": 1.095,
      "step": 1104
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.5940968466662249,
      "learning_rate": 5.2545732938238644e-05,
      "loss": 1.0344,
      "step": 1105
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.581806015846707,
      "learning_rate": 5.2531073034415906e-05,
      "loss": 1.0463,
      "step": 1106
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.716778315245542,
      "learning_rate": 5.251640077864282e-05,
      "loss": 1.0986,
      "step": 1107
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5360185118680745,
      "learning_rate": 5.250171617896297e-05,
      "loss": 1.0688,
      "step": 1108
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.606670508467804,
      "learning_rate": 5.248701924342675e-05,
      "loss": 1.1072,
      "step": 1109
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.6223311876870439,
      "learning_rate": 5.247230998009128e-05,
      "loss": 1.0129,
      "step": 1110
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.6991365084245404,
      "learning_rate": 5.2457588397020454e-05,
      "loss": 1.1428,
      "step": 1111
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5657029900273567,
      "learning_rate": 5.24428545022849e-05,
      "loss": 0.982,
      "step": 1112
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.449247484960036,
      "learning_rate": 5.242810830396204e-05,
      "loss": 0.9568,
      "step": 1113
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.6756081980557056,
      "learning_rate": 5.241334981013598e-05,
      "loss": 1.1103,
      "step": 1114
    },
    {
      "epoch": 1.27,
      "grad_norm": 2.0076773971841906,
      "learning_rate": 5.239857902889761e-05,
      "loss": 1.0839,
      "step": 1115
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.9033782881555408,
      "learning_rate": 5.238379596834456e-05,
      "loss": 1.111,
      "step": 1116
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.7058617612332132,
      "learning_rate": 5.2369000636581174e-05,
      "loss": 1.0473,
      "step": 1117
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.618283151527136,
      "learning_rate": 5.2354193041718495e-05,
      "loss": 1.0525,
      "step": 1118
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.8531834368141942,
      "learning_rate": 5.2339373191874364e-05,
      "loss": 1.0784,
      "step": 1119
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.8346005038407804,
      "learning_rate": 5.232454109517326e-05,
      "loss": 1.2449,
      "step": 1120
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.9600816384617774,
      "learning_rate": 5.230969675974645e-05,
      "loss": 1.1592,
      "step": 1121
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6068885060238642,
      "learning_rate": 5.2294840193731826e-05,
      "loss": 1.0575,
      "step": 1122
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.4135828277237152,
      "learning_rate": 5.2279971405274076e-05,
      "loss": 1.0074,
      "step": 1123
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.5921609156357326,
      "learning_rate": 5.226509040252451e-05,
      "loss": 1.0957,
      "step": 1124
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.5016126041599331,
      "learning_rate": 5.225019719364119e-05,
      "loss": 0.9725,
      "step": 1125
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.7959766282274638,
      "learning_rate": 5.223529178678885e-05,
      "loss": 1.1251,
      "step": 1126
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.7293667562685,
      "learning_rate": 5.2220374190138894e-05,
      "loss": 1.1424,
      "step": 1127
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6038526539766962,
      "learning_rate": 5.2205444411869425e-05,
      "loss": 1.0297,
      "step": 1128
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.599871116690515,
      "learning_rate": 5.2190502460165245e-05,
      "loss": 1.0104,
      "step": 1129
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6357078986891072,
      "learning_rate": 5.217554834321778e-05,
      "loss": 1.0041,
      "step": 1130
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.7993901021378045,
      "learning_rate": 5.2160582069225164e-05,
      "loss": 1.1538,
      "step": 1131
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.9515714829318325,
      "learning_rate": 5.2145603646392186e-05,
      "loss": 1.1511,
      "step": 1132
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.540535280216864,
      "learning_rate": 5.2130613082930307e-05,
      "loss": 1.0733,
      "step": 1133
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.7211364511292435,
      "learning_rate": 5.21156103870576e-05,
      "loss": 1.025,
      "step": 1134
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.7976599027867057,
      "learning_rate": 5.210059556699885e-05,
      "loss": 1.1547,
      "step": 1135
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.665628947296069,
      "learning_rate": 5.2085568630985445e-05,
      "loss": 1.0477,
      "step": 1136
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.3267488530175222,
      "learning_rate": 5.2070529587255425e-05,
      "loss": 0.931,
      "step": 1137
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.5611345456791814,
      "learning_rate": 5.205547844405348e-05,
      "loss": 0.9852,
      "step": 1138
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.7526763853702998,
      "learning_rate": 5.204041520963092e-05,
      "loss": 1.1041,
      "step": 1139
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.531424122605453,
      "learning_rate": 5.202533989224569e-05,
      "loss": 0.955,
      "step": 1140
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.8619198954053342,
      "learning_rate": 5.201025250016236e-05,
      "loss": 1.1154,
      "step": 1141
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.81352287196681,
      "learning_rate": 5.199515304165212e-05,
      "loss": 1.0955,
      "step": 1142
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5935328954919554,
      "learning_rate": 5.1980041524992755e-05,
      "loss": 0.9882,
      "step": 1143
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.730959815969166,
      "learning_rate": 5.1964917958468684e-05,
      "loss": 1.0925,
      "step": 1144
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.6365943237359362,
      "learning_rate": 5.194978235037094e-05,
      "loss": 1.0179,
      "step": 1145
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.6653022540736782,
      "learning_rate": 5.193463470899714e-05,
      "loss": 1.12,
      "step": 1146
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.426981428219737,
      "learning_rate": 5.191947504265148e-05,
      "loss": 0.9328,
      "step": 1147
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.6947542635767672,
      "learning_rate": 5.1904303359644794e-05,
      "loss": 1.0539,
      "step": 1148
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.787187231788914,
      "learning_rate": 5.1889119668294456e-05,
      "loss": 1.1481,
      "step": 1149
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.8820543345040652,
      "learning_rate": 5.187392397692447e-05,
      "loss": 1.1196,
      "step": 1150
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.608694390894053,
      "learning_rate": 5.185871629386537e-05,
      "loss": 1.0953,
      "step": 1151
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.7050855324136476,
      "learning_rate": 5.184349662745431e-05,
      "loss": 1.1078,
      "step": 1152
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.682468291032329,
      "learning_rate": 5.182826498603498e-05,
      "loss": 0.9784,
      "step": 1153
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.692587250772672,
      "learning_rate": 5.1813021377957645e-05,
      "loss": 1.0522,
      "step": 1154
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.6182055695396564,
      "learning_rate": 5.179776581157915e-05,
      "loss": 1.0239,
      "step": 1155
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.5794002352200016,
      "learning_rate": 5.178249829526287e-05,
      "loss": 1.0496,
      "step": 1156
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.535505055078387,
      "learning_rate": 5.176721883737872e-05,
      "loss": 1.0605,
      "step": 1157
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.6007786535827955,
      "learning_rate": 5.17519274463032e-05,
      "loss": 1.0466,
      "step": 1158
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.7291185035367436,
      "learning_rate": 5.173662413041933e-05,
      "loss": 1.07,
      "step": 1159
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.6161021936260749,
      "learning_rate": 5.1721308898116675e-05,
      "loss": 1.1091,
      "step": 1160
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.6283396077698598,
      "learning_rate": 5.170598175779132e-05,
      "loss": 0.9494,
      "step": 1161
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.7474971295961792,
      "learning_rate": 5.1690642717845885e-05,
      "loss": 1.0684,
      "step": 1162
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.843806743916585,
      "learning_rate": 5.167529178668953e-05,
      "loss": 1.0754,
      "step": 1163
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.90199143200939,
      "learning_rate": 5.16599289727379e-05,
      "loss": 1.0384,
      "step": 1164
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.8061929392147957,
      "learning_rate": 5.1644554284413183e-05,
      "loss": 1.0623,
      "step": 1165
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.7573613017236274,
      "learning_rate": 5.1629167730144074e-05,
      "loss": 1.0484,
      "step": 1166
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.6146342901186728,
      "learning_rate": 5.161376931836574e-05,
      "loss": 1.0039,
      "step": 1167
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.717022312881657,
      "learning_rate": 5.159835905751991e-05,
      "loss": 1.0909,
      "step": 1168
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.6658416728551724,
      "learning_rate": 5.158293695605475e-05,
      "loss": 1.0108,
      "step": 1169
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.7074739662326848,
      "learning_rate": 5.156750302242496e-05,
      "loss": 1.122,
      "step": 1170
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.6041935935461318,
      "learning_rate": 5.155205726509167e-05,
      "loss": 1.0617,
      "step": 1171
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.6111391898287866,
      "learning_rate": 5.153659969252257e-05,
      "loss": 0.9946,
      "step": 1172
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.6215036840057286,
      "learning_rate": 5.1521130313191766e-05,
      "loss": 1.0321,
      "step": 1173
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.7074631019568756,
      "learning_rate": 5.150564913557987e-05,
      "loss": 1.1695,
      "step": 1174
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.6903308945744033,
      "learning_rate": 5.149015616817393e-05,
      "loss": 0.9445,
      "step": 1175
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.8974480117677128,
      "learning_rate": 5.14746514194675e-05,
      "loss": 1.0954,
      "step": 1176
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.6315862076674523,
      "learning_rate": 5.145913489796055e-05,
      "loss": 0.979,
      "step": 1177
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.5644097106484196,
      "learning_rate": 5.144360661215954e-05,
      "loss": 0.9689,
      "step": 1178
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.628606097000505,
      "learning_rate": 5.142806657057735e-05,
      "loss": 1.0395,
      "step": 1179
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.776281617228734,
      "learning_rate": 5.141251478173332e-05,
      "loss": 1.1096,
      "step": 1180
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.5868355870615478,
      "learning_rate": 5.1396951254153234e-05,
      "loss": 0.972,
      "step": 1181
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.6985474212957843,
      "learning_rate": 5.13813759963693e-05,
      "loss": 1.0689,
      "step": 1182
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.6804826233290762,
      "learning_rate": 5.136578901692017e-05,
      "loss": 1.0638,
      "step": 1183
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.66420368599728,
      "learning_rate": 5.1350190324350906e-05,
      "loss": 1.1435,
      "step": 1184
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.5428088661327832,
      "learning_rate": 5.1334579927213e-05,
      "loss": 0.9487,
      "step": 1185
    },
    {
      "epoch": 1.36,
      "grad_norm": 2.0153101590314177,
      "learning_rate": 5.131895783406436e-05,
      "loss": 1.1165,
      "step": 1186
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.7527097329103802,
      "learning_rate": 5.1303324053469313e-05,
      "loss": 1.1412,
      "step": 1187
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.8290181784624369,
      "learning_rate": 5.128767859399858e-05,
      "loss": 1.1377,
      "step": 1188
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.5340897228040673,
      "learning_rate": 5.127202146422928e-05,
      "loss": 0.9179,
      "step": 1189
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.5132823633827057,
      "learning_rate": 5.125635267274496e-05,
      "loss": 0.9562,
      "step": 1190
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.6352119404368228,
      "learning_rate": 5.124067222813553e-05,
      "loss": 0.9989,
      "step": 1191
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.9537847311523202,
      "learning_rate": 5.1224980138997304e-05,
      "loss": 1.173,
      "step": 1192
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.97333665799883,
      "learning_rate": 5.120927641393297e-05,
      "loss": 1.115,
      "step": 1193
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.533704181863378,
      "learning_rate": 5.119356106155161e-05,
      "loss": 0.9599,
      "step": 1194
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.6089595033498774,
      "learning_rate": 5.117783409046866e-05,
      "loss": 1.0296,
      "step": 1195
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.74414710485083,
      "learning_rate": 5.116209550930594e-05,
      "loss": 1.0395,
      "step": 1196
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.6057563052155392,
      "learning_rate": 5.1146345326691634e-05,
      "loss": 0.9652,
      "step": 1197
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.9344052859597132,
      "learning_rate": 5.113058355126027e-05,
      "loss": 1.0633,
      "step": 1198
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.7044238221445687,
      "learning_rate": 5.111481019165276e-05,
      "loss": 1.0881,
      "step": 1199
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5152068170137316,
      "learning_rate": 5.109902525651635e-05,
      "loss": 0.9296,
      "step": 1200
    },
    {
      "epoch": 1.37,
      "eval_blimp_filtered_avg": 0.7086567164179104,
      "eval_blimp_filtered_std": 0.005046346825935137,
      "step": 1200
    },
    {
      "epoch": 1.37,
      "eval_blimp_supplement_avg": 0.8318965517241379,
      "eval_blimp_supplement_std": 0.016745275376264798,
      "step": 1200
    },
    {
      "epoch": 1.37,
      "eval_vqa_filtered_avg": 0.53,
      "eval_vqa_filtered_std": 0.0501613558046592,
      "step": 1200
    },
    {
      "epoch": 1.37,
      "eval_winoground_filtered_avg": 0.6,
      "eval_winoground_filtered_std": 0.04923659639173309,
      "step": 1200
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.8544663570712472,
      "learning_rate": 5.108322875450462e-05,
      "loss": 1.0911,
      "step": 1201
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.6540895439250312,
      "learning_rate": 5.106742069427753e-05,
      "loss": 0.9421,
      "step": 1202
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.954446015935895,
      "learning_rate": 5.1051601084501315e-05,
      "loss": 1.1488,
      "step": 1203
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.4727738370637697,
      "learning_rate": 5.103576993384859e-05,
      "loss": 0.8266,
      "step": 1204
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.5501803773599123,
      "learning_rate": 5.10199272509983e-05,
      "loss": 0.9708,
      "step": 1205
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.781261303553768,
      "learning_rate": 5.100407304463569e-05,
      "loss": 1.0109,
      "step": 1206
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.6643983664730877,
      "learning_rate": 5.098820732345232e-05,
      "loss": 1.0011,
      "step": 1207
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.6450581032388598,
      "learning_rate": 5.0972330096146075e-05,
      "loss": 0.9944,
      "step": 1208
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.699945496723741,
      "learning_rate": 5.095644137142114e-05,
      "loss": 1.0159,
      "step": 1209
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.811100109395967,
      "learning_rate": 5.094054115798802e-05,
      "loss": 1.024,
      "step": 1210
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.5396411950542797,
      "learning_rate": 5.092462946456349e-05,
      "loss": 0.9014,
      "step": 1211
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.5998364699440353,
      "learning_rate": 5.090870629987065e-05,
      "loss": 1.0233,
      "step": 1212
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7483126752518943,
      "learning_rate": 5.089277167263886e-05,
      "loss": 1.0948,
      "step": 1213
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.8327844490509417,
      "learning_rate": 5.087682559160378e-05,
      "loss": 1.1288,
      "step": 1214
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7313311389134325,
      "learning_rate": 5.086086806550735e-05,
      "loss": 1.1261,
      "step": 1215
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.809543562993296,
      "learning_rate": 5.0844899103097776e-05,
      "loss": 1.0716,
      "step": 1216
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.6158564892870835,
      "learning_rate": 5.0828918713129554e-05,
      "loss": 1.0233,
      "step": 1217
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7899614189381468,
      "learning_rate": 5.08129269043634e-05,
      "loss": 1.0682,
      "step": 1218
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.629501736527151,
      "learning_rate": 5.079692368556634e-05,
      "loss": 1.0194,
      "step": 1219
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.6877017212699217,
      "learning_rate": 5.0780909065511634e-05,
      "loss": 1.0407,
      "step": 1220
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.6995820800134132,
      "learning_rate": 5.076488305297878e-05,
      "loss": 1.0553,
      "step": 1221
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.3758568185895714,
      "learning_rate": 5.0748845656753536e-05,
      "loss": 0.8655,
      "step": 1222
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.6083201366048465,
      "learning_rate": 5.073279688562792e-05,
      "loss": 0.9952,
      "step": 1223
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.694385123908182,
      "learning_rate": 5.071673674840014e-05,
      "loss": 1.0261,
      "step": 1224
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.7781369544053127,
      "learning_rate": 5.0700665253874665e-05,
      "loss": 1.014,
      "step": 1225
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.7171079636614208,
      "learning_rate": 5.068458241086219e-05,
      "loss": 1.0463,
      "step": 1226
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5705742145778945,
      "learning_rate": 5.0668488228179636e-05,
      "loss": 0.9469,
      "step": 1227
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.613550216773075,
      "learning_rate": 5.065238271465011e-05,
      "loss": 1.0288,
      "step": 1228
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5719670427473011,
      "learning_rate": 5.063626587910297e-05,
      "loss": 0.9078,
      "step": 1229
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.828479812879375,
      "learning_rate": 5.0620137730373743e-05,
      "loss": 1.0839,
      "step": 1230
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.519115939417359,
      "learning_rate": 5.06039982773042e-05,
      "loss": 0.859,
      "step": 1231
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.8288992086033111,
      "learning_rate": 5.058784752874227e-05,
      "loss": 1.0557,
      "step": 1232
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.6825412788383167,
      "learning_rate": 5.0571685493542105e-05,
      "loss": 0.8577,
      "step": 1233
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.4744442095767734,
      "learning_rate": 5.0555512180564026e-05,
      "loss": 0.9221,
      "step": 1234
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.856696923001526,
      "learning_rate": 5.053932759867453e-05,
      "loss": 1.0165,
      "step": 1235
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.7353577176009232,
      "learning_rate": 5.0523131756746325e-05,
      "loss": 0.9818,
      "step": 1236
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.6099870062272406,
      "learning_rate": 5.0506924663658264e-05,
      "loss": 0.9596,
      "step": 1237
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.6139788385937262,
      "learning_rate": 5.049070632829537e-05,
      "loss": 0.9632,
      "step": 1238
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.680549994509992,
      "learning_rate": 5.047447675954882e-05,
      "loss": 0.9611,
      "step": 1239
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.7048279917115787,
      "learning_rate": 5.0458235966316e-05,
      "loss": 0.9646,
      "step": 1240
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.5420716927385807,
      "learning_rate": 5.044198395750038e-05,
      "loss": 0.9284,
      "step": 1241
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.8195325251686554,
      "learning_rate": 5.042572074201163e-05,
      "loss": 1.1198,
      "step": 1242
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.485340064536036,
      "learning_rate": 5.040944632876552e-05,
      "loss": 0.9187,
      "step": 1243
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.5554586523599723,
      "learning_rate": 5.039316072668403e-05,
      "loss": 0.9699,
      "step": 1244
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.8140165026702841,
      "learning_rate": 5.0376863944695176e-05,
      "loss": 1.0975,
      "step": 1245
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.8458600631175988,
      "learning_rate": 5.036055599173318e-05,
      "loss": 1.0255,
      "step": 1246
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.8156366517899951,
      "learning_rate": 5.034423687673837e-05,
      "loss": 1.0028,
      "step": 1247
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.8097749847404128,
      "learning_rate": 5.032790660865718e-05,
      "loss": 0.9616,
      "step": 1248
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.8174563475703018,
      "learning_rate": 5.031156519644215e-05,
      "loss": 1.1156,
      "step": 1249
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.8411842580739477,
      "learning_rate": 5.029521264905197e-05,
      "loss": 1.0501,
      "step": 1250
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.713479272408884,
      "learning_rate": 5.0278848975451384e-05,
      "loss": 1.0473,
      "step": 1251
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.7335061901125457,
      "learning_rate": 5.0262474184611276e-05,
      "loss": 1.0038,
      "step": 1252
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.6319057687788001,
      "learning_rate": 5.024608828550861e-05,
      "loss": 1.0101,
      "step": 1253
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.7700602789448585,
      "learning_rate": 5.022969128712642e-05,
      "loss": 1.0429,
      "step": 1254
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.7811899965719522,
      "learning_rate": 5.021328319845386e-05,
      "loss": 1.0519,
      "step": 1255
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6675448052900101,
      "learning_rate": 5.019686402848615e-05,
      "loss": 1.0572,
      "step": 1256
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6232436578095046,
      "learning_rate": 5.0180433786224555e-05,
      "loss": 0.9806,
      "step": 1257
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6045391894295171,
      "learning_rate": 5.0163992480676474e-05,
      "loss": 1.0212,
      "step": 1258
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.9181240362393215,
      "learning_rate": 5.0147540120855315e-05,
      "loss": 0.968,
      "step": 1259
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.8037303723183211,
      "learning_rate": 5.0131076715780563e-05,
      "loss": 1.0973,
      "step": 1260
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.7635486889817504,
      "learning_rate": 5.011460227447777e-05,
      "loss": 1.0849,
      "step": 1261
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.8395607222805435,
      "learning_rate": 5.009811680597853e-05,
      "loss": 1.0843,
      "step": 1262
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.5966194984449098,
      "learning_rate": 5.0081620319320464e-05,
      "loss": 0.986,
      "step": 1263
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.8025186042540533,
      "learning_rate": 5.006511282354727e-05,
      "loss": 1.0429,
      "step": 1264
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.604994453064332,
      "learning_rate": 5.004859432770865e-05,
      "loss": 1.0518,
      "step": 1265
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.6988719909191496,
      "learning_rate": 5.0032064840860354e-05,
      "loss": 1.0757,
      "step": 1266
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.5743243842562589,
      "learning_rate": 5.001552437206416e-05,
      "loss": 0.9957,
      "step": 1267
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.7853065447439143,
      "learning_rate": 4.999897293038783e-05,
      "loss": 1.1463,
      "step": 1268
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.5986296644080582,
      "learning_rate": 4.9982410524905204e-05,
      "loss": 0.9498,
      "step": 1269
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.6577220874808092,
      "learning_rate": 4.996583716469608e-05,
      "loss": 0.9664,
      "step": 1270
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.4777143169287965,
      "learning_rate": 4.994925285884629e-05,
      "loss": 0.9613,
      "step": 1271
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3810650420792376,
      "learning_rate": 4.993265761644765e-05,
      "loss": 0.9388,
      "step": 1272
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.442637482470765,
      "learning_rate": 4.991605144659797e-05,
      "loss": 0.9017,
      "step": 1273
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.8204006434373043,
      "learning_rate": 4.989943435840107e-05,
      "loss": 1.1572,
      "step": 1274
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.6652139773639314,
      "learning_rate": 4.9882806360966756e-05,
      "loss": 0.9987,
      "step": 1275
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.6984381790084275,
      "learning_rate": 4.9866167463410773e-05,
      "loss": 0.9344,
      "step": 1276
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.5550615686019407,
      "learning_rate": 4.984951767485491e-05,
      "loss": 1.0318,
      "step": 1277
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.7689913014727845,
      "learning_rate": 4.9832857004426856e-05,
      "loss": 0.9757,
      "step": 1278
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.6446835541890545,
      "learning_rate": 4.981618546126031e-05,
      "loss": 1.0059,
      "step": 1279
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.698703989193505,
      "learning_rate": 4.979950305449491e-05,
      "loss": 0.9728,
      "step": 1280
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.713172458423019,
      "learning_rate": 4.9782809793276284e-05,
      "loss": 0.9957,
      "step": 1281
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.7251176822990617,
      "learning_rate": 4.976610568675597e-05,
      "loss": 0.9697,
      "step": 1282
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.8795057648420033,
      "learning_rate": 4.974939074409146e-05,
      "loss": 1.0984,
      "step": 1283
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.720795003510563,
      "learning_rate": 4.973266497444621e-05,
      "loss": 1.004,
      "step": 1284
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.7015990868725706,
      "learning_rate": 4.9715928386989593e-05,
      "loss": 1.0467,
      "step": 1285
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.6872741879379727,
      "learning_rate": 4.9699180990896905e-05,
      "loss": 0.9872,
      "step": 1286
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.4535074096168135,
      "learning_rate": 4.9682422795349386e-05,
      "loss": 0.9865,
      "step": 1287
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.667097644194145,
      "learning_rate": 4.9665653809534195e-05,
      "loss": 0.9938,
      "step": 1288
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.982594836642423,
      "learning_rate": 4.964887404264438e-05,
      "loss": 1.1672,
      "step": 1289
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.625949669592417,
      "learning_rate": 4.963208350387893e-05,
      "loss": 0.9124,
      "step": 1290
    },
    {
      "epoch": 1.48,
      "grad_norm": 2.0438421851233115,
      "learning_rate": 4.961528220244274e-05,
      "loss": 1.084,
      "step": 1291
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.7667787744076966,
      "learning_rate": 4.9598470147546574e-05,
      "loss": 1.0765,
      "step": 1292
    },
    {
      "epoch": 1.48,
      "grad_norm": 2.0533582227293508,
      "learning_rate": 4.958164734840713e-05,
      "loss": 1.1239,
      "step": 1293
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.7153618410938956,
      "learning_rate": 4.956481381424696e-05,
      "loss": 1.0541,
      "step": 1294
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.7887508171432167,
      "learning_rate": 4.9547969554294514e-05,
      "loss": 1.0898,
      "step": 1295
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.5849010417831215,
      "learning_rate": 4.9531114577784154e-05,
      "loss": 0.9479,
      "step": 1296
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.811411116492653,
      "learning_rate": 4.951424889395607e-05,
      "loss": 1.0902,
      "step": 1297
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.6251088279930217,
      "learning_rate": 4.949737251205634e-05,
      "loss": 0.8716,
      "step": 1298
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.5650065078121858,
      "learning_rate": 4.9480485441336916e-05,
      "loss": 0.9737,
      "step": 1299
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.7178316985489392,
      "learning_rate": 4.946358769105559e-05,
      "loss": 1.0072,
      "step": 1300
    },
    {
      "epoch": 1.49,
      "eval_blimp_filtered_avg": 0.7013432835820895,
      "eval_blimp_filtered_std": 0.005076466707385747,
      "step": 1300
    },
    {
      "epoch": 1.49,
      "eval_blimp_supplement_avg": 0.8254310344827587,
      "eval_blimp_supplement_std": 0.017088747032722257,
      "step": 1300
    },
    {
      "epoch": 1.49,
      "eval_vqa_filtered_avg": 0.53,
      "eval_vqa_filtered_std": 0.0501613558046592,
      "step": 1300
    },
    {
      "epoch": 1.49,
      "eval_winoground_filtered_avg": 0.64,
      "eval_winoground_filtered_std": 0.048241815132442176,
      "step": 1300
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.7564242912196177,
      "learning_rate": 4.944667927047604e-05,
      "loss": 1.0596,
      "step": 1301
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.8078259033710278,
      "learning_rate": 4.9429760188867764e-05,
      "loss": 1.0595,
      "step": 1302
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.810030513284643,
      "learning_rate": 4.9412830455506114e-05,
      "loss": 1.0494,
      "step": 1303
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.74597563857238,
      "learning_rate": 4.9395890079672277e-05,
      "loss": 0.9424,
      "step": 1304
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.5426904386165736,
      "learning_rate": 4.937893907065329e-05,
      "loss": 0.9055,
      "step": 1305
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.423555775669211,
      "learning_rate": 4.9361977437742e-05,
      "loss": 0.7992,
      "step": 1306
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.7593691847412536,
      "learning_rate": 4.9345005190237105e-05,
      "loss": 1.0526,
      "step": 1307
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.9233121881383441,
      "learning_rate": 4.9328022337443086e-05,
      "loss": 1.0773,
      "step": 1308
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.7426916304504436,
      "learning_rate": 4.931102888867026e-05,
      "loss": 1.0113,
      "step": 1309
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.823971827375318,
      "learning_rate": 4.929402485323476e-05,
      "loss": 1.0694,
      "step": 1310
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.5201299014504215,
      "learning_rate": 4.9277010240458495e-05,
      "loss": 0.8881,
      "step": 1311
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.5856511790933685,
      "learning_rate": 4.9259985059669214e-05,
      "loss": 0.887,
      "step": 1312
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.608240711326309,
      "learning_rate": 4.9242949320200406e-05,
      "loss": 0.9591,
      "step": 1313
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.6174743092885837,
      "learning_rate": 4.92259030313914e-05,
      "loss": 0.9818,
      "step": 1314
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.824276248825837,
      "learning_rate": 4.920884620258727e-05,
      "loss": 1.0681,
      "step": 1315
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.5307303669623644,
      "learning_rate": 4.919177884313891e-05,
      "loss": 0.8964,
      "step": 1316
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.670927755660054,
      "learning_rate": 4.9174700962402936e-05,
      "loss": 1.0021,
      "step": 1317
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.8304606397969936,
      "learning_rate": 4.9157612569741775e-05,
      "loss": 1.0308,
      "step": 1318
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.9248866094653747,
      "learning_rate": 4.914051367452358e-05,
      "loss": 1.0515,
      "step": 1319
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.7794864750808854,
      "learning_rate": 4.9123404286122304e-05,
      "loss": 0.9271,
      "step": 1320
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.6544279029587163,
      "learning_rate": 4.910628441391762e-05,
      "loss": 0.9961,
      "step": 1321
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.9396948088317298,
      "learning_rate": 4.908915406729494e-05,
      "loss": 1.0996,
      "step": 1322
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.4787313354485454,
      "learning_rate": 4.9072013255645476e-05,
      "loss": 0.8933,
      "step": 1323
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.6042788559709478,
      "learning_rate": 4.90548619883661e-05,
      "loss": 0.9209,
      "step": 1324
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.4978174422658417,
      "learning_rate": 4.9037700274859474e-05,
      "loss": 0.867,
      "step": 1325
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.8582466894191032,
      "learning_rate": 4.9020528124533954e-05,
      "loss": 1.0328,
      "step": 1326
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.9232693318862748,
      "learning_rate": 4.9003345546803646e-05,
      "loss": 0.9048,
      "step": 1327
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.5649022036157152,
      "learning_rate": 4.8986152551088334e-05,
      "loss": 0.9604,
      "step": 1328
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.6230836152955854,
      "learning_rate": 4.896894914681356e-05,
      "loss": 1.0114,
      "step": 1329
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4480741409877889,
      "learning_rate": 4.895173534341054e-05,
      "loss": 0.9563,
      "step": 1330
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.547661699233136,
      "learning_rate": 4.893451115031619e-05,
      "loss": 0.9655,
      "step": 1331
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.6843488559741597,
      "learning_rate": 4.891727657697313e-05,
      "loss": 1.0778,
      "step": 1332
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.7367684156448366,
      "learning_rate": 4.8900031632829686e-05,
      "loss": 1.0238,
      "step": 1333
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.737334303504081,
      "learning_rate": 4.888277632733984e-05,
      "loss": 1.0133,
      "step": 1334
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.7768400600429264,
      "learning_rate": 4.886551066996329e-05,
      "loss": 1.0257,
      "step": 1335
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.6867628280756912,
      "learning_rate": 4.884823467016536e-05,
      "loss": 0.9725,
      "step": 1336
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.4380452481194135,
      "learning_rate": 4.883094833741709e-05,
      "loss": 0.8638,
      "step": 1337
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.5814998086415863,
      "learning_rate": 4.8813651681195154e-05,
      "loss": 0.8954,
      "step": 1338
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.7864936357623085,
      "learning_rate": 4.879634471098192e-05,
      "loss": 1.0318,
      "step": 1339
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.5681692139329506,
      "learning_rate": 4.877902743626537e-05,
      "loss": 0.8241,
      "step": 1340
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.4574729952980203,
      "learning_rate": 4.8761699866539155e-05,
      "loss": 0.8465,
      "step": 1341
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.2667304725477289,
      "learning_rate": 4.874436201130259e-05,
      "loss": 0.7304,
      "step": 1342
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.6384692329221817,
      "learning_rate": 4.8727013880060576e-05,
      "loss": 0.8945,
      "step": 1343
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.8402940821900626,
      "learning_rate": 4.87096554823237e-05,
      "loss": 1.0268,
      "step": 1344
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.6135505291128451,
      "learning_rate": 4.8692286827608154e-05,
      "loss": 0.9454,
      "step": 1345
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.445992359120777,
      "learning_rate": 4.867490792543575e-05,
      "loss": 0.8583,
      "step": 1346
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.5514626913272305,
      "learning_rate": 4.865751878533392e-05,
      "loss": 0.8917,
      "step": 1347
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.547649130254401,
      "learning_rate": 4.8640119416835715e-05,
      "loss": 0.8804,
      "step": 1348
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.6822037097554143,
      "learning_rate": 4.86227098294798e-05,
      "loss": 0.9919,
      "step": 1349
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.459608010832908,
      "learning_rate": 4.860529003281042e-05,
      "loss": 0.8131,
      "step": 1350
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.7069200790019363,
      "learning_rate": 4.858786003637743e-05,
      "loss": 0.9424,
      "step": 1351
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.706434761136003,
      "learning_rate": 4.857041984973628e-05,
      "loss": 1.0155,
      "step": 1352
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5637092031541717,
      "learning_rate": 4.8552969482448e-05,
      "loss": 0.918,
      "step": 1353
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5036152017341191,
      "learning_rate": 4.853550894407921e-05,
      "loss": 0.8653,
      "step": 1354
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.7324980213984802,
      "learning_rate": 4.85180382442021e-05,
      "loss": 1.0218,
      "step": 1355
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.9162444604876634,
      "learning_rate": 4.850055739239441e-05,
      "loss": 1.0209,
      "step": 1356
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.6706711116380952,
      "learning_rate": 4.848306639823949e-05,
      "loss": 0.8998,
      "step": 1357
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5858744713136999,
      "learning_rate": 4.846556527132621e-05,
      "loss": 0.8494,
      "step": 1358
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.6347441880043818,
      "learning_rate": 4.844805402124903e-05,
      "loss": 0.9597,
      "step": 1359
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.4764201877004797,
      "learning_rate": 4.8430532657607924e-05,
      "loss": 0.9145,
      "step": 1360
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.585237797775831,
      "learning_rate": 4.841300119000843e-05,
      "loss": 0.9028,
      "step": 1361
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.797340212784329,
      "learning_rate": 4.839545962806162e-05,
      "loss": 0.9228,
      "step": 1362
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.8310572910060163,
      "learning_rate": 4.837790798138412e-05,
      "loss": 1.0014,
      "step": 1363
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.6255025531451495,
      "learning_rate": 4.836034625959804e-05,
      "loss": 0.8664,
      "step": 1364
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.7625247525240582,
      "learning_rate": 4.834277447233106e-05,
      "loss": 1.01,
      "step": 1365
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.4735440273265845,
      "learning_rate": 4.832519262921634e-05,
      "loss": 0.876,
      "step": 1366
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.7110316910770746,
      "learning_rate": 4.8307600739892595e-05,
      "loss": 0.9755,
      "step": 1367
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.685324007194489,
      "learning_rate": 4.8289998814004e-05,
      "loss": 0.9529,
      "step": 1368
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.379988935684883,
      "learning_rate": 4.827238686120028e-05,
      "loss": 0.8584,
      "step": 1369
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.6957951874041441,
      "learning_rate": 4.825476489113662e-05,
      "loss": 1.0482,
      "step": 1370
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.6210127607139255,
      "learning_rate": 4.8237132913473704e-05,
      "loss": 0.8829,
      "step": 1371
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.6225232093587842,
      "learning_rate": 4.821949093787771e-05,
      "loss": 0.9991,
      "step": 1372
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.4307944712005054,
      "learning_rate": 4.8201838974020297e-05,
      "loss": 0.9392,
      "step": 1373
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.6065320829280063,
      "learning_rate": 4.8184177031578604e-05,
      "loss": 0.9446,
      "step": 1374
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.4887417729987247,
      "learning_rate": 4.816650512023523e-05,
      "loss": 0.8221,
      "step": 1375
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.3691726249407214,
      "learning_rate": 4.814882324967824e-05,
      "loss": 0.8325,
      "step": 1376
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.7114951911422331,
      "learning_rate": 4.813113142960117e-05,
      "loss": 0.9605,
      "step": 1377
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.7378387844454724,
      "learning_rate": 4.8113429669702986e-05,
      "loss": 0.9777,
      "step": 1378
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.5574739006287899,
      "learning_rate": 4.8095717979688126e-05,
      "loss": 0.883,
      "step": 1379
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.6336312910604611,
      "learning_rate": 4.8077996369266485e-05,
      "loss": 0.8867,
      "step": 1380
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.823327199308181,
      "learning_rate": 4.8060264848153354e-05,
      "loss": 0.9923,
      "step": 1381
    },
    {
      "epoch": 1.58,
      "grad_norm": 2.0452263581176586,
      "learning_rate": 4.8042523426069494e-05,
      "loss": 1.0765,
      "step": 1382
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.7596822491697843,
      "learning_rate": 4.8024772112741066e-05,
      "loss": 0.942,
      "step": 1383
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.4988701946497511,
      "learning_rate": 4.8007010917899685e-05,
      "loss": 0.8797,
      "step": 1384
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.5414614745345605,
      "learning_rate": 4.798923985128235e-05,
      "loss": 0.8798,
      "step": 1385
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.4878387023593527,
      "learning_rate": 4.7971458922631497e-05,
      "loss": 0.869,
      "step": 1386
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4662429089545088,
      "learning_rate": 4.795366814169496e-05,
      "loss": 0.8927,
      "step": 1387
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4309486560753406,
      "learning_rate": 4.7935867518225964e-05,
      "loss": 0.8318,
      "step": 1388
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.510415207994616,
      "learning_rate": 4.7918057061983154e-05,
      "loss": 0.8975,
      "step": 1389
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4437978281584951,
      "learning_rate": 4.790023678273055e-05,
      "loss": 0.8796,
      "step": 1390
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.73536331579817,
      "learning_rate": 4.788240669023753e-05,
      "loss": 0.9233,
      "step": 1391
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.8027642004965732,
      "learning_rate": 4.7864566794278925e-05,
      "loss": 0.9452,
      "step": 1392
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.9202451891843415,
      "learning_rate": 4.784671710463486e-05,
      "loss": 0.9597,
      "step": 1393
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.5582967179739577,
      "learning_rate": 4.782885763109088e-05,
      "loss": 0.9104,
      "step": 1394
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.8509954577556187,
      "learning_rate": 4.781098838343788e-05,
      "loss": 1.0588,
      "step": 1395
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.4337531107407828,
      "learning_rate": 4.77931093714721e-05,
      "loss": 0.8585,
      "step": 1396
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.7394634648050262,
      "learning_rate": 4.777522060499515e-05,
      "loss": 0.9507,
      "step": 1397
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6308028070317244,
      "learning_rate": 4.775732209381399e-05,
      "loss": 0.8982,
      "step": 1398
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.5600523839288876,
      "learning_rate": 4.773941384774091e-05,
      "loss": 0.8989,
      "step": 1399
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.61981309522277,
      "learning_rate": 4.7721495876593537e-05,
      "loss": 0.9308,
      "step": 1400
    },
    {
      "epoch": 1.6,
      "eval_blimp_filtered_avg": 0.7116417910447761,
      "eval_blimp_filtered_std": 0.0049947599851199054,
      "step": 1400
    },
    {
      "epoch": 1.6,
      "eval_blimp_supplement_avg": 0.8448275862068966,
      "eval_blimp_supplement_std": 0.016256557740155556,
      "step": 1400
    },
    {
      "epoch": 1.6,
      "eval_vqa_filtered_avg": 0.53,
      "eval_vqa_filtered_std": 0.0501613558046592,
      "step": 1400
    },
    {
      "epoch": 1.6,
      "eval_winoground_filtered_avg": 0.6,
      "eval_winoground_filtered_std": 0.04923659639173309,
      "step": 1400
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.5994323269365207,
      "learning_rate": 4.770356819019483e-05,
      "loss": 0.8718,
      "step": 1401
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.3966033766162596,
      "learning_rate": 4.7685630798373085e-05,
      "loss": 0.8614,
      "step": 1402
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6688052053494857,
      "learning_rate": 4.766768371096192e-05,
      "loss": 0.9481,
      "step": 1403
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.8301816822022643,
      "learning_rate": 4.7649726937800236e-05,
      "loss": 1.0669,
      "step": 1404
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.578213502358382,
      "learning_rate": 4.763176048873229e-05,
      "loss": 0.9304,
      "step": 1405
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6143150654582008,
      "learning_rate": 4.761378437360761e-05,
      "loss": 0.8494,
      "step": 1406
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6405741818956894,
      "learning_rate": 4.759579860228104e-05,
      "loss": 0.9621,
      "step": 1407
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.768306001555728,
      "learning_rate": 4.75778031846127e-05,
      "loss": 0.9472,
      "step": 1408
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.7830655738611016,
      "learning_rate": 4.755979813046804e-05,
      "loss": 0.9212,
      "step": 1409
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6230540351265126,
      "learning_rate": 4.754178344971772e-05,
      "loss": 0.8791,
      "step": 1410
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.792726352981723,
      "learning_rate": 4.752375915223775e-05,
      "loss": 0.8951,
      "step": 1411
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.7696743595933486,
      "learning_rate": 4.750572524790939e-05,
      "loss": 0.957,
      "step": 1412
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.5009370986981443,
      "learning_rate": 4.748768174661913e-05,
      "loss": 0.8339,
      "step": 1413
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.641464409331535,
      "learning_rate": 4.746962865825877e-05,
      "loss": 0.8822,
      "step": 1414
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.6449675388156872,
      "learning_rate": 4.745156599272534e-05,
      "loss": 0.9401,
      "step": 1415
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.7119688576750811,
      "learning_rate": 4.743349375992113e-05,
      "loss": 0.9612,
      "step": 1416
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.5854116782081484,
      "learning_rate": 4.741541196975368e-05,
      "loss": 0.8728,
      "step": 1417
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.683766316367014,
      "learning_rate": 4.7397320632135744e-05,
      "loss": 0.9901,
      "step": 1418
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.6479238409126953,
      "learning_rate": 4.7379219756985336e-05,
      "loss": 0.9357,
      "step": 1419
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.8188868444890702,
      "learning_rate": 4.736110935422569e-05,
      "loss": 1.0002,
      "step": 1420
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.7436553771487318,
      "learning_rate": 4.7342989433785265e-05,
      "loss": 1.0673,
      "step": 1421
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6021703428804868,
      "learning_rate": 4.732486000559773e-05,
      "loss": 0.9301,
      "step": 1422
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6484309155002626,
      "learning_rate": 4.730672107960199e-05,
      "loss": 0.9327,
      "step": 1423
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.8968799081345202,
      "learning_rate": 4.7288572665742116e-05,
      "loss": 1.0042,
      "step": 1424
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6161067343881526,
      "learning_rate": 4.7270414773967417e-05,
      "loss": 0.85,
      "step": 1425
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.588772764162754,
      "learning_rate": 4.725224741423237e-05,
      "loss": 0.8156,
      "step": 1426
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.619917905166706,
      "learning_rate": 4.723407059649668e-05,
      "loss": 0.8889,
      "step": 1427
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.90712158692843,
      "learning_rate": 4.72158843307252e-05,
      "loss": 0.9952,
      "step": 1428
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.758083860919103,
      "learning_rate": 4.7197688626887984e-05,
      "loss": 0.8714,
      "step": 1429
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.9005680618289889,
      "learning_rate": 4.717948349496024e-05,
      "loss": 1.0261,
      "step": 1430
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.6245096376329258,
      "learning_rate": 4.716126894492237e-05,
      "loss": 0.9422,
      "step": 1431
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.4924574120363516,
      "learning_rate": 4.7143044986759915e-05,
      "loss": 0.8207,
      "step": 1432
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.5836438917537368,
      "learning_rate": 4.712481163046361e-05,
      "loss": 0.917,
      "step": 1433
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.6031189459195574,
      "learning_rate": 4.710656888602928e-05,
      "loss": 0.9458,
      "step": 1434
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.986252501164731,
      "learning_rate": 4.7088316763457966e-05,
      "loss": 0.9365,
      "step": 1435
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.7760134985289833,
      "learning_rate": 4.7070055272755795e-05,
      "loss": 1.0221,
      "step": 1436
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.6314504818011493,
      "learning_rate": 4.705178442393409e-05,
      "loss": 0.8642,
      "step": 1437
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.718106613231125,
      "learning_rate": 4.703350422700922e-05,
      "loss": 0.9561,
      "step": 1438
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.770383228764284,
      "learning_rate": 4.701521469200277e-05,
      "loss": 0.9777,
      "step": 1439
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.5488112517709751,
      "learning_rate": 4.699691582894137e-05,
      "loss": 0.916,
      "step": 1440
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.567049317478263,
      "learning_rate": 4.697860764785681e-05,
      "loss": 0.8259,
      "step": 1441
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.6371999803461017,
      "learning_rate": 4.6960290158785966e-05,
      "loss": 0.9476,
      "step": 1442
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.7677535016430783,
      "learning_rate": 4.694196337177082e-05,
      "loss": 0.9783,
      "step": 1443
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.6110254922243958,
      "learning_rate": 4.692362729685847e-05,
      "loss": 0.8435,
      "step": 1444
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.5933991512142578,
      "learning_rate": 4.690528194410109e-05,
      "loss": 0.8497,
      "step": 1445
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.5492863233493097,
      "learning_rate": 4.688692732355592e-05,
      "loss": 0.794,
      "step": 1446
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.5096862910370323,
      "learning_rate": 4.6868563445285315e-05,
      "loss": 0.8506,
      "step": 1447
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.5564278319041585,
      "learning_rate": 4.685019031935671e-05,
      "loss": 0.8664,
      "step": 1448
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.8204798670710853,
      "learning_rate": 4.683180795584256e-05,
      "loss": 0.9591,
      "step": 1449
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.734141248596898,
      "learning_rate": 4.681341636482044e-05,
      "loss": 0.9089,
      "step": 1450
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.796787771706782,
      "learning_rate": 4.6795015556372944e-05,
      "loss": 0.9253,
      "step": 1451
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.726893168143386,
      "learning_rate": 4.6776605540587746e-05,
      "loss": 0.8627,
      "step": 1452
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.368527493743542,
      "learning_rate": 4.675818632755755e-05,
      "loss": 0.8266,
      "step": 1453
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.7502471366220016,
      "learning_rate": 4.673975792738011e-05,
      "loss": 0.9794,
      "step": 1454
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.5222611160513144,
      "learning_rate": 4.672132035015823e-05,
      "loss": 0.8975,
      "step": 1455
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.638634607620076,
      "learning_rate": 4.67028736059997e-05,
      "loss": 0.9375,
      "step": 1456
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.610741292032599,
      "learning_rate": 4.668441770501739e-05,
      "loss": 0.9117,
      "step": 1457
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.5362516989685095,
      "learning_rate": 4.666595265732917e-05,
      "loss": 0.8439,
      "step": 1458
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.6304407442602293,
      "learning_rate": 4.6647478473057896e-05,
      "loss": 0.8871,
      "step": 1459
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.9876828909186621,
      "learning_rate": 4.662899516233147e-05,
      "loss": 1.0492,
      "step": 1460
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.60954846272114,
      "learning_rate": 4.661050273528279e-05,
      "loss": 0.8679,
      "step": 1461
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.6754517783964606,
      "learning_rate": 4.6592001202049735e-05,
      "loss": 0.8765,
      "step": 1462
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.9570761059799089,
      "learning_rate": 4.657349057277518e-05,
      "loss": 1.0294,
      "step": 1463
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.9305684243375512,
      "learning_rate": 4.655497085760703e-05,
      "loss": 0.9702,
      "step": 1464
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.711325855798691,
      "learning_rate": 4.6536442066698104e-05,
      "loss": 0.9293,
      "step": 1465
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.4907015542833537,
      "learning_rate": 4.651790421020623e-05,
      "loss": 0.7949,
      "step": 1466
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.484618861560881,
      "learning_rate": 4.6499357298294206e-05,
      "loss": 0.9283,
      "step": 1467
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.5426722125144767,
      "learning_rate": 4.64808013411298e-05,
      "loss": 0.9085,
      "step": 1468
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.4772613168371718,
      "learning_rate": 4.6462236348885724e-05,
      "loss": 0.8637,
      "step": 1469
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.4991844491413344,
      "learning_rate": 4.644366233173965e-05,
      "loss": 0.8983,
      "step": 1470
    },
    {
      "epoch": 1.68,
      "grad_norm": 2.035471020113462,
      "learning_rate": 4.642507929987419e-05,
      "loss": 1.0788,
      "step": 1471
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.9021205946696405,
      "learning_rate": 4.640648726347692e-05,
      "loss": 1.0487,
      "step": 1472
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.4714494276832044,
      "learning_rate": 4.6387886232740326e-05,
      "loss": 0.8371,
      "step": 1473
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.374898744572044,
      "learning_rate": 4.636927621786184e-05,
      "loss": 0.816,
      "step": 1474
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.839588184089671,
      "learning_rate": 4.635065722904381e-05,
      "loss": 0.9846,
      "step": 1475
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.5119201378081346,
      "learning_rate": 4.6332029276493525e-05,
      "loss": 0.8722,
      "step": 1476
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.5603886785635268,
      "learning_rate": 4.631339237042316e-05,
      "loss": 0.9265,
      "step": 1477
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.8231686444513178,
      "learning_rate": 4.629474652104982e-05,
      "loss": 0.9942,
      "step": 1478
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.8745068628917059,
      "learning_rate": 4.627609173859549e-05,
      "loss": 0.9899,
      "step": 1479
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.987530158304971,
      "learning_rate": 4.6257428033287086e-05,
      "loss": 1.0011,
      "step": 1480
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.7177616373923081,
      "learning_rate": 4.6238755415356374e-05,
      "loss": 0.8837,
      "step": 1481
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.7291467325644365,
      "learning_rate": 4.622007389504003e-05,
      "loss": 0.9498,
      "step": 1482
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.6414332796657072,
      "learning_rate": 4.6201383482579635e-05,
      "loss": 0.9658,
      "step": 1483
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.4188674574523719,
      "learning_rate": 4.618268418822159e-05,
      "loss": 0.7811,
      "step": 1484
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.6509422702769447,
      "learning_rate": 4.616397602221721e-05,
      "loss": 0.9631,
      "step": 1485
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.4646996467918032,
      "learning_rate": 4.614525899482266e-05,
      "loss": 0.8679,
      "step": 1486
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.5838249321623894,
      "learning_rate": 4.612653311629893e-05,
      "loss": 0.8523,
      "step": 1487
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.5660591516283624,
      "learning_rate": 4.610779839691192e-05,
      "loss": 0.9523,
      "step": 1488
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.5389363870629298,
      "learning_rate": 4.608905484693235e-05,
      "loss": 0.839,
      "step": 1489
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.6609581093252064,
      "learning_rate": 4.607030247663578e-05,
      "loss": 0.8485,
      "step": 1490
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.4651521335944375,
      "learning_rate": 4.605154129630258e-05,
      "loss": 0.8541,
      "step": 1491
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.6803958723227412,
      "learning_rate": 4.6032771316218004e-05,
      "loss": 1.0116,
      "step": 1492
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.9312598032806265,
      "learning_rate": 4.601399254667209e-05,
      "loss": 0.9456,
      "step": 1493
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.5486832156237615,
      "learning_rate": 4.599520499795971e-05,
      "loss": 0.8889,
      "step": 1494
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.7726343945279441,
      "learning_rate": 4.597640868038054e-05,
      "loss": 1.0097,
      "step": 1495
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.6959469540135081,
      "learning_rate": 4.595760360423908e-05,
      "loss": 0.9245,
      "step": 1496
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.967191968981989,
      "learning_rate": 4.59387897798446e-05,
      "loss": 1.0502,
      "step": 1497
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.5873884264412195,
      "learning_rate": 4.591996721751121e-05,
      "loss": 0.8529,
      "step": 1498
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.7094800270362576,
      "learning_rate": 4.5901135927557756e-05,
      "loss": 0.9664,
      "step": 1499
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.7666652655343271,
      "learning_rate": 4.588229592030794e-05,
      "loss": 0.9328,
      "step": 1500
    },
    {
      "epoch": 1.71,
      "eval_blimp_filtered_avg": 0.7104477611940299,
      "eval_blimp_filtered_std": 0.004991695604314401,
      "step": 1500
    },
    {
      "epoch": 1.71,
      "eval_blimp_supplement_avg": 0.8211206896551724,
      "eval_blimp_supplement_std": 0.01729291315622075,
      "step": 1500
    },
    {
      "epoch": 1.71,
      "eval_vqa_filtered_avg": 0.53,
      "eval_vqa_filtered_std": 0.0501613558046592,
      "step": 1500
    },
    {
      "epoch": 1.71,
      "eval_winoground_filtered_avg": 0.61,
      "eval_winoground_filtered_std": 0.04902071300001975,
      "step": 1500
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.660552174803649,
      "learning_rate": 4.5863447206090164e-05,
      "loss": 0.9293,
      "step": 1501
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.6975652399879069,
      "learning_rate": 4.584458979523766e-05,
      "loss": 0.8825,
      "step": 1502
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.844163007434069,
      "learning_rate": 4.582572369808841e-05,
      "loss": 0.9671,
      "step": 1503
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.5086948287135693,
      "learning_rate": 4.580684892498515e-05,
      "loss": 0.8495,
      "step": 1504
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.5308215844269868,
      "learning_rate": 4.578796548627538e-05,
      "loss": 0.8573,
      "step": 1505
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.6985540533626,
      "learning_rate": 4.5769073392311354e-05,
      "loss": 0.9188,
      "step": 1506
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2502122745455266,
      "learning_rate": 4.575017265345006e-05,
      "loss": 0.7313,
      "step": 1507
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.5743749239713212,
      "learning_rate": 4.5731263280053235e-05,
      "loss": 0.8775,
      "step": 1508
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.6604010070162538,
      "learning_rate": 4.5712345282487354e-05,
      "loss": 0.9045,
      "step": 1509
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.7674869576742955,
      "learning_rate": 4.56934186711236e-05,
      "loss": 1.0048,
      "step": 1510
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.4894967840777413,
      "learning_rate": 4.56744834563379e-05,
      "loss": 0.8595,
      "step": 1511
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.8848833052588594,
      "learning_rate": 4.565553964851088e-05,
      "loss": 0.9361,
      "step": 1512
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.6222113020833702,
      "learning_rate": 4.563658725802791e-05,
      "loss": 0.8463,
      "step": 1513
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.7940862586347823,
      "learning_rate": 4.5617626295279005e-05,
      "loss": 0.9209,
      "step": 1514
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.753226599037123,
      "learning_rate": 4.559865677065893e-05,
      "loss": 0.9271,
      "step": 1515
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.6874593089610457,
      "learning_rate": 4.557967869456713e-05,
      "loss": 0.8471,
      "step": 1516
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.9855236898384365,
      "learning_rate": 4.556069207740775e-05,
      "loss": 0.9474,
      "step": 1517
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.9755529392583575,
      "learning_rate": 4.554169692958959e-05,
      "loss": 0.9251,
      "step": 1518
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.6377661984286818,
      "learning_rate": 4.552269326152612e-05,
      "loss": 0.888,
      "step": 1519
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.5595781242785014,
      "learning_rate": 4.550368108363555e-05,
      "loss": 0.9161,
      "step": 1520
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.588227879974706,
      "learning_rate": 4.548466040634067e-05,
      "loss": 0.8649,
      "step": 1521
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.5651856548467602,
      "learning_rate": 4.5465631240068984e-05,
      "loss": 0.8696,
      "step": 1522
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.6695788370844171,
      "learning_rate": 4.544659359525261e-05,
      "loss": 0.8777,
      "step": 1523
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4315097968608326,
      "learning_rate": 4.5427547482328363e-05,
      "loss": 0.7649,
      "step": 1524
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4934713353994402,
      "learning_rate": 4.540849291173767e-05,
      "loss": 0.8501,
      "step": 1525
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.647791868930416,
      "learning_rate": 4.5389429893926584e-05,
      "loss": 0.9898,
      "step": 1526
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.591096418753412,
      "learning_rate": 4.5370358439345805e-05,
      "loss": 0.8532,
      "step": 1527
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6722507192739797,
      "learning_rate": 4.535127855845067e-05,
      "loss": 0.8662,
      "step": 1528
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.9233514694952836,
      "learning_rate": 4.533219026170112e-05,
      "loss": 0.8964,
      "step": 1529
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.5549883461345677,
      "learning_rate": 4.531309355956171e-05,
      "loss": 0.8657,
      "step": 1530
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.8824212431786214,
      "learning_rate": 4.529398846250158e-05,
      "loss": 1.0229,
      "step": 1531
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6117653496802473,
      "learning_rate": 4.527487498099454e-05,
      "loss": 0.7773,
      "step": 1532
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.8584062372701253,
      "learning_rate": 4.5255753125518925e-05,
      "loss": 0.9805,
      "step": 1533
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.5152213816419582,
      "learning_rate": 4.52366229065577e-05,
      "loss": 0.8517,
      "step": 1534
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6764143616238096,
      "learning_rate": 4.521748433459839e-05,
      "loss": 0.8442,
      "step": 1535
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.8942315883125231,
      "learning_rate": 4.519833742013313e-05,
      "loss": 0.9616,
      "step": 1536
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.6193265457079418,
      "learning_rate": 4.5179182173658614e-05,
      "loss": 0.9595,
      "step": 1537
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.7339291145076061,
      "learning_rate": 4.516001860567609e-05,
      "loss": 0.9966,
      "step": 1538
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.4977231351290377,
      "learning_rate": 4.514084672669137e-05,
      "loss": 0.8692,
      "step": 1539
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5947544675080423,
      "learning_rate": 4.512166654721487e-05,
      "loss": 0.8801,
      "step": 1540
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.4774203196449927,
      "learning_rate": 4.5102478077761484e-05,
      "loss": 0.8846,
      "step": 1541
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.6240277471620401,
      "learning_rate": 4.508328132885069e-05,
      "loss": 0.8595,
      "step": 1542
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5484651157349392,
      "learning_rate": 4.506407631100651e-05,
      "loss": 0.8542,
      "step": 1543
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.333138966794168,
      "learning_rate": 4.5044863034757494e-05,
      "loss": 0.7336,
      "step": 1544
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.702326416282578,
      "learning_rate": 4.502564151063671e-05,
      "loss": 0.9754,
      "step": 1545
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.701139651502326,
      "learning_rate": 4.500641174918173e-05,
      "loss": 0.9544,
      "step": 1546
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.57153160556335,
      "learning_rate": 4.4987173760934706e-05,
      "loss": 0.837,
      "step": 1547
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.70641827850704,
      "learning_rate": 4.4967927556442215e-05,
      "loss": 0.8832,
      "step": 1548
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.4662364202794993,
      "learning_rate": 4.494867314625542e-05,
      "loss": 0.7879,
      "step": 1549
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.7017062184203287,
      "learning_rate": 4.492941054092992e-05,
      "loss": 0.9172,
      "step": 1550
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6184344165330637,
      "learning_rate": 4.4910139751025844e-05,
      "loss": 0.8216,
      "step": 1551
    },
    {
      "epoch": 1.77,
      "grad_norm": 2.0143365463348712,
      "learning_rate": 4.489086078710778e-05,
      "loss": 1.0556,
      "step": 1552
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3949673067471176,
      "learning_rate": 4.487157365974483e-05,
      "loss": 0.7847,
      "step": 1553
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5397539317266333,
      "learning_rate": 4.485227837951054e-05,
      "loss": 0.8787,
      "step": 1554
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.6926243700356431,
      "learning_rate": 4.4832974956982945e-05,
      "loss": 0.9849,
      "step": 1555
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.682617278247559,
      "learning_rate": 4.481366340274454e-05,
      "loss": 0.9734,
      "step": 1556
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5814915929750368,
      "learning_rate": 4.479434372738227e-05,
      "loss": 0.8754,
      "step": 1557
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.8124050378446064,
      "learning_rate": 4.4775015941487534e-05,
      "loss": 0.9658,
      "step": 1558
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.7511254080279446,
      "learning_rate": 4.47556800556562e-05,
      "loss": 0.9349,
      "step": 1559
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.7314765639575387,
      "learning_rate": 4.473633608048853e-05,
      "loss": 0.9105,
      "step": 1560
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5363326692770694,
      "learning_rate": 4.471698402658928e-05,
      "loss": 0.8711,
      "step": 1561
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.5598606753812005,
      "learning_rate": 4.469762390456757e-05,
      "loss": 0.7792,
      "step": 1562
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.8019374059814344,
      "learning_rate": 4.4678255725037e-05,
      "loss": 0.9122,
      "step": 1563
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.8883286259376844,
      "learning_rate": 4.465887949861556e-05,
      "loss": 0.9208,
      "step": 1564
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.9721821272658977,
      "learning_rate": 4.463949523592565e-05,
      "loss": 0.8676,
      "step": 1565
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.9328273991262495,
      "learning_rate": 4.462010294759408e-05,
      "loss": 0.9752,
      "step": 1566
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.7245165768031239,
      "learning_rate": 4.460070264425207e-05,
      "loss": 0.9169,
      "step": 1567
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.8243509183654614,
      "learning_rate": 4.458129433653522e-05,
      "loss": 0.8354,
      "step": 1568
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.7790228154941434,
      "learning_rate": 4.456187803508351e-05,
      "loss": 0.9465,
      "step": 1569
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.8258091414302446,
      "learning_rate": 4.454245375054133e-05,
      "loss": 0.976,
      "step": 1570
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4949344048290432,
      "learning_rate": 4.452302149355743e-05,
      "loss": 0.7822,
      "step": 1571
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2924695009652558,
      "learning_rate": 4.450358127478493e-05,
      "loss": 0.6996,
      "step": 1572
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.7169299106932057,
      "learning_rate": 4.448413310488131e-05,
      "loss": 0.9814,
      "step": 1573
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.463020539168944,
      "learning_rate": 4.4464676994508426e-05,
      "loss": 0.7991,
      "step": 1574
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.491987232399926,
      "learning_rate": 4.444521295433248e-05,
      "loss": 0.8181,
      "step": 1575
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.550278189406592,
      "learning_rate": 4.4425740995024e-05,
      "loss": 0.8811,
      "step": 1576
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4972068062477413,
      "learning_rate": 4.44062611272579e-05,
      "loss": 0.8491,
      "step": 1577
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.6361962676218051,
      "learning_rate": 4.438677336171337e-05,
      "loss": 0.8591,
      "step": 1578
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.708434546166381,
      "learning_rate": 4.4367277709074e-05,
      "loss": 0.9032,
      "step": 1579
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.6851056527998325,
      "learning_rate": 4.4347774180027635e-05,
      "loss": 0.883,
      "step": 1580
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.691672750325262,
      "learning_rate": 4.432826278526649e-05,
      "loss": 0.8576,
      "step": 1581
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.9205883164929938,
      "learning_rate": 4.430874353548706e-05,
      "loss": 0.9795,
      "step": 1582
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.8658512048351386,
      "learning_rate": 4.4289216441390176e-05,
      "loss": 0.9499,
      "step": 1583
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.5998111436991937,
      "learning_rate": 4.426968151368093e-05,
      "loss": 0.8021,
      "step": 1584
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.7542937699045236,
      "learning_rate": 4.425013876306874e-05,
      "loss": 0.8585,
      "step": 1585
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.6455688716049426,
      "learning_rate": 4.4230588200267297e-05,
      "loss": 0.8549,
      "step": 1586
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.6351481083019088,
      "learning_rate": 4.42110298359946e-05,
      "loss": 0.9068,
      "step": 1587
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.578777601831848,
      "learning_rate": 4.419146368097288e-05,
      "loss": 0.8396,
      "step": 1588
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4766293496626073,
      "learning_rate": 4.4171889745928675e-05,
      "loss": 0.8203,
      "step": 1589
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4211306973263806,
      "learning_rate": 4.415230804159278e-05,
      "loss": 0.7835,
      "step": 1590
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4786654629739024,
      "learning_rate": 4.413271857870025e-05,
      "loss": 0.8116,
      "step": 1591
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4430976094256125,
      "learning_rate": 4.411312136799038e-05,
      "loss": 0.7743,
      "step": 1592
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.5109587947276857,
      "learning_rate": 4.4093516420206734e-05,
      "loss": 0.8149,
      "step": 1593
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.7871554349484635,
      "learning_rate": 4.407390374609709e-05,
      "loss": 0.9205,
      "step": 1594
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.8977780229116645,
      "learning_rate": 4.40542833564135e-05,
      "loss": 0.9323,
      "step": 1595
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.64307999557667,
      "learning_rate": 4.403465526191222e-05,
      "loss": 0.876,
      "step": 1596
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.8351798797190293,
      "learning_rate": 4.4015019473353714e-05,
      "loss": 0.9343,
      "step": 1597
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.8380325017342625,
      "learning_rate": 4.399537600150271e-05,
      "loss": 0.9434,
      "step": 1598
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.7096240926722508,
      "learning_rate": 4.3975724857128095e-05,
      "loss": 0.8207,
      "step": 1599
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.794809853203688,
      "learning_rate": 4.3956066051003e-05,
      "loss": 0.8963,
      "step": 1600
    },
    {
      "epoch": 1.83,
      "eval_blimp_filtered_avg": 0.7071641791044776,
      "eval_blimp_filtered_std": 0.0050181316163269525,
      "step": 1600
    },
    {
      "epoch": 1.83,
      "eval_blimp_supplement_avg": 0.8232758620689655,
      "eval_blimp_supplement_std": 0.017096387185946922,
      "step": 1600
    },
    {
      "epoch": 1.83,
      "eval_vqa_filtered_avg": 0.55,
      "eval_vqa_filtered_std": 0.05,
      "step": 1600
    },
    {
      "epoch": 1.83,
      "eval_winoground_filtered_avg": 0.63,
      "eval_winoground_filtered_std": 0.048523658709391,
      "step": 1600
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.7910224461543487,
      "learning_rate": 4.393639959390476e-05,
      "loss": 0.8888,
      "step": 1601
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.6306007955355601,
      "learning_rate": 4.3916725496614876e-05,
      "loss": 0.8347,
      "step": 1602
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5397840614680196,
      "learning_rate": 4.3897043769919055e-05,
      "loss": 0.8555,
      "step": 1603
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.793833683059475,
      "learning_rate": 4.387735442460718e-05,
      "loss": 0.8825,
      "step": 1604
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.484472260016709,
      "learning_rate": 4.385765747147332e-05,
      "loss": 0.7674,
      "step": 1605
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.8527321628852644,
      "learning_rate": 4.3837952921315695e-05,
      "loss": 0.8402,
      "step": 1606
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.49668144093315,
      "learning_rate": 4.3818240784936734e-05,
      "loss": 0.7773,
      "step": 1607
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.77881439023542,
      "learning_rate": 4.379852107314296e-05,
      "loss": 0.8591,
      "step": 1608
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.4081600779573809,
      "learning_rate": 4.377879379674509e-05,
      "loss": 0.7149,
      "step": 1609
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3750515560490657,
      "learning_rate": 4.3759058966557994e-05,
      "loss": 0.7334,
      "step": 1610
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.7452510179769003,
      "learning_rate": 4.373931659340067e-05,
      "loss": 0.881,
      "step": 1611
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6104374961618835,
      "learning_rate": 4.3719566688096235e-05,
      "loss": 0.8322,
      "step": 1612
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.4713969339556416,
      "learning_rate": 4.369980926147197e-05,
      "loss": 0.8072,
      "step": 1613
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.8114195396888808,
      "learning_rate": 4.368004432435925e-05,
      "loss": 0.943,
      "step": 1614
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.4524963164376974,
      "learning_rate": 4.366027188759359e-05,
      "loss": 0.7677,
      "step": 1615
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.5119813209870063,
      "learning_rate": 4.3640491962014585e-05,
      "loss": 0.8012,
      "step": 1616
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.6762015210840284,
      "learning_rate": 4.362070455846598e-05,
      "loss": 0.9143,
      "step": 1617
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3630726283377508,
      "learning_rate": 4.360090968779557e-05,
      "loss": 0.7715,
      "step": 1618
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.8817299233355316,
      "learning_rate": 4.358110736085529e-05,
      "loss": 1.0111,
      "step": 1619
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.581410296291099,
      "learning_rate": 4.356129758850113e-05,
      "loss": 0.8768,
      "step": 1620
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.7527539192953836,
      "learning_rate": 4.3541480381593176e-05,
      "loss": 0.9413,
      "step": 1621
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.5943129594542564,
      "learning_rate": 4.352165575099558e-05,
      "loss": 0.8114,
      "step": 1622
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.7297340287827,
      "learning_rate": 4.3501823707576596e-05,
      "loss": 0.8839,
      "step": 1623
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5242788569425738,
      "learning_rate": 4.348198426220847e-05,
      "loss": 0.819,
      "step": 1624
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.9241736085423058,
      "learning_rate": 4.34621374257676e-05,
      "loss": 0.9662,
      "step": 1625
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5023846348769319,
      "learning_rate": 4.344228320913437e-05,
      "loss": 0.73,
      "step": 1626
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6523791534870884,
      "learning_rate": 4.342242162319322e-05,
      "loss": 0.8388,
      "step": 1627
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.560474755192733,
      "learning_rate": 4.340255267883264e-05,
      "loss": 0.6738,
      "step": 1628
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.9853039420517555,
      "learning_rate": 4.3382676386945164e-05,
      "loss": 0.9003,
      "step": 1629
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3130403666876511,
      "learning_rate": 4.336279275842734e-05,
      "loss": 0.7035,
      "step": 1630
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5751628806166837,
      "learning_rate": 4.334290180417973e-05,
      "loss": 0.8437,
      "step": 1631
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.6917080334551688,
      "learning_rate": 4.3323003535106925e-05,
      "loss": 0.8886,
      "step": 1632
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.6127608869870482,
      "learning_rate": 4.3303097962117525e-05,
      "loss": 0.8629,
      "step": 1633
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.5238849390700522,
      "learning_rate": 4.328318509612414e-05,
      "loss": 0.8316,
      "step": 1634
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.929799615501741,
      "learning_rate": 4.326326494804337e-05,
      "loss": 0.9029,
      "step": 1635
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.573608250201829,
      "learning_rate": 4.324333752879579e-05,
      "loss": 0.8142,
      "step": 1636
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.5294167379122685,
      "learning_rate": 4.3223402849306013e-05,
      "loss": 0.776,
      "step": 1637
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.746621086445882,
      "learning_rate": 4.320346092050257e-05,
      "loss": 0.8881,
      "step": 1638
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.5049553873134345,
      "learning_rate": 4.3183511753317995e-05,
      "loss": 0.7917,
      "step": 1639
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.8446777209415686,
      "learning_rate": 4.31635553586888e-05,
      "loss": 0.8342,
      "step": 1640
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.477963021141527,
      "learning_rate": 4.3143591747555446e-05,
      "loss": 0.7154,
      "step": 1641
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.6315631494288498,
      "learning_rate": 4.3123620930862364e-05,
      "loss": 0.8203,
      "step": 1642
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.649579132483942,
      "learning_rate": 4.310364291955791e-05,
      "loss": 0.7811,
      "step": 1643
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.780213523741584,
      "learning_rate": 4.3083657724594394e-05,
      "loss": 0.87,
      "step": 1644
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.6401441990735033,
      "learning_rate": 4.306366535692809e-05,
      "loss": 0.8389,
      "step": 1645
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3377468105259562,
      "learning_rate": 4.304366582751917e-05,
      "loss": 0.7316,
      "step": 1646
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.5528394712797147,
      "learning_rate": 4.302365914733174e-05,
      "loss": 0.8251,
      "step": 1647
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4899907540554982,
      "learning_rate": 4.300364532733384e-05,
      "loss": 0.9002,
      "step": 1648
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.6009522178446867,
      "learning_rate": 4.298362437849743e-05,
      "loss": 0.8466,
      "step": 1649
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.4656193222940246,
      "learning_rate": 4.296359631179833e-05,
      "loss": 0.8782,
      "step": 1650
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.697952756259932,
      "learning_rate": 4.294356113821632e-05,
      "loss": 0.8911,
      "step": 1651
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.6469057775946696,
      "learning_rate": 4.292351886873505e-05,
      "loss": 0.9697,
      "step": 1652
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.6644172785353617,
      "learning_rate": 4.290346951434206e-05,
      "loss": 0.8711,
      "step": 1653
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.662933157855988,
      "learning_rate": 4.288341308602877e-05,
      "loss": 0.9188,
      "step": 1654
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5169637725890288,
      "learning_rate": 4.286334959479049e-05,
      "loss": 0.7871,
      "step": 1655
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.9545243854212686,
      "learning_rate": 4.284327905162639e-05,
      "loss": 0.9989,
      "step": 1656
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.554367033421096,
      "learning_rate": 4.282320146753953e-05,
      "loss": 0.829,
      "step": 1657
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.6799497700745964,
      "learning_rate": 4.28031168535368e-05,
      "loss": 0.8196,
      "step": 1658
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.6400387999635666,
      "learning_rate": 4.278302522062895e-05,
      "loss": 0.8764,
      "step": 1659
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.723432614592233,
      "learning_rate": 4.27629265798306e-05,
      "loss": 0.8946,
      "step": 1660
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.8856838633565154,
      "learning_rate": 4.2742820942160177e-05,
      "loss": 0.8079,
      "step": 1661
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.6275022030622441,
      "learning_rate": 4.272270831863998e-05,
      "loss": 0.6965,
      "step": 1662
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.6478339120550947,
      "learning_rate": 4.270258872029611e-05,
      "loss": 0.7879,
      "step": 1663
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.6452735523461475,
      "learning_rate": 4.268246215815851e-05,
      "loss": 0.8277,
      "step": 1664
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.6902573846610423,
      "learning_rate": 4.266232864326093e-05,
      "loss": 0.7943,
      "step": 1665
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2023603144845407,
      "learning_rate": 4.2642188186640935e-05,
      "loss": 0.6459,
      "step": 1666
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.9380524362214928,
      "learning_rate": 4.262204079933989e-05,
      "loss": 0.8373,
      "step": 1667
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.6590197285178945,
      "learning_rate": 4.260188649240297e-05,
      "loss": 0.8295,
      "step": 1668
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.6814847194047817,
      "learning_rate": 4.258172527687914e-05,
      "loss": 0.9197,
      "step": 1669
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.6677338862923374,
      "learning_rate": 4.256155716382115e-05,
      "loss": 0.926,
      "step": 1670
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.5178626918067655,
      "learning_rate": 4.254138216428552e-05,
      "loss": 0.8121,
      "step": 1671
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.6279332977665157,
      "learning_rate": 4.252120028933258e-05,
      "loss": 0.8081,
      "step": 1672
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.6875689804072966,
      "learning_rate": 4.250101155002639e-05,
      "loss": 0.8586,
      "step": 1673
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.6406141746251122,
      "learning_rate": 4.24808159574348e-05,
      "loss": 0.8582,
      "step": 1674
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.5062524052612773,
      "learning_rate": 4.2460613522629385e-05,
      "loss": 0.8283,
      "step": 1675
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.5371278667201784,
      "learning_rate": 4.244040425668552e-05,
      "loss": 0.8434,
      "step": 1676
    },
    {
      "epoch": 1.92,
      "grad_norm": 2.1498195791458614,
      "learning_rate": 4.242018817068229e-05,
      "loss": 1.0003,
      "step": 1677
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.6901976028161299,
      "learning_rate": 4.2399965275702514e-05,
      "loss": 0.8332,
      "step": 1678
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.5018443886176653,
      "learning_rate": 4.2379735582832765e-05,
      "loss": 0.7664,
      "step": 1679
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.571509252311741,
      "learning_rate": 4.235949910316334e-05,
      "loss": 0.785,
      "step": 1680
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.6437732327890073,
      "learning_rate": 4.233925584778825e-05,
      "loss": 0.8623,
      "step": 1681
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4750167684189492,
      "learning_rate": 4.231900582780521e-05,
      "loss": 0.7699,
      "step": 1682
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.688587753245439,
      "learning_rate": 4.2298749054315665e-05,
      "loss": 0.9303,
      "step": 1683
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.6446112923678773,
      "learning_rate": 4.2278485538424744e-05,
      "loss": 0.8677,
      "step": 1684
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.6863144086965303,
      "learning_rate": 4.22582152912413e-05,
      "loss": 0.8552,
      "step": 1685
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5472495884001822,
      "learning_rate": 4.2237938323877835e-05,
      "loss": 0.8007,
      "step": 1686
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.6366560587421803,
      "learning_rate": 4.221765464745056e-05,
      "loss": 0.7711,
      "step": 1687
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.6852545603808264,
      "learning_rate": 4.2197364273079375e-05,
      "loss": 0.8844,
      "step": 1688
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5051018802249267,
      "learning_rate": 4.217706721188783e-05,
      "loss": 0.7778,
      "step": 1689
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5367180242923455,
      "learning_rate": 4.215676347500314e-05,
      "loss": 0.8045,
      "step": 1690
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.8673731430188012,
      "learning_rate": 4.21364530735562e-05,
      "loss": 0.9343,
      "step": 1691
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.6873637871462401,
      "learning_rate": 4.211613601868154e-05,
      "loss": 0.8825,
      "step": 1692
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.6331201343439263,
      "learning_rate": 4.209581232151735e-05,
      "loss": 0.8764,
      "step": 1693
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6971686582182126,
      "learning_rate": 4.207548199320545e-05,
      "loss": 0.8333,
      "step": 1694
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6311129726458755,
      "learning_rate": 4.2055145044891316e-05,
      "loss": 0.912,
      "step": 1695
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5435623908917093,
      "learning_rate": 4.203480148772403e-05,
      "loss": 0.8978,
      "step": 1696
    },
    {
      "epoch": 1.94,
      "grad_norm": 2.112477123251343,
      "learning_rate": 4.2014451332856296e-05,
      "loss": 0.9779,
      "step": 1697
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.652692481311726,
      "learning_rate": 4.199409459144446e-05,
      "loss": 0.8988,
      "step": 1698
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5785850590728057,
      "learning_rate": 4.197373127464845e-05,
      "loss": 0.8008,
      "step": 1699
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6355626223872652,
      "learning_rate": 4.195336139363183e-05,
      "loss": 0.841,
      "step": 1700
    },
    {
      "epoch": 1.94,
      "eval_blimp_filtered_avg": 0.7064179104477611,
      "eval_blimp_filtered_std": 0.005030246423315235,
      "step": 1700
    },
    {
      "epoch": 1.94,
      "eval_blimp_supplement_avg": 0.8275862068965517,
      "eval_blimp_supplement_std": 0.017064688286622314,
      "step": 1700
    },
    {
      "epoch": 1.94,
      "eval_vqa_filtered_avg": 0.55,
      "eval_vqa_filtered_std": 0.05,
      "step": 1700
    },
    {
      "epoch": 1.94,
      "eval_winoground_filtered_avg": 0.63,
      "eval_winoground_filtered_std": 0.048523658709391,
      "step": 1700
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2852798792361706,
      "learning_rate": 4.193298495956172e-05,
      "loss": 0.7144,
      "step": 1701
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.575875511897742,
      "learning_rate": 4.191260198360887e-05,
      "loss": 0.8901,
      "step": 1702
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5313641908749038,
      "learning_rate": 4.1892212476947615e-05,
      "loss": 0.8271,
      "step": 1703
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.8330980584941994,
      "learning_rate": 4.187181645075585e-05,
      "loss": 0.9364,
      "step": 1704
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.9173348112145319,
      "learning_rate": 4.185141391621505e-05,
      "loss": 0.8883,
      "step": 1705
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.898200969475029,
      "learning_rate": 4.183100488451026e-05,
      "loss": 0.8894,
      "step": 1706
    },
    {
      "epoch": 1.95,
      "grad_norm": 2.0197086622443794,
      "learning_rate": 4.1810589366830094e-05,
      "loss": 0.8419,
      "step": 1707
    },
    {
      "epoch": 1.95,
      "grad_norm": 2.2907810045976187,
      "learning_rate": 4.1790167374366724e-05,
      "loss": 0.8218,
      "step": 1708
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.7048899348860067,
      "learning_rate": 4.1769738918315837e-05,
      "loss": 0.8461,
      "step": 1709
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5113758817906493,
      "learning_rate": 4.174930400987671e-05,
      "loss": 0.7277,
      "step": 1710
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.6580652807297145,
      "learning_rate": 4.1728862660252135e-05,
      "loss": 0.7478,
      "step": 1711
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.8117585672752115,
      "learning_rate": 4.170841488064844e-05,
      "loss": 0.9123,
      "step": 1712
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5754706517552086,
      "learning_rate": 4.168796068227547e-05,
      "loss": 0.8602,
      "step": 1713
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5782041718732558,
      "learning_rate": 4.166750007634659e-05,
      "loss": 0.8014,
      "step": 1714
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5505693458508156,
      "learning_rate": 4.1647033074078696e-05,
      "loss": 0.8236,
      "step": 1715
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.7226668554668836,
      "learning_rate": 4.162655968669216e-05,
      "loss": 0.9003,
      "step": 1716
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.6146084622164485,
      "learning_rate": 4.160607992541089e-05,
      "loss": 0.9097,
      "step": 1717
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5674670157242054,
      "learning_rate": 4.158559380146224e-05,
      "loss": 0.8347,
      "step": 1718
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.6869087815367534,
      "learning_rate": 4.156510132607711e-05,
      "loss": 0.8783,
      "step": 1719
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.4916380990718798,
      "learning_rate": 4.154460251048986e-05,
      "loss": 0.7469,
      "step": 1720
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.4823379728009878,
      "learning_rate": 4.1524097365938274e-05,
      "loss": 0.7965,
      "step": 1721
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.5966796402585466,
      "learning_rate": 4.15035859036637e-05,
      "loss": 0.837,
      "step": 1722
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.6655686876986038,
      "learning_rate": 4.1483068134910876e-05,
      "loss": 0.7556,
      "step": 1723
    },
    {
      "epoch": 1.97,
      "grad_norm": 2.0962513632942508,
      "learning_rate": 4.1462544070928024e-05,
      "loss": 0.9732,
      "step": 1724
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.8900788194787885,
      "learning_rate": 4.1442013722966815e-05,
      "loss": 0.921,
      "step": 1725
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.501386550669223,
      "learning_rate": 4.1421477102282366e-05,
      "loss": 0.7597,
      "step": 1726
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7557868771308582,
      "learning_rate": 4.1400934220133236e-05,
      "loss": 0.872,
      "step": 1727
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7102104742158446,
      "learning_rate": 4.138038508778141e-05,
      "loss": 0.8849,
      "step": 1728
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.630247325442754,
      "learning_rate": 4.13598297164923e-05,
      "loss": 0.8007,
      "step": 1729
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.7692974097647751,
      "learning_rate": 4.133926811753474e-05,
      "loss": 0.8867,
      "step": 1730
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.489506221168437,
      "learning_rate": 4.131870030218097e-05,
      "loss": 0.8054,
      "step": 1731
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.6098864746892834,
      "learning_rate": 4.129812628170667e-05,
      "loss": 0.8305,
      "step": 1732
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2288975381875042,
      "learning_rate": 4.127754606739088e-05,
      "loss": 0.6837,
      "step": 1733
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.4672846547853817,
      "learning_rate": 4.1256959670516054e-05,
      "loss": 0.7909,
      "step": 1734
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.6206689616815615,
      "learning_rate": 4.123636710236804e-05,
      "loss": 0.8767,
      "step": 1735
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.6430956589221068,
      "learning_rate": 4.121576837423606e-05,
      "loss": 0.8788,
      "step": 1736
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.7832523351706069,
      "learning_rate": 4.1195163497412726e-05,
      "loss": 0.9274,
      "step": 1737
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.7654746087560718,
      "learning_rate": 4.1174552483193996e-05,
      "loss": 0.8156,
      "step": 1738
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3960641646879544,
      "learning_rate": 4.115393534287922e-05,
      "loss": 0.7507,
      "step": 1739
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.618404461306914,
      "learning_rate": 4.113331208777111e-05,
      "loss": 0.797,
      "step": 1740
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.787785989654667,
      "learning_rate": 4.111268272917567e-05,
      "loss": 0.9259,
      "step": 1741
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.735148437707804,
      "learning_rate": 4.109204727840234e-05,
      "loss": 0.9532,
      "step": 1742
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.5284594329861714,
      "learning_rate": 4.1071405746763845e-05,
      "loss": 0.6987,
      "step": 1743
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.575436205017517,
      "learning_rate": 4.1050758145576234e-05,
      "loss": 0.8041,
      "step": 1744
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.52851816236907,
      "learning_rate": 4.103010448615892e-05,
      "loss": 0.8429,
      "step": 1745
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5776385357281564,
      "learning_rate": 4.1009444779834626e-05,
      "loss": 0.7524,
      "step": 1746
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.476222138117243,
      "learning_rate": 4.098877903792937e-05,
      "loss": 0.7853,
      "step": 1747
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.8787143193935252,
      "learning_rate": 4.09681072717725e-05,
      "loss": 0.9016,
      "step": 1748
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.7658179757982908,
      "learning_rate": 4.094742949269666e-05,
      "loss": 0.8158,
      "step": 1749
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3769269708011436,
      "learning_rate": 4.092674571203779e-05,
      "loss": 0.6298,
      "step": 1750
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.27719540268342,
      "learning_rate": 4.0906055941135124e-05,
      "loss": 0.6627,
      "step": 1751
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.254250828713255,
      "learning_rate": 4.088536019133117e-05,
      "loss": 0.6489,
      "step": 1752
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4252839694726176,
      "learning_rate": 4.0864658473971726e-05,
      "loss": 0.6921,
      "step": 1753
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.353479526149188,
      "learning_rate": 4.0843950800405844e-05,
      "loss": 0.6224,
      "step": 1754
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3108579734746832,
      "learning_rate": 4.0823237181985876e-05,
      "loss": 0.5753,
      "step": 1755
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4073459620187412,
      "learning_rate": 4.080251763006738e-05,
      "loss": 0.6072,
      "step": 1756
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.563882106305504,
      "learning_rate": 4.0781792156009226e-05,
      "loss": 0.5667,
      "step": 1757
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.7045923666303613,
      "learning_rate": 4.0761060771173485e-05,
      "loss": 0.6129,
      "step": 1758
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.6204428963286786,
      "learning_rate": 4.07403234869255e-05,
      "loss": 0.675,
      "step": 1759
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.590186054426488,
      "learning_rate": 4.071958031463381e-05,
      "loss": 0.5799,
      "step": 1760
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.6061164039807845,
      "learning_rate": 4.069883126567022e-05,
      "loss": 0.6184,
      "step": 1761
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.9272533819406508,
      "learning_rate": 4.067807635140974e-05,
      "loss": 0.6595,
      "step": 1762
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.709882173087582,
      "learning_rate": 4.0657315583230596e-05,
      "loss": 0.6558,
      "step": 1763
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.8406818477624047,
      "learning_rate": 4.0636548972514225e-05,
      "loss": 0.6624,
      "step": 1764
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.681139357056113,
      "learning_rate": 4.061577653064526e-05,
      "loss": 0.7092,
      "step": 1765
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4365310368190067,
      "learning_rate": 4.059499826901155e-05,
      "loss": 0.6041,
      "step": 1766
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.5561588651403697,
      "learning_rate": 4.057421419900412e-05,
      "loss": 0.6829,
      "step": 1767
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3356484737860963,
      "learning_rate": 4.055342433201717e-05,
      "loss": 0.5891,
      "step": 1768
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.484684549413198,
      "learning_rate": 4.05326286794481e-05,
      "loss": 0.5826,
      "step": 1769
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2209337711009192,
      "learning_rate": 4.0511827252697475e-05,
      "loss": 0.5481,
      "step": 1770
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.562944802435354,
      "learning_rate": 4.0491020063169e-05,
      "loss": 0.6267,
      "step": 1771
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.5645135940504025,
      "learning_rate": 4.0470207122269575e-05,
      "loss": 0.6333,
      "step": 1772
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6490396029124927,
      "learning_rate": 4.044938844140924e-05,
      "loss": 0.5895,
      "step": 1773
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.4350852071408482,
      "learning_rate": 4.0428564032001185e-05,
      "loss": 0.5649,
      "step": 1774
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.326757287189493,
      "learning_rate": 4.040773390546172e-05,
      "loss": 0.5252,
      "step": 1775
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.7295070659076157,
      "learning_rate": 4.038689807321031e-05,
      "loss": 0.6693,
      "step": 1776
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.7091069773216934,
      "learning_rate": 4.036605654666955e-05,
      "loss": 0.6628,
      "step": 1777
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6950744378534655,
      "learning_rate": 4.034520933726514e-05,
      "loss": 0.64,
      "step": 1778
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.797780499160475,
      "learning_rate": 4.032435645642592e-05,
      "loss": 0.6866,
      "step": 1779
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.5475829065039217,
      "learning_rate": 4.030349791558381e-05,
      "loss": 0.6445,
      "step": 1780
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.6421655205763233,
      "learning_rate": 4.028263372617386e-05,
      "loss": 0.605,
      "step": 1781
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.5315082253008234,
      "learning_rate": 4.02617638996342e-05,
      "loss": 0.5773,
      "step": 1782
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3901789249874648,
      "learning_rate": 4.024088844740604e-05,
      "loss": 0.5728,
      "step": 1783
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.321237707083159,
      "learning_rate": 4.022000738093369e-05,
      "loss": 0.5675,
      "step": 1784
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.4304096214355457,
      "learning_rate": 4.019912071166457e-05,
      "loss": 0.6056,
      "step": 1785
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3579469309743086,
      "learning_rate": 4.01782284510491e-05,
      "loss": 0.5803,
      "step": 1786
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3897748406919983,
      "learning_rate": 4.015733061054083e-05,
      "loss": 0.6082,
      "step": 1787
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.4493818623804178,
      "learning_rate": 4.013642720159631e-05,
      "loss": 0.6179,
      "step": 1788
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.5311565704424595,
      "learning_rate": 4.011551823567521e-05,
      "loss": 0.6557,
      "step": 1789
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.457996682720405,
      "learning_rate": 4.009460372424019e-05,
      "loss": 0.589,
      "step": 1790
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3576443974935615,
      "learning_rate": 4.007368367875698e-05,
      "loss": 0.5767,
      "step": 1791
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.4398716488403656,
      "learning_rate": 4.005275811069432e-05,
      "loss": 0.5662,
      "step": 1792
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.5991953379388104,
      "learning_rate": 4.003182703152402e-05,
      "loss": 0.654,
      "step": 1793
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.4780898007140728,
      "learning_rate": 4.0010890452720854e-05,
      "loss": 0.5689,
      "step": 1794
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.5060432763139104,
      "learning_rate": 3.998994838576267e-05,
      "loss": 0.6294,
      "step": 1795
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.9407747689342838,
      "learning_rate": 3.9969000842130276e-05,
      "loss": 0.745,
      "step": 1796
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.5111036976465857,
      "learning_rate": 3.994804783330751e-05,
      "loss": 0.6221,
      "step": 1797
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.4986198272703706,
      "learning_rate": 3.9927089370781194e-05,
      "loss": 0.5816,
      "step": 1798
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3398386515766492,
      "learning_rate": 3.990612546604115e-05,
      "loss": 0.5469,
      "step": 1799
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4610483411443436,
      "learning_rate": 3.9885156130580166e-05,
      "loss": 0.5812,
      "step": 1800
    },
    {
      "epoch": 2.06,
      "eval_blimp_filtered_avg": 0.7091044776119403,
      "eval_blimp_filtered_std": 0.005013443565044148,
      "step": 1800
    },
    {
      "epoch": 2.06,
      "eval_blimp_supplement_avg": 0.8383620689655172,
      "eval_blimp_supplement_std": 0.016522363806281668,
      "step": 1800
    },
    {
      "epoch": 2.06,
      "eval_vqa_filtered_avg": 0.57,
      "eval_vqa_filtered_std": 0.04975698519562428,
      "step": 1800
    },
    {
      "epoch": 2.06,
      "eval_winoground_filtered_avg": 0.61,
      "eval_winoground_filtered_std": 0.04902071300001975,
      "step": 1800
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.5671137861681546,
      "learning_rate": 3.986418137589404e-05,
      "loss": 0.6354,
      "step": 1801
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.541348883516256,
      "learning_rate": 3.98432012134815e-05,
      "loss": 0.6536,
      "step": 1802
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.703972862821284,
      "learning_rate": 3.9822215654844276e-05,
      "loss": 0.6459,
      "step": 1803
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.5017024328590154,
      "learning_rate": 3.980122471148702e-05,
      "loss": 0.631,
      "step": 1804
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4950395435844843,
      "learning_rate": 3.978022839491736e-05,
      "loss": 0.5491,
      "step": 1805
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.563205663604122,
      "learning_rate": 3.975922671664588e-05,
      "loss": 0.6601,
      "step": 1806
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.6608960947933231,
      "learning_rate": 3.973821968818607e-05,
      "loss": 0.6598,
      "step": 1807
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3615395935044217,
      "learning_rate": 3.971720732105437e-05,
      "loss": 0.5414,
      "step": 1808
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3446096948387403,
      "learning_rate": 3.9696189626770165e-05,
      "loss": 0.5391,
      "step": 1809
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3449485831077412,
      "learning_rate": 3.967516661685573e-05,
      "loss": 0.589,
      "step": 1810
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.555241307059899,
      "learning_rate": 3.9654138302836275e-05,
      "loss": 0.6477,
      "step": 1811
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.5962148045421787,
      "learning_rate": 3.9633104696239896e-05,
      "loss": 0.7241,
      "step": 1812
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.814225166585558,
      "learning_rate": 3.961206580859762e-05,
      "loss": 0.6827,
      "step": 1813
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.6836246168072084,
      "learning_rate": 3.959102165144335e-05,
      "loss": 0.675,
      "step": 1814
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.4778553764477365,
      "learning_rate": 3.956997223631387e-05,
      "loss": 0.6149,
      "step": 1815
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6301715925943712,
      "learning_rate": 3.954891757474888e-05,
      "loss": 0.6703,
      "step": 1816
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6716614485122554,
      "learning_rate": 3.9527857678290906e-05,
      "loss": 0.6228,
      "step": 1817
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2760483842665267,
      "learning_rate": 3.9506792558485405e-05,
      "loss": 0.5025,
      "step": 1818
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.7062785736304542,
      "learning_rate": 3.9485722226880635e-05,
      "loss": 0.5875,
      "step": 1819
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6031333654527684,
      "learning_rate": 3.9464646695027756e-05,
      "loss": 0.607,
      "step": 1820
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.4750631843517297,
      "learning_rate": 3.944356597448077e-05,
      "loss": 0.6048,
      "step": 1821
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.509260882616605,
      "learning_rate": 3.9422480076796496e-05,
      "loss": 0.5659,
      "step": 1822
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.5389016428157782,
      "learning_rate": 3.940138901353464e-05,
      "loss": 0.594,
      "step": 1823
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.4900008323073546,
      "learning_rate": 3.938029279625768e-05,
      "loss": 0.5965,
      "step": 1824
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.8887522863257686,
      "learning_rate": 3.935919143653098e-05,
      "loss": 0.7,
      "step": 1825
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.392674451488079,
      "learning_rate": 3.933808494592269e-05,
      "loss": 0.5742,
      "step": 1826
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.5459970839972959,
      "learning_rate": 3.931697333600376e-05,
      "loss": 0.6184,
      "step": 1827
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.4569970045816887,
      "learning_rate": 3.929585661834799e-05,
      "loss": 0.6206,
      "step": 1828
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.5093704752899733,
      "learning_rate": 3.9274734804531936e-05,
      "loss": 0.6268,
      "step": 1829
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.5332085954941796,
      "learning_rate": 3.9253607906134975e-05,
      "loss": 0.6656,
      "step": 1830
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3822087116650297,
      "learning_rate": 3.923247593473926e-05,
      "loss": 0.5676,
      "step": 1831
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.7059837428085172,
      "learning_rate": 3.9211338901929714e-05,
      "loss": 0.6317,
      "step": 1832
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3206540888791303,
      "learning_rate": 3.919019681929408e-05,
      "loss": 0.5502,
      "step": 1833
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.5893301247283158,
      "learning_rate": 3.916904969842282e-05,
      "loss": 0.6385,
      "step": 1834
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.5451620890597624,
      "learning_rate": 3.914789755090916e-05,
      "loss": 0.6142,
      "step": 1835
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.361231863937087,
      "learning_rate": 3.912674038834913e-05,
      "loss": 0.5622,
      "step": 1836
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.6471308595685599,
      "learning_rate": 3.910557822234146e-05,
      "loss": 0.618,
      "step": 1837
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3792928629277628,
      "learning_rate": 3.9084411064487645e-05,
      "loss": 0.5759,
      "step": 1838
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.5264654443567038,
      "learning_rate": 3.90632389263919e-05,
      "loss": 0.5836,
      "step": 1839
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.9078360140186807,
      "learning_rate": 3.90420618196612e-05,
      "loss": 0.7196,
      "step": 1840
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.6865929406957914,
      "learning_rate": 3.902087975590521e-05,
      "loss": 0.6777,
      "step": 1841
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.6896207849150109,
      "learning_rate": 3.899969274673634e-05,
      "loss": 0.6473,
      "step": 1842
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.014541353033349,
      "learning_rate": 3.89785008037697e-05,
      "loss": 0.6986,
      "step": 1843
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.690281343459732,
      "learning_rate": 3.89573039386231e-05,
      "loss": 0.6613,
      "step": 1844
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.547367577355236,
      "learning_rate": 3.893610216291706e-05,
      "loss": 0.6009,
      "step": 1845
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.7348028521539531,
      "learning_rate": 3.891489548827477e-05,
      "loss": 0.7191,
      "step": 1846
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.5763712863618082,
      "learning_rate": 3.8893683926322126e-05,
      "loss": 0.6582,
      "step": 1847
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.5915979814095604,
      "learning_rate": 3.8872467488687704e-05,
      "loss": 0.5786,
      "step": 1848
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.684765551694086,
      "learning_rate": 3.885124618700274e-05,
      "loss": 0.5531,
      "step": 1849
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4163932132690171,
      "learning_rate": 3.883002003290115e-05,
      "loss": 0.5525,
      "step": 1850
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4695818023574603,
      "learning_rate": 3.8808789038019504e-05,
      "loss": 0.5804,
      "step": 1851
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4586461057522562,
      "learning_rate": 3.8787553213997025e-05,
      "loss": 0.5918,
      "step": 1852
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.9528012558253844,
      "learning_rate": 3.876631257247558e-05,
      "loss": 0.665,
      "step": 1853
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.5651479234701375,
      "learning_rate": 3.874506712509968e-05,
      "loss": 0.6716,
      "step": 1854
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.644457100650583,
      "learning_rate": 3.872381688351647e-05,
      "loss": 0.6346,
      "step": 1855
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3578712976360015,
      "learning_rate": 3.870256185937573e-05,
      "loss": 0.5686,
      "step": 1856
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.9708875421942278,
      "learning_rate": 3.8681302064329866e-05,
      "loss": 0.5524,
      "step": 1857
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4825295181949523,
      "learning_rate": 3.866003751003386e-05,
      "loss": 0.5959,
      "step": 1858
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.5617445895653634,
      "learning_rate": 3.8638768208145376e-05,
      "loss": 0.6245,
      "step": 1859
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.6748586459494719,
      "learning_rate": 3.861749417032461e-05,
      "loss": 0.6361,
      "step": 1860
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.6645672733044394,
      "learning_rate": 3.859621540823438e-05,
      "loss": 0.6303,
      "step": 1861
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.966072030643582,
      "learning_rate": 3.8574931933540104e-05,
      "loss": 0.712,
      "step": 1862
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.7212616892548163,
      "learning_rate": 3.855364375790979e-05,
      "loss": 0.6197,
      "step": 1863
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.7378794580574242,
      "learning_rate": 3.853235089301399e-05,
      "loss": 0.6089,
      "step": 1864
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.9100674014280186,
      "learning_rate": 3.851105335052585e-05,
      "loss": 0.7341,
      "step": 1865
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.5890328609742752,
      "learning_rate": 3.848975114212107e-05,
      "loss": 0.6631,
      "step": 1866
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4680666925087,
      "learning_rate": 3.846844427947793e-05,
      "loss": 0.5753,
      "step": 1867
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.6143506860055097,
      "learning_rate": 3.844713277427723e-05,
      "loss": 0.6105,
      "step": 1868
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3873665628683955,
      "learning_rate": 3.8425816638202345e-05,
      "loss": 0.5699,
      "step": 1869
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.427421256818506,
      "learning_rate": 3.8404495882939156e-05,
      "loss": 0.5811,
      "step": 1870
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3909388191268985,
      "learning_rate": 3.8383170520176126e-05,
      "loss": 0.5399,
      "step": 1871
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.6719495612773312,
      "learning_rate": 3.8361840561604174e-05,
      "loss": 0.7637,
      "step": 1872
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.5709387639825327,
      "learning_rate": 3.83405060189168e-05,
      "loss": 0.6278,
      "step": 1873
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.462267466678255,
      "learning_rate": 3.831916690380999e-05,
      "loss": 0.5563,
      "step": 1874
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.71863531490377,
      "learning_rate": 3.8297823227982246e-05,
      "loss": 0.6424,
      "step": 1875
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.847699763179429,
      "learning_rate": 3.827647500313456e-05,
      "loss": 0.6361,
      "step": 1876
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.521018812599912,
      "learning_rate": 3.8255122240970424e-05,
      "loss": 0.6282,
      "step": 1877
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.7623173557200904,
      "learning_rate": 3.823376495319581e-05,
      "loss": 0.602,
      "step": 1878
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.5325720906614468,
      "learning_rate": 3.821240315151919e-05,
      "loss": 0.607,
      "step": 1879
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4827604777377639,
      "learning_rate": 3.819103684765149e-05,
      "loss": 0.6151,
      "step": 1880
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.6520811125097752,
      "learning_rate": 3.8169666053306116e-05,
      "loss": 0.607,
      "step": 1881
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.680752654939389,
      "learning_rate": 3.814829078019893e-05,
      "loss": 0.5848,
      "step": 1882
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.5381897041544572,
      "learning_rate": 3.8126911040048246e-05,
      "loss": 0.6175,
      "step": 1883
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.6936744354024493,
      "learning_rate": 3.8105526844574845e-05,
      "loss": 0.6202,
      "step": 1884
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.720873419517308,
      "learning_rate": 3.8084138205501923e-05,
      "loss": 0.7061,
      "step": 1885
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.6956119921055277,
      "learning_rate": 3.806274513455514e-05,
      "loss": 0.7226,
      "step": 1886
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.6725715594291635,
      "learning_rate": 3.804134764346258e-05,
      "loss": 0.6482,
      "step": 1887
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.5667639406412426,
      "learning_rate": 3.801994574395472e-05,
      "loss": 0.5689,
      "step": 1888
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4805978584110613,
      "learning_rate": 3.799853944776448e-05,
      "loss": 0.608,
      "step": 1889
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.548299568638704,
      "learning_rate": 3.7977128766627195e-05,
      "loss": 0.6591,
      "step": 1890
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.8006469364678979,
      "learning_rate": 3.7955713712280606e-05,
      "loss": 0.7291,
      "step": 1891
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.7358924379325198,
      "learning_rate": 3.793429429646483e-05,
      "loss": 0.7093,
      "step": 1892
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.5613678112850877,
      "learning_rate": 3.791287053092238e-05,
      "loss": 0.6579,
      "step": 1893
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4040800790887704,
      "learning_rate": 3.7891442427398164e-05,
      "loss": 0.58,
      "step": 1894
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.252563986394296,
      "learning_rate": 3.787000999763948e-05,
      "loss": 0.5267,
      "step": 1895
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.612084842218785,
      "learning_rate": 3.784857325339595e-05,
      "loss": 0.6268,
      "step": 1896
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3923140638642784,
      "learning_rate": 3.782713220641962e-05,
      "loss": 0.538,
      "step": 1897
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.8100911126872532,
      "learning_rate": 3.7805686868464864e-05,
      "loss": 0.6516,
      "step": 1898
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.6201600771195623,
      "learning_rate": 3.778423725128841e-05,
      "loss": 0.5987,
      "step": 1899
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.754755607002754,
      "learning_rate": 3.776278336664933e-05,
      "loss": 0.634,
      "step": 1900
    },
    {
      "epoch": 2.17,
      "eval_blimp_filtered_avg": 0.7071641791044776,
      "eval_blimp_filtered_std": 0.0050305371780157366,
      "step": 1900
    },
    {
      "epoch": 2.17,
      "eval_blimp_supplement_avg": 0.834051724137931,
      "eval_blimp_supplement_std": 0.016742471147719547,
      "step": 1900
    },
    {
      "epoch": 2.17,
      "eval_vqa_filtered_avg": 0.54,
      "eval_vqa_filtered_std": 0.05009082659620333,
      "step": 1900
    },
    {
      "epoch": 2.17,
      "eval_winoground_filtered_avg": 0.62,
      "eval_winoground_filtered_std": 0.048783173121456316,
      "step": 1900
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.7522164323143774,
      "learning_rate": 3.774132522630903e-05,
      "loss": 0.6049,
      "step": 1901
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.67446472863105,
      "learning_rate": 3.771986284203128e-05,
      "loss": 0.6263,
      "step": 1902
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.527962641366566,
      "learning_rate": 3.769839622558215e-05,
      "loss": 0.5865,
      "step": 1903
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.568128251788967,
      "learning_rate": 3.767692538873002e-05,
      "loss": 0.5342,
      "step": 1904
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.637427112687361,
      "learning_rate": 3.7655450343245616e-05,
      "loss": 0.5693,
      "step": 1905
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.557172276701044,
      "learning_rate": 3.763397110090195e-05,
      "loss": 0.5352,
      "step": 1906
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4871770064080814,
      "learning_rate": 3.761248767347435e-05,
      "loss": 0.5765,
      "step": 1907
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5554253862351257,
      "learning_rate": 3.7591000072740404e-05,
      "loss": 0.5941,
      "step": 1908
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.0527583612259823,
      "learning_rate": 3.7569508310480024e-05,
      "loss": 0.704,
      "step": 1909
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3191665023808952,
      "learning_rate": 3.7548012398475394e-05,
      "loss": 0.5247,
      "step": 1910
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.593690514376665,
      "learning_rate": 3.752651234851097e-05,
      "loss": 0.6286,
      "step": 1911
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.5077359096324667,
      "learning_rate": 3.750500817237347e-05,
      "loss": 0.5943,
      "step": 1912
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.5574021316029742,
      "learning_rate": 3.748349988185188e-05,
      "loss": 0.6067,
      "step": 1913
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.9265025371413702,
      "learning_rate": 3.7461987488737454e-05,
      "loss": 0.7306,
      "step": 1914
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3983425553115068,
      "learning_rate": 3.744047100482367e-05,
      "loss": 0.5904,
      "step": 1915
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3722120892943799,
      "learning_rate": 3.741895044190627e-05,
      "loss": 0.5837,
      "step": 1916
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.5828288691846297,
      "learning_rate": 3.739742581178322e-05,
      "loss": 0.6205,
      "step": 1917
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.700462615686894,
      "learning_rate": 3.737589712625473e-05,
      "loss": 0.6,
      "step": 1918
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.6068070265189023,
      "learning_rate": 3.73543643971232e-05,
      "loss": 0.6445,
      "step": 1919
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4600964577044566,
      "learning_rate": 3.7332827636193284e-05,
      "loss": 0.5859,
      "step": 1920
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.653476293166639,
      "learning_rate": 3.731128685527184e-05,
      "loss": 0.6714,
      "step": 1921
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4150859645057219,
      "learning_rate": 3.728974206616791e-05,
      "loss": 0.6769,
      "step": 1922
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5785047771674217,
      "learning_rate": 3.726819328069275e-05,
      "loss": 0.641,
      "step": 1923
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.6447758347316808,
      "learning_rate": 3.72466405106598e-05,
      "loss": 0.6329,
      "step": 1924
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.7316245610423502,
      "learning_rate": 3.722508376788469e-05,
      "loss": 0.6284,
      "step": 1925
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.9084197203774342,
      "learning_rate": 3.7203523064185216e-05,
      "loss": 0.6872,
      "step": 1926
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.6907251279970354,
      "learning_rate": 3.7181958411381374e-05,
      "loss": 0.5878,
      "step": 1927
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.6597487465922427,
      "learning_rate": 3.716038982129528e-05,
      "loss": 0.6083,
      "step": 1928
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3653515536074117,
      "learning_rate": 3.713881730575125e-05,
      "loss": 0.5421,
      "step": 1929
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.6857247452998358,
      "learning_rate": 3.711724087657572e-05,
      "loss": 0.6529,
      "step": 1930
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.6014399492683096,
      "learning_rate": 3.7095660545597315e-05,
      "loss": 0.6328,
      "step": 1931
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.6298533293660642,
      "learning_rate": 3.707407632464674e-05,
      "loss": 0.6061,
      "step": 1932
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5495652177121795,
      "learning_rate": 3.7052488225556876e-05,
      "loss": 0.6935,
      "step": 1933
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.630491142772625,
      "learning_rate": 3.703089626016273e-05,
      "loss": 0.6652,
      "step": 1934
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5081911761952356,
      "learning_rate": 3.700930044030139e-05,
      "loss": 0.5869,
      "step": 1935
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5668757566839353,
      "learning_rate": 3.69877007778121e-05,
      "loss": 0.6259,
      "step": 1936
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.6247829703082857,
      "learning_rate": 3.69660972845362e-05,
      "loss": 0.5821,
      "step": 1937
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.6693017142423776,
      "learning_rate": 3.694448997231711e-05,
      "loss": 0.5932,
      "step": 1938
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.7947464541444516,
      "learning_rate": 3.6922878853000374e-05,
      "loss": 0.7134,
      "step": 1939
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.6405163239601823,
      "learning_rate": 3.690126393843357e-05,
      "loss": 0.6436,
      "step": 1940
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.6791250911193798,
      "learning_rate": 3.687964524046645e-05,
      "loss": 0.6286,
      "step": 1941
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4236449177499695,
      "learning_rate": 3.685802277095073e-05,
      "loss": 0.5285,
      "step": 1942
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.697992711962431,
      "learning_rate": 3.683639654174028e-05,
      "loss": 0.6727,
      "step": 1943
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.623387275322384,
      "learning_rate": 3.681476656469098e-05,
      "loss": 0.6202,
      "step": 1944
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.6463424785585201,
      "learning_rate": 3.6793132851660796e-05,
      "loss": 0.5919,
      "step": 1945
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.7216005282020308,
      "learning_rate": 3.677149541450973e-05,
      "loss": 0.6501,
      "step": 1946
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.6208151511735687,
      "learning_rate": 3.674985426509982e-05,
      "loss": 0.6459,
      "step": 1947
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4509398927469235,
      "learning_rate": 3.672820941529513e-05,
      "loss": 0.5905,
      "step": 1948
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.6001530154163461,
      "learning_rate": 3.670656087696179e-05,
      "loss": 0.6024,
      "step": 1949
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.5734324904889583,
      "learning_rate": 3.668490866196791e-05,
      "loss": 0.6115,
      "step": 1950
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4809202698514716,
      "learning_rate": 3.6663252782183645e-05,
      "loss": 0.6108,
      "step": 1951
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.5151920551407485,
      "learning_rate": 3.6641593249481146e-05,
      "loss": 0.6177,
      "step": 1952
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.700666463740946,
      "learning_rate": 3.6619930075734565e-05,
      "loss": 0.6657,
      "step": 1953
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.5037253460235382,
      "learning_rate": 3.6598263272820066e-05,
      "loss": 0.6507,
      "step": 1954
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2545725233412692,
      "learning_rate": 3.657659285261577e-05,
      "loss": 0.5088,
      "step": 1955
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.883689744311875,
      "learning_rate": 3.655491882700182e-05,
      "loss": 0.6434,
      "step": 1956
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.8681131630967227,
      "learning_rate": 3.6533241207860296e-05,
      "loss": 0.6535,
      "step": 1957
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3666802355330794,
      "learning_rate": 3.65115600070753e-05,
      "loss": 0.5581,
      "step": 1958
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.5480308285072937,
      "learning_rate": 3.648987523653284e-05,
      "loss": 0.6598,
      "step": 1959
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.8154272115265837,
      "learning_rate": 3.646818690812092e-05,
      "loss": 0.6409,
      "step": 1960
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.6648747001405297,
      "learning_rate": 3.644649503372948e-05,
      "loss": 0.6333,
      "step": 1961
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.6912450175204907,
      "learning_rate": 3.64247996252504e-05,
      "loss": 0.6893,
      "step": 1962
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.8256286852806207,
      "learning_rate": 3.640310069457752e-05,
      "loss": 0.5902,
      "step": 1963
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.5888500035848976,
      "learning_rate": 3.638139825360657e-05,
      "loss": 0.5785,
      "step": 1964
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.754336724016476,
      "learning_rate": 3.635969231423526e-05,
      "loss": 0.6444,
      "step": 1965
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.7071349897777632,
      "learning_rate": 3.633798288836316e-05,
      "loss": 0.6218,
      "step": 1966
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5591667402444933,
      "learning_rate": 3.631626998789179e-05,
      "loss": 0.6059,
      "step": 1967
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.740670377806425,
      "learning_rate": 3.629455362472456e-05,
      "loss": 0.632,
      "step": 1968
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.6734499066923818,
      "learning_rate": 3.6272833810766784e-05,
      "loss": 0.6535,
      "step": 1969
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5465788629169497,
      "learning_rate": 3.625111055792566e-05,
      "loss": 0.6327,
      "step": 1970
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5591013985245126,
      "learning_rate": 3.622938387811028e-05,
      "loss": 0.6173,
      "step": 1971
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.681912852378223,
      "learning_rate": 3.620765378323161e-05,
      "loss": 0.6501,
      "step": 1972
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5156394801053266,
      "learning_rate": 3.6185920285202495e-05,
      "loss": 0.5982,
      "step": 1973
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.48515154128942,
      "learning_rate": 3.616418339593763e-05,
      "loss": 0.6899,
      "step": 1974
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4236883143935226,
      "learning_rate": 3.614244312735359e-05,
      "loss": 0.6019,
      "step": 1975
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3913927105513824,
      "learning_rate": 3.612069949136878e-05,
      "loss": 0.5825,
      "step": 1976
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.335115620713137,
      "learning_rate": 3.609895249990348e-05,
      "loss": 0.526,
      "step": 1977
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4575926675068571,
      "learning_rate": 3.6077202164879774e-05,
      "loss": 0.5678,
      "step": 1978
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.9176312126151371,
      "learning_rate": 3.605544849822159e-05,
      "loss": 0.7242,
      "step": 1979
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.540835993658818,
      "learning_rate": 3.603369151185473e-05,
      "loss": 0.6116,
      "step": 1980
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.5262758604478146,
      "learning_rate": 3.601193121770674e-05,
      "loss": 0.5581,
      "step": 1981
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.542769798020007,
      "learning_rate": 3.599016762770703e-05,
      "loss": 0.5597,
      "step": 1982
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2389052424986384,
      "learning_rate": 3.5968400753786776e-05,
      "loss": 0.4763,
      "step": 1983
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.544613636605701,
      "learning_rate": 3.594663060787903e-05,
      "loss": 0.5631,
      "step": 1984
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.7771831027293836,
      "learning_rate": 3.592485720191854e-05,
      "loss": 0.6475,
      "step": 1985
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.8369504403004395,
      "learning_rate": 3.590308054784193e-05,
      "loss": 0.6888,
      "step": 1986
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.9129642079046802,
      "learning_rate": 3.588130065758752e-05,
      "loss": 0.6238,
      "step": 1987
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.6899432938338024,
      "learning_rate": 3.585951754309548e-05,
      "loss": 0.5833,
      "step": 1988
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.572074828313111,
      "learning_rate": 3.583773121630771e-05,
      "loss": 0.5954,
      "step": 1989
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5082753649907616,
      "learning_rate": 3.581594168916787e-05,
      "loss": 0.6296,
      "step": 1990
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4725359875355337,
      "learning_rate": 3.579414897362138e-05,
      "loss": 0.5899,
      "step": 1991
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.7011239259834303,
      "learning_rate": 3.577235308161542e-05,
      "loss": 0.6737,
      "step": 1992
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5322006903345393,
      "learning_rate": 3.57505540250989e-05,
      "loss": 0.5609,
      "step": 1993
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4698148591698155,
      "learning_rate": 3.572875181602245e-05,
      "loss": 0.5772,
      "step": 1994
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5690271343194773,
      "learning_rate": 3.5706946466338456e-05,
      "loss": 0.5181,
      "step": 1995
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.624650287324326,
      "learning_rate": 3.568513798800101e-05,
      "loss": 0.5909,
      "step": 1996
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.6563338154299914,
      "learning_rate": 3.5663326392965925e-05,
      "loss": 0.6132,
      "step": 1997
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.7417074631242964,
      "learning_rate": 3.564151169319072e-05,
      "loss": 0.6446,
      "step": 1998
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.8311963502077062,
      "learning_rate": 3.561969390063461e-05,
      "loss": 0.6485,
      "step": 1999
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.8090993773342088,
      "learning_rate": 3.5597873027258516e-05,
      "loss": 0.6295,
      "step": 2000
    },
    {
      "epoch": 2.29,
      "eval_blimp_filtered_avg": 0.7071641791044776,
      "eval_blimp_filtered_std": 0.005040100350838152,
      "step": 2000
    },
    {
      "epoch": 2.29,
      "eval_blimp_supplement_avg": 0.8426724137931034,
      "eval_blimp_supplement_std": 0.016359577926956526,
      "step": 2000
    },
    {
      "epoch": 2.29,
      "eval_vqa_filtered_avg": 0.54,
      "eval_vqa_filtered_std": 0.05009082659620333,
      "step": 2000
    },
    {
      "epoch": 2.29,
      "eval_winoground_filtered_avg": 0.62,
      "eval_winoground_filtered_std": 0.048783173121456316,
      "step": 2000
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.906909015370143,
      "learning_rate": 3.557604908502505e-05,
      "loss": 0.6458,
      "step": 2001
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.8479658275027262,
      "learning_rate": 3.555422208589849e-05,
      "loss": 0.6658,
      "step": 2002
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.9642826378176788,
      "learning_rate": 3.553239204184481e-05,
      "loss": 0.7154,
      "step": 2003
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5526785434597115,
      "learning_rate": 3.5510558964831625e-05,
      "loss": 0.5928,
      "step": 2004
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.1618187830864906,
      "learning_rate": 3.548872286682827e-05,
      "loss": 0.6688,
      "step": 2005
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5482303978559542,
      "learning_rate": 3.546688375980565e-05,
      "loss": 0.6389,
      "step": 2006
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5443139069988245,
      "learning_rate": 3.5445041655736386e-05,
      "loss": 0.605,
      "step": 2007
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4442394204121018,
      "learning_rate": 3.542319656659473e-05,
      "loss": 0.5591,
      "step": 2008
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4605369808575883,
      "learning_rate": 3.540134850435657e-05,
      "loss": 0.5846,
      "step": 2009
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3735093001506637,
      "learning_rate": 3.537949748099938e-05,
      "loss": 0.5497,
      "step": 2010
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.7749206974071505,
      "learning_rate": 3.535764350850235e-05,
      "loss": 0.6421,
      "step": 2011
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.5757227000712357,
      "learning_rate": 3.533578659884618e-05,
      "loss": 0.6272,
      "step": 2012
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.7857163592669485,
      "learning_rate": 3.531392676401327e-05,
      "loss": 0.7044,
      "step": 2013
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.682167542646007,
      "learning_rate": 3.529206401598754e-05,
      "loss": 0.6168,
      "step": 2014
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.383443231873138,
      "learning_rate": 3.52701983667546e-05,
      "loss": 0.5152,
      "step": 2015
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.824318443907499,
      "learning_rate": 3.5248329828301585e-05,
      "loss": 0.6538,
      "step": 2016
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.633200285918954,
      "learning_rate": 3.5226458412617224e-05,
      "loss": 0.6354,
      "step": 2017
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.61916202204473,
      "learning_rate": 3.520458413169184e-05,
      "loss": 0.6428,
      "step": 2018
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.582441847508157,
      "learning_rate": 3.518270699751731e-05,
      "loss": 0.5245,
      "step": 2019
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.6827183893892,
      "learning_rate": 3.516082702208709e-05,
      "loss": 0.6455,
      "step": 2020
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.714834092518746,
      "learning_rate": 3.513894421739619e-05,
      "loss": 0.6658,
      "step": 2021
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.8164126178371935,
      "learning_rate": 3.511705859544115e-05,
      "loss": 0.7103,
      "step": 2022
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.576198025023223,
      "learning_rate": 3.5095170168220086e-05,
      "loss": 0.5675,
      "step": 2023
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3627163424753084,
      "learning_rate": 3.5073278947732644e-05,
      "loss": 0.5113,
      "step": 2024
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.6279606945271168,
      "learning_rate": 3.5051384945979986e-05,
      "loss": 0.5969,
      "step": 2025
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5890617018741637,
      "learning_rate": 3.50294881749648e-05,
      "loss": 0.6307,
      "step": 2026
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.6324889053459886,
      "learning_rate": 3.500758864669131e-05,
      "loss": 0.6453,
      "step": 2027
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.6580544090075982,
      "learning_rate": 3.498568637316526e-05,
      "loss": 0.6115,
      "step": 2028
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5182701312097575,
      "learning_rate": 3.496378136639384e-05,
      "loss": 0.5999,
      "step": 2029
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.7841516204924803,
      "learning_rate": 3.494187363838581e-05,
      "loss": 0.6434,
      "step": 2030
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.7726866225974722,
      "learning_rate": 3.491996320115138e-05,
      "loss": 0.6492,
      "step": 2031
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5094655507020232,
      "learning_rate": 3.4898050066702266e-05,
      "loss": 0.6794,
      "step": 2032
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.664093008403517,
      "learning_rate": 3.487613424705164e-05,
      "loss": 0.5964,
      "step": 2033
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4395654733674308,
      "learning_rate": 3.4854215754214166e-05,
      "loss": 0.5503,
      "step": 2034
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5363681122692825,
      "learning_rate": 3.483229460020597e-05,
      "loss": 0.6075,
      "step": 2035
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4771534378893452,
      "learning_rate": 3.4810370797044626e-05,
      "loss": 0.612,
      "step": 2036
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.7700539204649994,
      "learning_rate": 3.478844435674917e-05,
      "loss": 0.6009,
      "step": 2037
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4869757946451019,
      "learning_rate": 3.476651529134008e-05,
      "loss": 0.5675,
      "step": 2038
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.8061455054395035,
      "learning_rate": 3.474458361283929e-05,
      "loss": 0.6317,
      "step": 2039
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.943177465975602,
      "learning_rate": 3.472264933327013e-05,
      "loss": 0.7423,
      "step": 2040
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4371130788826885,
      "learning_rate": 3.470071246465738e-05,
      "loss": 0.5695,
      "step": 2041
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.355208996528763,
      "learning_rate": 3.467877301902725e-05,
      "loss": 0.5323,
      "step": 2042
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.6461879058240587,
      "learning_rate": 3.465683100840733e-05,
      "loss": 0.6066,
      "step": 2043
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.6099986155840897,
      "learning_rate": 3.463488644482666e-05,
      "loss": 0.6212,
      "step": 2044
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.6605255250576227,
      "learning_rate": 3.4612939340315634e-05,
      "loss": 0.5661,
      "step": 2045
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5635286420107481,
      "learning_rate": 3.459098970690607e-05,
      "loss": 0.5772,
      "step": 2046
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.553649025100253,
      "learning_rate": 3.4569037556631166e-05,
      "loss": 0.5979,
      "step": 2047
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5523613961647764,
      "learning_rate": 3.454708290152548e-05,
      "loss": 0.5555,
      "step": 2048
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.7421835983061589,
      "learning_rate": 3.452512575362498e-05,
      "loss": 0.5862,
      "step": 2049
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4682355141545789,
      "learning_rate": 3.450316612496696e-05,
      "loss": 0.5251,
      "step": 2050
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.515234024419513,
      "learning_rate": 3.448120402759011e-05,
      "loss": 0.6272,
      "step": 2051
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5956823494838306,
      "learning_rate": 3.445923947353445e-05,
      "loss": 0.6133,
      "step": 2052
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4397936927095083,
      "learning_rate": 3.443727247484135e-05,
      "loss": 0.5787,
      "step": 2053
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5857127186120368,
      "learning_rate": 3.441530304355354e-05,
      "loss": 0.5848,
      "step": 2054
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.6164718338403576,
      "learning_rate": 3.439333119171505e-05,
      "loss": 0.5818,
      "step": 2055
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4423376048113998,
      "learning_rate": 3.437135693137127e-05,
      "loss": 0.5377,
      "step": 2056
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5823909082854675,
      "learning_rate": 3.434938027456887e-05,
      "loss": 0.5457,
      "step": 2057
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.6673003965210176,
      "learning_rate": 3.4327401233355905e-05,
      "loss": 0.6502,
      "step": 2058
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.7158999264473453,
      "learning_rate": 3.4305419819781645e-05,
      "loss": 0.6615,
      "step": 2059
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.35970108312716,
      "learning_rate": 3.4283436045896734e-05,
      "loss": 0.5542,
      "step": 2060
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.72499264294075,
      "learning_rate": 3.4261449923753056e-05,
      "loss": 0.6248,
      "step": 2061
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5693611687816325,
      "learning_rate": 3.423946146540385e-05,
      "loss": 0.54,
      "step": 2062
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.7029467737781632,
      "learning_rate": 3.421747068290356e-05,
      "loss": 0.6357,
      "step": 2063
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.571486484614695,
      "learning_rate": 3.4195477588307965e-05,
      "loss": 0.6085,
      "step": 2064
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3883667788844416,
      "learning_rate": 3.417348219367405e-05,
      "loss": 0.5309,
      "step": 2065
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.853751273730986,
      "learning_rate": 3.415148451106014e-05,
      "loss": 0.6366,
      "step": 2066
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5367309027046938,
      "learning_rate": 3.412948455252573e-05,
      "loss": 0.6214,
      "step": 2067
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5285284289154997,
      "learning_rate": 3.4107482330131636e-05,
      "loss": 0.5782,
      "step": 2068
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.8623313259520622,
      "learning_rate": 3.4085477855939866e-05,
      "loss": 0.6445,
      "step": 2069
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5158706193276017,
      "learning_rate": 3.4063471142013684e-05,
      "loss": 0.5499,
      "step": 2070
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.496088860634882,
      "learning_rate": 3.404146220041758e-05,
      "loss": 0.5549,
      "step": 2071
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.8073110876126843,
      "learning_rate": 3.4019451043217255e-05,
      "loss": 0.7,
      "step": 2072
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5674731937597197,
      "learning_rate": 3.3997437682479635e-05,
      "loss": 0.5639,
      "step": 2073
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.6059257556705693,
      "learning_rate": 3.397542213027285e-05,
      "loss": 0.5952,
      "step": 2074
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.750904610071589,
      "learning_rate": 3.395340439866623e-05,
      "loss": 0.6817,
      "step": 2075
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5535281751295507,
      "learning_rate": 3.393138449973031e-05,
      "loss": 0.5767,
      "step": 2076
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5685290727303702,
      "learning_rate": 3.39093624455368e-05,
      "loss": 0.5975,
      "step": 2077
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.971831989007339,
      "learning_rate": 3.38873382481586e-05,
      "loss": 0.728,
      "step": 2078
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.7883265909581119,
      "learning_rate": 3.3865311919669776e-05,
      "loss": 0.6758,
      "step": 2079
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.7184695839763098,
      "learning_rate": 3.384328347214557e-05,
      "loss": 0.674,
      "step": 2080
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.7241081379237548,
      "learning_rate": 3.3821252917662386e-05,
      "loss": 0.597,
      "step": 2081
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.7987431880168845,
      "learning_rate": 3.3799220268297775e-05,
      "loss": 0.6756,
      "step": 2082
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5606380832172093,
      "learning_rate": 3.3777185536130464e-05,
      "loss": 0.5656,
      "step": 2083
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4663392407796516,
      "learning_rate": 3.375514873324027e-05,
      "loss": 0.5904,
      "step": 2084
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3139727148289904,
      "learning_rate": 3.373310987170818e-05,
      "loss": 0.4875,
      "step": 2085
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.65093566969236,
      "learning_rate": 3.371106896361633e-05,
      "loss": 0.6314,
      "step": 2086
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.7658766232208067,
      "learning_rate": 3.368902602104793e-05,
      "loss": 0.5856,
      "step": 2087
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.7960706862637021,
      "learning_rate": 3.366698105608734e-05,
      "loss": 0.6877,
      "step": 2088
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4901057703220846,
      "learning_rate": 3.364493408082001e-05,
      "loss": 0.6095,
      "step": 2089
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5942280724208684,
      "learning_rate": 3.36228851073325e-05,
      "loss": 0.6087,
      "step": 2090
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.000728492341386,
      "learning_rate": 3.3600834147712454e-05,
      "loss": 0.675,
      "step": 2091
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.6220261477013789,
      "learning_rate": 3.357878121404864e-05,
      "loss": 0.6166,
      "step": 2092
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.7145265950355806,
      "learning_rate": 3.3556726318430864e-05,
      "loss": 0.6134,
      "step": 2093
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5930472850425905,
      "learning_rate": 3.353466947295003e-05,
      "loss": 0.5733,
      "step": 2094
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5062186919302791,
      "learning_rate": 3.351261068969812e-05,
      "loss": 0.568,
      "step": 2095
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.7683954028374957,
      "learning_rate": 3.349054998076815e-05,
      "loss": 0.6818,
      "step": 2096
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6070220446925243,
      "learning_rate": 3.3468487358254216e-05,
      "loss": 0.6065,
      "step": 2097
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.7678833104499005,
      "learning_rate": 3.344642283425146e-05,
      "loss": 0.5841,
      "step": 2098
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.7977689627834363,
      "learning_rate": 3.3424356420856056e-05,
      "loss": 0.6817,
      "step": 2099
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.7853193561948264,
      "learning_rate": 3.340228813016521e-05,
      "loss": 0.6253,
      "step": 2100
    },
    {
      "epoch": 2.4,
      "eval_blimp_filtered_avg": 0.7061194029850746,
      "eval_blimp_filtered_std": 0.005055679814065885,
      "step": 2100
    },
    {
      "epoch": 2.4,
      "eval_blimp_supplement_avg": 0.8383620689655172,
      "eval_blimp_supplement_std": 0.01652369585751099,
      "step": 2100
    },
    {
      "epoch": 2.4,
      "eval_vqa_filtered_avg": 0.55,
      "eval_vqa_filtered_std": 0.04999999999999999,
      "step": 2100
    },
    {
      "epoch": 2.4,
      "eval_winoground_filtered_avg": 0.61,
      "eval_winoground_filtered_std": 0.04902071300001975,
      "step": 2100
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.477959327303386,
      "learning_rate": 3.338021797427718e-05,
      "loss": 0.5589,
      "step": 2101
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.797738764372941,
      "learning_rate": 3.335814596529125e-05,
      "loss": 0.5919,
      "step": 2102
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.7235875292214002,
      "learning_rate": 3.333607211530765e-05,
      "loss": 0.6143,
      "step": 2103
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.675976846598579,
      "learning_rate": 3.331399643642771e-05,
      "loss": 0.571,
      "step": 2104
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.535536575939603,
      "learning_rate": 3.3291918940753736e-05,
      "loss": 0.5728,
      "step": 2105
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.522446153202001,
      "learning_rate": 3.326983964038899e-05,
      "loss": 0.5708,
      "step": 2106
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.7761159520893113,
      "learning_rate": 3.3247758547437775e-05,
      "loss": 0.6375,
      "step": 2107
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6553019462617637,
      "learning_rate": 3.322567567400533e-05,
      "loss": 0.613,
      "step": 2108
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.8882323339364047,
      "learning_rate": 3.3203591032197915e-05,
      "loss": 0.6479,
      "step": 2109
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6258348658934234,
      "learning_rate": 3.3181504634122724e-05,
      "loss": 0.586,
      "step": 2110
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5979068228368023,
      "learning_rate": 3.315941649188795e-05,
      "loss": 0.6133,
      "step": 2111
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5319126372090632,
      "learning_rate": 3.3137326617602686e-05,
      "loss": 0.5849,
      "step": 2112
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.8625475413035768,
      "learning_rate": 3.3115235023377046e-05,
      "loss": 0.6771,
      "step": 2113
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.8529845245886993,
      "learning_rate": 3.3093141721322025e-05,
      "loss": 0.6443,
      "step": 2114
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.6589968275587204,
      "learning_rate": 3.3071046723549585e-05,
      "loss": 0.6167,
      "step": 2115
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.6226859776516775,
      "learning_rate": 3.304895004217261e-05,
      "loss": 0.5808,
      "step": 2116
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.364791284884427,
      "learning_rate": 3.302685168930492e-05,
      "loss": 0.5326,
      "step": 2117
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.7659284758569143,
      "learning_rate": 3.300475167706123e-05,
      "loss": 0.6271,
      "step": 2118
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5070658163407333,
      "learning_rate": 3.298265001755718e-05,
      "loss": 0.5955,
      "step": 2119
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5535693458418718,
      "learning_rate": 3.29605467229093e-05,
      "loss": 0.5616,
      "step": 2120
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.579616633964049,
      "learning_rate": 3.293844180523503e-05,
      "loss": 0.6075,
      "step": 2121
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.547070789382176,
      "learning_rate": 3.291633527665271e-05,
      "loss": 0.6322,
      "step": 2122
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5961731926803764,
      "learning_rate": 3.2894227149281515e-05,
      "loss": 0.6041,
      "step": 2123
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.569380677598229,
      "learning_rate": 3.287211743524155e-05,
      "loss": 0.6147,
      "step": 2124
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3283231859908622,
      "learning_rate": 3.2850006146653755e-05,
      "loss": 0.4932,
      "step": 2125
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5341808509666306,
      "learning_rate": 3.282789329563997e-05,
      "loss": 0.5806,
      "step": 2126
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.7295659595108583,
      "learning_rate": 3.280577889432285e-05,
      "loss": 0.6184,
      "step": 2127
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6878537926178614,
      "learning_rate": 3.278366295482591e-05,
      "loss": 0.6115,
      "step": 2128
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.71308219197936,
      "learning_rate": 3.2761545489273536e-05,
      "loss": 0.6069,
      "step": 2129
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.7466849542248457,
      "learning_rate": 3.2739426509790924e-05,
      "loss": 0.6392,
      "step": 2130
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6897556878275952,
      "learning_rate": 3.27173060285041e-05,
      "loss": 0.5793,
      "step": 2131
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4688912400791365,
      "learning_rate": 3.269518405753993e-05,
      "loss": 0.5728,
      "step": 2132
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7918002719308384,
      "learning_rate": 3.267306060902609e-05,
      "loss": 0.6158,
      "step": 2133
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7768877953421118,
      "learning_rate": 3.265093569509106e-05,
      "loss": 0.6366,
      "step": 2134
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.882381927443951,
      "learning_rate": 3.2628809327864105e-05,
      "loss": 0.6883,
      "step": 2135
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4509769765342038,
      "learning_rate": 3.260668151947534e-05,
      "loss": 0.5711,
      "step": 2136
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.452132481570875,
      "learning_rate": 3.2584552282055624e-05,
      "loss": 0.5884,
      "step": 2137
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.56302559531155,
      "learning_rate": 3.2562421627736615e-05,
      "loss": 0.6362,
      "step": 2138
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.8341935387359343,
      "learning_rate": 3.254028956865073e-05,
      "loss": 0.6284,
      "step": 2139
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.596598033810024,
      "learning_rate": 3.251815611693121e-05,
      "loss": 0.5983,
      "step": 2140
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.6085717994264996,
      "learning_rate": 3.249602128471199e-05,
      "loss": 0.6534,
      "step": 2141
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5241758091554216,
      "learning_rate": 3.24738850841278e-05,
      "loss": 0.5816,
      "step": 2142
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5094425215561855,
      "learning_rate": 3.245174752731413e-05,
      "loss": 0.5453,
      "step": 2143
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.7701542469182436,
      "learning_rate": 3.242960862640717e-05,
      "loss": 0.5817,
      "step": 2144
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.6645809112224577,
      "learning_rate": 3.2407468393543905e-05,
      "loss": 0.5929,
      "step": 2145
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.7361647710536416,
      "learning_rate": 3.2385326840862e-05,
      "loss": 0.6074,
      "step": 2146
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.7520796479231404,
      "learning_rate": 3.236318398049987e-05,
      "loss": 0.6281,
      "step": 2147
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5542002067165006,
      "learning_rate": 3.234103982459663e-05,
      "loss": 0.6196,
      "step": 2148
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.736807973250598,
      "learning_rate": 3.231889438529213e-05,
      "loss": 0.6162,
      "step": 2149
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.234381371141111,
      "learning_rate": 3.22967476747269e-05,
      "loss": 0.6813,
      "step": 2150
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.762179094985595,
      "learning_rate": 3.2274599705042166e-05,
      "loss": 0.6708,
      "step": 2151
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.364454814987001,
      "learning_rate": 3.225245048837988e-05,
      "loss": 0.5902,
      "step": 2152
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4802869716896028,
      "learning_rate": 3.2230300036882636e-05,
      "loss": 0.5898,
      "step": 2153
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4331092910823215,
      "learning_rate": 3.220814836269371e-05,
      "loss": 0.5347,
      "step": 2154
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.7705715492119347,
      "learning_rate": 3.218599547795708e-05,
      "loss": 0.5829,
      "step": 2155
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.406565733690675,
      "learning_rate": 3.2163841394817355e-05,
      "loss": 0.4911,
      "step": 2156
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7185716877536763,
      "learning_rate": 3.214168612541981e-05,
      "loss": 0.6564,
      "step": 2157
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.6037394024190172,
      "learning_rate": 3.211952968191038e-05,
      "loss": 0.642,
      "step": 2158
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.8340748184047697,
      "learning_rate": 3.2097372076435635e-05,
      "loss": 0.6101,
      "step": 2159
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7338208384296765,
      "learning_rate": 3.207521332114279e-05,
      "loss": 0.6226,
      "step": 2160
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5485491912611549,
      "learning_rate": 3.2053053428179666e-05,
      "loss": 0.5787,
      "step": 2161
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.9657890506553024,
      "learning_rate": 3.2030892409694736e-05,
      "loss": 0.6427,
      "step": 2162
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2880625676919653,
      "learning_rate": 3.2008730277837086e-05,
      "loss": 0.4954,
      "step": 2163
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5842979035241112,
      "learning_rate": 3.19865670447564e-05,
      "loss": 0.5839,
      "step": 2164
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5669068167698976,
      "learning_rate": 3.196440272260297e-05,
      "loss": 0.6153,
      "step": 2165
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7640608610556148,
      "learning_rate": 3.1942237323527685e-05,
      "loss": 0.6109,
      "step": 2166
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6872140289371926,
      "learning_rate": 3.1920070859682027e-05,
      "loss": 0.6763,
      "step": 2167
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4590133132398424,
      "learning_rate": 3.1897903343218075e-05,
      "loss": 0.5551,
      "step": 2168
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4947824446746636,
      "learning_rate": 3.1875734786288444e-05,
      "loss": 0.5878,
      "step": 2169
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3986189189383356,
      "learning_rate": 3.185356520104637e-05,
      "loss": 0.5257,
      "step": 2170
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3789807815846855,
      "learning_rate": 3.183139459964561e-05,
      "loss": 0.5606,
      "step": 2171
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.607218977435853,
      "learning_rate": 3.180922299424052e-05,
      "loss": 0.5903,
      "step": 2172
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4980038676555534,
      "learning_rate": 3.178705039698597e-05,
      "loss": 0.5933,
      "step": 2173
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.710263501311292,
      "learning_rate": 3.1764876820037376e-05,
      "loss": 0.6129,
      "step": 2174
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.374176991569631,
      "learning_rate": 3.174270227555073e-05,
      "loss": 0.4934,
      "step": 2175
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.9482538021853866,
      "learning_rate": 3.17205267756825e-05,
      "loss": 0.6121,
      "step": 2176
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.583045247839717,
      "learning_rate": 3.169835033258973e-05,
      "loss": 0.5307,
      "step": 2177
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.7818310358902465,
      "learning_rate": 3.167617295842992e-05,
      "loss": 0.6522,
      "step": 2178
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6394973367250358,
      "learning_rate": 3.1653994665361154e-05,
      "loss": 0.6522,
      "step": 2179
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.8097185904919115,
      "learning_rate": 3.163181546554196e-05,
      "loss": 0.5924,
      "step": 2180
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5524910102988887,
      "learning_rate": 3.1609635371131395e-05,
      "loss": 0.5489,
      "step": 2181
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.8491664033601507,
      "learning_rate": 3.158745439428899e-05,
      "loss": 0.5817,
      "step": 2182
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6362995669642655,
      "learning_rate": 3.156527254717477e-05,
      "loss": 0.643,
      "step": 2183
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6668565626382479,
      "learning_rate": 3.154308984194922e-05,
      "loss": 0.5855,
      "step": 2184
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4831942318806546,
      "learning_rate": 3.1520906290773334e-05,
      "loss": 0.5681,
      "step": 2185
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.549675743180217,
      "learning_rate": 3.1498721905808524e-05,
      "loss": 0.5708,
      "step": 2186
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5704015783188319,
      "learning_rate": 3.1476536699216693e-05,
      "loss": 0.5853,
      "step": 2187
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.694578588645736,
      "learning_rate": 3.145435068316017e-05,
      "loss": 0.6597,
      "step": 2188
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5177087002321548,
      "learning_rate": 3.143216386980175e-05,
      "loss": 0.5756,
      "step": 2189
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6606919941260252,
      "learning_rate": 3.140997627130463e-05,
      "loss": 0.5953,
      "step": 2190
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6209763650079285,
      "learning_rate": 3.138778789983249e-05,
      "loss": 0.6397,
      "step": 2191
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6010130924274728,
      "learning_rate": 3.136559876754939e-05,
      "loss": 0.5906,
      "step": 2192
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.615672055527984,
      "learning_rate": 3.134340888661983e-05,
      "loss": 0.6156,
      "step": 2193
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6067325418911393,
      "learning_rate": 3.13212182692087e-05,
      "loss": 0.5893,
      "step": 2194
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.7656802946553205,
      "learning_rate": 3.1299026927481315e-05,
      "loss": 0.5559,
      "step": 2195
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.7340631218232923,
      "learning_rate": 3.1276834873603386e-05,
      "loss": 0.6211,
      "step": 2196
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.596317280669042,
      "learning_rate": 3.125464211974099e-05,
      "loss": 0.5716,
      "step": 2197
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.8938070296850018,
      "learning_rate": 3.123244867806061e-05,
      "loss": 0.6793,
      "step": 2198
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.556094328764993,
      "learning_rate": 3.1210254560729105e-05,
      "loss": 0.5639,
      "step": 2199
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.7096575256991033,
      "learning_rate": 3.1188059779913706e-05,
      "loss": 0.6072,
      "step": 2200
    },
    {
      "epoch": 2.51,
      "eval_blimp_filtered_avg": 0.706865671641791,
      "eval_blimp_filtered_std": 0.005042108990337878,
      "step": 2200
    },
    {
      "epoch": 2.51,
      "eval_blimp_supplement_avg": 0.8275862068965517,
      "eval_blimp_supplement_std": 0.017002752918024062,
      "step": 2200
    },
    {
      "epoch": 2.51,
      "eval_vqa_filtered_avg": 0.54,
      "eval_vqa_filtered_std": 0.05009082659620333,
      "step": 2200
    },
    {
      "epoch": 2.51,
      "eval_winoground_filtered_avg": 0.61,
      "eval_winoground_filtered_std": 0.04902071300001974,
      "step": 2200
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6404581266782507,
      "learning_rate": 3.1165864347781985e-05,
      "loss": 0.5436,
      "step": 2201
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.9964334301100275,
      "learning_rate": 3.11436682765019e-05,
      "loss": 0.6015,
      "step": 2202
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4477894616728546,
      "learning_rate": 3.112147157824175e-05,
      "loss": 0.4955,
      "step": 2203
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.7882931323942197,
      "learning_rate": 3.109927426517016e-05,
      "loss": 0.6035,
      "step": 2204
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6837551758691582,
      "learning_rate": 3.1077076349456116e-05,
      "loss": 0.5494,
      "step": 2205
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.42876500514236,
      "learning_rate": 3.105487784326893e-05,
      "loss": 0.5033,
      "step": 2206
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.563220229500069,
      "learning_rate": 3.103267875877823e-05,
      "loss": 0.5664,
      "step": 2207
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.648681658123781,
      "learning_rate": 3.1010479108153964e-05,
      "loss": 0.5818,
      "step": 2208
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6276815085546306,
      "learning_rate": 3.0988278903566375e-05,
      "loss": 0.5793,
      "step": 2209
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.8602034870800492,
      "learning_rate": 3.0966078157186053e-05,
      "loss": 0.6543,
      "step": 2210
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6299562563338967,
      "learning_rate": 3.094387688118384e-05,
      "loss": 0.5486,
      "step": 2211
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5082567064481156,
      "learning_rate": 3.092167508773089e-05,
      "loss": 0.4983,
      "step": 2212
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4446721249983223,
      "learning_rate": 3.089947278899863e-05,
      "loss": 0.536,
      "step": 2213
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4906983606217687,
      "learning_rate": 3.0877269997158775e-05,
      "loss": 0.5623,
      "step": 2214
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5756692118370474,
      "learning_rate": 3.085506672438331e-05,
      "loss": 0.5437,
      "step": 2215
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.8247361415423604,
      "learning_rate": 3.083286298284447e-05,
      "loss": 0.661,
      "step": 2216
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4024750178656167,
      "learning_rate": 3.081065878471477e-05,
      "loss": 0.5723,
      "step": 2217
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.533300049355846,
      "learning_rate": 3.0788454142166944e-05,
      "loss": 0.5455,
      "step": 2218
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.9906911485641967,
      "learning_rate": 3.076624906737401e-05,
      "loss": 0.6936,
      "step": 2219
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.534767332369444,
      "learning_rate": 3.074404357250917e-05,
      "loss": 0.5747,
      "step": 2220
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8512939975764506,
      "learning_rate": 3.0721837669745905e-05,
      "loss": 0.6926,
      "step": 2221
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.526553381710776,
      "learning_rate": 3.06996313712579e-05,
      "loss": 0.5359,
      "step": 2222
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4450321567777644,
      "learning_rate": 3.067742468921906e-05,
      "loss": 0.5051,
      "step": 2223
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.7065560718192576,
      "learning_rate": 3.065521763580349e-05,
      "loss": 0.5809,
      "step": 2224
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.7843799007267527,
      "learning_rate": 3.063301022318551e-05,
      "loss": 0.5395,
      "step": 2225
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6443749550103184,
      "learning_rate": 3.0610802463539635e-05,
      "loss": 0.5511,
      "step": 2226
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5897749945495985,
      "learning_rate": 3.058859436904057e-05,
      "loss": 0.5791,
      "step": 2227
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6250461043920728,
      "learning_rate": 3.056638595186321e-05,
      "loss": 0.5517,
      "step": 2228
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.688461170562411,
      "learning_rate": 3.05441772241826e-05,
      "loss": 0.598,
      "step": 2229
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.7260116235240255,
      "learning_rate": 3.052196819817399e-05,
      "loss": 0.618,
      "step": 2230
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5253922622666771,
      "learning_rate": 3.049975888601277e-05,
      "loss": 0.5448,
      "step": 2231
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6105773534616024,
      "learning_rate": 3.04775492998745e-05,
      "loss": 0.6364,
      "step": 2232
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.660744475073298,
      "learning_rate": 3.0455339451934886e-05,
      "loss": 0.5889,
      "step": 2233
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.511656056758855,
      "learning_rate": 3.0433129354369777e-05,
      "loss": 0.5419,
      "step": 2234
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5121680477474664,
      "learning_rate": 3.0410919019355153e-05,
      "loss": 0.5828,
      "step": 2235
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.518476946138721,
      "learning_rate": 3.0388708459067135e-05,
      "loss": 0.5546,
      "step": 2236
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.9303856994161501,
      "learning_rate": 3.036649768568196e-05,
      "loss": 0.6466,
      "step": 2237
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6494697025945217,
      "learning_rate": 3.034428671137599e-05,
      "loss": 0.5538,
      "step": 2238
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.7535897813523658,
      "learning_rate": 3.032207554832568e-05,
      "loss": 0.595,
      "step": 2239
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.8591299301079802,
      "learning_rate": 3.029986420870761e-05,
      "loss": 0.7014,
      "step": 2240
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6719626013350106,
      "learning_rate": 3.027765270469844e-05,
      "loss": 0.601,
      "step": 2241
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.635521364237335,
      "learning_rate": 3.0255441048474934e-05,
      "loss": 0.6618,
      "step": 2242
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.704751201825541,
      "learning_rate": 3.0233229252213926e-05,
      "loss": 0.6082,
      "step": 2243
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5086036653774302,
      "learning_rate": 3.0211017328092333e-05,
      "loss": 0.6113,
      "step": 2244
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.7679001044409794,
      "learning_rate": 3.018880528828714e-05,
      "loss": 0.6076,
      "step": 2245
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5842171087527677,
      "learning_rate": 3.0166593144975406e-05,
      "loss": 0.5959,
      "step": 2246
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.7854032544945193,
      "learning_rate": 3.0144380910334234e-05,
      "loss": 0.6312,
      "step": 2247
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.786657891885551,
      "learning_rate": 3.0122168596540768e-05,
      "loss": 0.6697,
      "step": 2248
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4968901110550112,
      "learning_rate": 3.009995621577224e-05,
      "loss": 0.538,
      "step": 2249
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6892871185340872,
      "learning_rate": 3.007774378020586e-05,
      "loss": 0.6304,
      "step": 2250
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.7289255588592345,
      "learning_rate": 3.0055531302018907e-05,
      "loss": 0.5502,
      "step": 2251
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5120558904084502,
      "learning_rate": 3.003331879338866e-05,
      "loss": 0.5818,
      "step": 2252
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4820762562021461,
      "learning_rate": 3.0011106266492447e-05,
      "loss": 0.5639,
      "step": 2253
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.7540688271305798,
      "learning_rate": 2.998889373350757e-05,
      "loss": 0.6624,
      "step": 2254
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4967303759544022,
      "learning_rate": 2.9966681206611346e-05,
      "loss": 0.5914,
      "step": 2255
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.599638919473847,
      "learning_rate": 2.9944468697981095e-05,
      "loss": 0.607,
      "step": 2256
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6155425286708518,
      "learning_rate": 2.992225621979414e-05,
      "loss": 0.5873,
      "step": 2257
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6969571957698861,
      "learning_rate": 2.990004378422777e-05,
      "loss": 0.6249,
      "step": 2258
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.7380130222545993,
      "learning_rate": 2.9877831403459237e-05,
      "loss": 0.5525,
      "step": 2259
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5590368028917374,
      "learning_rate": 2.985561908966577e-05,
      "loss": 0.5861,
      "step": 2260
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.8579976688562498,
      "learning_rate": 2.9833406855024596e-05,
      "loss": 0.6555,
      "step": 2261
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.655514733745974,
      "learning_rate": 2.981119471171287e-05,
      "loss": 0.5337,
      "step": 2262
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.8454033589297425,
      "learning_rate": 2.9788982671907676e-05,
      "loss": 0.6202,
      "step": 2263
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.573424962004041,
      "learning_rate": 2.9766770747786082e-05,
      "loss": 0.5906,
      "step": 2264
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.566518208115982,
      "learning_rate": 2.974455895152507e-05,
      "loss": 0.5838,
      "step": 2265
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.590546654166428,
      "learning_rate": 2.9722347295301557e-05,
      "loss": 0.5855,
      "step": 2266
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6119063570189078,
      "learning_rate": 2.9700135791292395e-05,
      "loss": 0.6056,
      "step": 2267
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.437436535868399,
      "learning_rate": 2.9677924451674324e-05,
      "loss": 0.534,
      "step": 2268
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5071619184201588,
      "learning_rate": 2.9655713288624014e-05,
      "loss": 0.5501,
      "step": 2269
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.530730061159738,
      "learning_rate": 2.963350231431804e-05,
      "loss": 0.5869,
      "step": 2270
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6237072482067696,
      "learning_rate": 2.961129154093287e-05,
      "loss": 0.572,
      "step": 2271
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4345213322281698,
      "learning_rate": 2.9589080980644852e-05,
      "loss": 0.5665,
      "step": 2272
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5902856799643306,
      "learning_rate": 2.9566870645630225e-05,
      "loss": 0.5783,
      "step": 2273
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4545185913310683,
      "learning_rate": 2.9544660548065112e-05,
      "loss": 0.5339,
      "step": 2274
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5792865639290172,
      "learning_rate": 2.952245070012551e-05,
      "loss": 0.5301,
      "step": 2275
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6699998738731663,
      "learning_rate": 2.9500241113987234e-05,
      "loss": 0.604,
      "step": 2276
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.7917662882755,
      "learning_rate": 2.947803180182602e-05,
      "loss": 0.5806,
      "step": 2277
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.7071380692142004,
      "learning_rate": 2.9455822775817406e-05,
      "loss": 0.5894,
      "step": 2278
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.7826469964617144,
      "learning_rate": 2.9433614048136805e-05,
      "loss": 0.5795,
      "step": 2279
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4682528809318345,
      "learning_rate": 2.9411405630959435e-05,
      "loss": 0.5168,
      "step": 2280
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5022223702187345,
      "learning_rate": 2.938919753646037e-05,
      "loss": 0.5216,
      "step": 2281
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.0109890413801033,
      "learning_rate": 2.9366989776814497e-05,
      "loss": 0.7192,
      "step": 2282
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6669425833874223,
      "learning_rate": 2.9344782364196515e-05,
      "loss": 0.6156,
      "step": 2283
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5866200946554407,
      "learning_rate": 2.932257531078095e-05,
      "loss": 0.5575,
      "step": 2284
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5925282612210412,
      "learning_rate": 2.9300368628742105e-05,
      "loss": 0.5729,
      "step": 2285
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5366562572774263,
      "learning_rate": 2.9278162330254104e-05,
      "loss": 0.5743,
      "step": 2286
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4896181221827096,
      "learning_rate": 2.9255956427490836e-05,
      "loss": 0.5496,
      "step": 2287
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.620566335977769,
      "learning_rate": 2.9233750932626007e-05,
      "loss": 0.5065,
      "step": 2288
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5293480326183142,
      "learning_rate": 2.921154585783306e-05,
      "loss": 0.5416,
      "step": 2289
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5596327051513459,
      "learning_rate": 2.9189341215285235e-05,
      "loss": 0.5362,
      "step": 2290
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.338113006167212,
      "learning_rate": 2.916713701715553e-05,
      "loss": 0.4983,
      "step": 2291
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5700016627043991,
      "learning_rate": 2.91449332756167e-05,
      "loss": 0.5449,
      "step": 2292
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.43899130137767,
      "learning_rate": 2.912273000284123e-05,
      "loss": 0.4932,
      "step": 2293
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5661309165504989,
      "learning_rate": 2.910052721100138e-05,
      "loss": 0.6114,
      "step": 2294
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5877739265782556,
      "learning_rate": 2.9078324912269112e-05,
      "loss": 0.5773,
      "step": 2295
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5637777068560315,
      "learning_rate": 2.9056123118816172e-05,
      "loss": 0.5669,
      "step": 2296
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4508441860432086,
      "learning_rate": 2.9033921842813958e-05,
      "loss": 0.5277,
      "step": 2297
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.936932488154957,
      "learning_rate": 2.9011721096433627e-05,
      "loss": 0.6772,
      "step": 2298
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4948941435515057,
      "learning_rate": 2.898952089184604e-05,
      "loss": 0.5489,
      "step": 2299
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6518444203545415,
      "learning_rate": 2.8967321241221767e-05,
      "loss": 0.6168,
      "step": 2300
    },
    {
      "epoch": 2.63,
      "eval_blimp_filtered_avg": 0.7061194029850746,
      "eval_blimp_filtered_std": 0.005048375266210442,
      "step": 2300
    },
    {
      "epoch": 2.63,
      "eval_blimp_supplement_avg": 0.834051724137931,
      "eval_blimp_supplement_std": 0.01671729969470888,
      "step": 2300
    },
    {
      "epoch": 2.63,
      "eval_vqa_filtered_avg": 0.53,
      "eval_vqa_filtered_std": 0.0501613558046592,
      "step": 2300
    },
    {
      "epoch": 2.63,
      "eval_winoground_filtered_avg": 0.63,
      "eval_winoground_filtered_std": 0.04852365870939099,
      "step": 2300
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6414032492929658,
      "learning_rate": 2.894512215673108e-05,
      "loss": 0.5978,
      "step": 2301
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6331120531579102,
      "learning_rate": 2.892292365054389e-05,
      "loss": 0.6017,
      "step": 2302
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4424008382975424,
      "learning_rate": 2.890072573482984e-05,
      "loss": 0.5374,
      "step": 2303
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.376235423146149,
      "learning_rate": 2.8878528421758255e-05,
      "loss": 0.4864,
      "step": 2304
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.504198733527538,
      "learning_rate": 2.885633172349811e-05,
      "loss": 0.5358,
      "step": 2305
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6231380001640416,
      "learning_rate": 2.8834135652218017e-05,
      "loss": 0.5944,
      "step": 2306
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7167790841110846,
      "learning_rate": 2.88119402200863e-05,
      "loss": 0.5743,
      "step": 2307
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7345762743280357,
      "learning_rate": 2.8789745439270893e-05,
      "loss": 0.6109,
      "step": 2308
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6547235500374302,
      "learning_rate": 2.8767551321939403e-05,
      "loss": 0.5816,
      "step": 2309
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7150975309417988,
      "learning_rate": 2.874535788025902e-05,
      "loss": 0.5509,
      "step": 2310
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3846295951497132,
      "learning_rate": 2.8723165126396625e-05,
      "loss": 0.4705,
      "step": 2311
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6606684662449667,
      "learning_rate": 2.8700973072518686e-05,
      "loss": 0.5743,
      "step": 2312
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7153087615906182,
      "learning_rate": 2.8678781730791314e-05,
      "loss": 0.578,
      "step": 2313
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.9098793963101643,
      "learning_rate": 2.865659111338018e-05,
      "loss": 0.6947,
      "step": 2314
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.609554586089545,
      "learning_rate": 2.8634401232450612e-05,
      "loss": 0.5421,
      "step": 2315
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4612158852750046,
      "learning_rate": 2.8612212100167514e-05,
      "loss": 0.4946,
      "step": 2316
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5856325946529002,
      "learning_rate": 2.859002372869537e-05,
      "loss": 0.6068,
      "step": 2317
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6842504713271629,
      "learning_rate": 2.8567836130198264e-05,
      "loss": 0.5552,
      "step": 2318
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6949911639852746,
      "learning_rate": 2.8545649316839834e-05,
      "loss": 0.5711,
      "step": 2319
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.9135446272426693,
      "learning_rate": 2.8523463300783308e-05,
      "loss": 0.6036,
      "step": 2320
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6814295936588917,
      "learning_rate": 2.850127809419147e-05,
      "loss": 0.5706,
      "step": 2321
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5598080442968034,
      "learning_rate": 2.847909370922667e-05,
      "loss": 0.6015,
      "step": 2322
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.532151863146623,
      "learning_rate": 2.8456910158050777e-05,
      "loss": 0.5641,
      "step": 2323
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3945554460921492,
      "learning_rate": 2.8434727452825238e-05,
      "loss": 0.4932,
      "step": 2324
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6589998623210682,
      "learning_rate": 2.8412545605711013e-05,
      "loss": 0.5911,
      "step": 2325
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6430566664440158,
      "learning_rate": 2.8390364628868614e-05,
      "loss": 0.6077,
      "step": 2326
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.824344433652133,
      "learning_rate": 2.8368184534458044e-05,
      "loss": 0.6204,
      "step": 2327
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6432225193357877,
      "learning_rate": 2.834600533463885e-05,
      "loss": 0.5799,
      "step": 2328
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.557228216975582,
      "learning_rate": 2.8323827041570082e-05,
      "loss": 0.557,
      "step": 2329
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6941553042052697,
      "learning_rate": 2.8301649667410288e-05,
      "loss": 0.5888,
      "step": 2330
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.59524931493396,
      "learning_rate": 2.8279473224317505e-05,
      "loss": 0.5937,
      "step": 2331
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.8520272998543785,
      "learning_rate": 2.8257297724449282e-05,
      "loss": 0.6258,
      "step": 2332
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.7385529079179611,
      "learning_rate": 2.8235123179962625e-05,
      "loss": 0.6125,
      "step": 2333
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4692219467363037,
      "learning_rate": 2.821294960301404e-05,
      "loss": 0.5085,
      "step": 2334
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.8938148373210808,
      "learning_rate": 2.819077700575949e-05,
      "loss": 0.6165,
      "step": 2335
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6185865910819133,
      "learning_rate": 2.8168605400354395e-05,
      "loss": 0.5893,
      "step": 2336
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.795443811460055,
      "learning_rate": 2.814643479895364e-05,
      "loss": 0.6391,
      "step": 2337
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6791661255405617,
      "learning_rate": 2.812426521371156e-05,
      "loss": 0.6095,
      "step": 2338
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4504658723613981,
      "learning_rate": 2.8102096656781943e-05,
      "loss": 0.5403,
      "step": 2339
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4021559110375863,
      "learning_rate": 2.807992914031798e-05,
      "loss": 0.5303,
      "step": 2340
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6973625609847574,
      "learning_rate": 2.8057762676472323e-05,
      "loss": 0.5511,
      "step": 2341
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6518172173986865,
      "learning_rate": 2.803559727739703e-05,
      "loss": 0.573,
      "step": 2342
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4359125203594734,
      "learning_rate": 2.801343295524361e-05,
      "loss": 0.5395,
      "step": 2343
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4156167492097242,
      "learning_rate": 2.799126972216292e-05,
      "loss": 0.5009,
      "step": 2344
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.445608568951133,
      "learning_rate": 2.7969107590305266e-05,
      "loss": 0.5106,
      "step": 2345
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6374658137347486,
      "learning_rate": 2.794694657182033e-05,
      "loss": 0.5611,
      "step": 2346
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.7148409911641542,
      "learning_rate": 2.792478667885722e-05,
      "loss": 0.599,
      "step": 2347
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6456052977365658,
      "learning_rate": 2.790262792356437e-05,
      "loss": 0.5665,
      "step": 2348
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4754215451210724,
      "learning_rate": 2.7880470318089623e-05,
      "loss": 0.565,
      "step": 2349
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4818464983049628,
      "learning_rate": 2.7858313874580186e-05,
      "loss": 0.5331,
      "step": 2350
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3275234266626774,
      "learning_rate": 2.7836158605182657e-05,
      "loss": 0.5502,
      "step": 2351
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5693103304677283,
      "learning_rate": 2.7814004522042932e-05,
      "loss": 0.5707,
      "step": 2352
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3292300269161954,
      "learning_rate": 2.7791851637306292e-05,
      "loss": 0.4499,
      "step": 2353
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6623766768049064,
      "learning_rate": 2.776969996311737e-05,
      "loss": 0.5735,
      "step": 2354
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6580557783866272,
      "learning_rate": 2.7747549511620123e-05,
      "loss": 0.6196,
      "step": 2355
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4389577321129061,
      "learning_rate": 2.7725400294957842e-05,
      "loss": 0.5189,
      "step": 2356
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.7151682746412018,
      "learning_rate": 2.7703252325273108e-05,
      "loss": 0.615,
      "step": 2357
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.8000642545281413,
      "learning_rate": 2.7681105614707875e-05,
      "loss": 0.6059,
      "step": 2358
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.8232603197073112,
      "learning_rate": 2.765896017540337e-05,
      "loss": 0.5778,
      "step": 2359
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.7125590097399948,
      "learning_rate": 2.7636816019500145e-05,
      "loss": 0.5755,
      "step": 2360
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4489324916547686,
      "learning_rate": 2.7614673159138003e-05,
      "loss": 0.5641,
      "step": 2361
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6586428028097826,
      "learning_rate": 2.7592531606456094e-05,
      "loss": 0.5494,
      "step": 2362
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5428337823634852,
      "learning_rate": 2.7570391373592824e-05,
      "loss": 0.5624,
      "step": 2363
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.470567526608812,
      "learning_rate": 2.754825247268588e-05,
      "loss": 0.5441,
      "step": 2364
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4230809479237576,
      "learning_rate": 2.7526114915872198e-05,
      "loss": 0.524,
      "step": 2365
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.582295315975592,
      "learning_rate": 2.7503978715288014e-05,
      "loss": 0.6043,
      "step": 2366
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.883041270144907,
      "learning_rate": 2.7481843883068794e-05,
      "loss": 0.6161,
      "step": 2367
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4744744322406345,
      "learning_rate": 2.745971043134927e-05,
      "loss": 0.5691,
      "step": 2368
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5493133492619338,
      "learning_rate": 2.7437578372263397e-05,
      "loss": 0.5373,
      "step": 2369
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6755287305585607,
      "learning_rate": 2.7415447717944384e-05,
      "loss": 0.5791,
      "step": 2370
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5575643495839127,
      "learning_rate": 2.7393318480524664e-05,
      "loss": 0.5586,
      "step": 2371
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4517736359683862,
      "learning_rate": 2.7371190672135893e-05,
      "loss": 0.4933,
      "step": 2372
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.550584491307284,
      "learning_rate": 2.7349064304908958e-05,
      "loss": 0.5267,
      "step": 2373
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5155331449818732,
      "learning_rate": 2.7326939390973916e-05,
      "loss": 0.5022,
      "step": 2374
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5602757267570024,
      "learning_rate": 2.7304815942460073e-05,
      "loss": 0.5875,
      "step": 2375
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.896222616137569,
      "learning_rate": 2.7282693971495904e-05,
      "loss": 0.602,
      "step": 2376
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.824388892747857,
      "learning_rate": 2.7260573490209087e-05,
      "loss": 0.5896,
      "step": 2377
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.650657394524827,
      "learning_rate": 2.7238454510726475e-05,
      "loss": 0.5718,
      "step": 2378
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.637544199011248,
      "learning_rate": 2.7216337045174097e-05,
      "loss": 0.5192,
      "step": 2379
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4978668514258793,
      "learning_rate": 2.7194221105677164e-05,
      "loss": 0.4875,
      "step": 2380
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5575113203055093,
      "learning_rate": 2.7172106704360042e-05,
      "loss": 0.5701,
      "step": 2381
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.693494600295127,
      "learning_rate": 2.714999385334625e-05,
      "loss": 0.5652,
      "step": 2382
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.864022389042997,
      "learning_rate": 2.712788256475846e-05,
      "loss": 0.5353,
      "step": 2383
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5856588451640898,
      "learning_rate": 2.7105772850718493e-05,
      "loss": 0.5176,
      "step": 2384
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3520634926216935,
      "learning_rate": 2.7083664723347303e-05,
      "loss": 0.4642,
      "step": 2385
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6441117450760026,
      "learning_rate": 2.706155819476497e-05,
      "loss": 0.5542,
      "step": 2386
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5152770868260028,
      "learning_rate": 2.7039453277090704e-05,
      "loss": 0.5323,
      "step": 2387
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.589930284813545,
      "learning_rate": 2.7017349982442826e-05,
      "loss": 0.5435,
      "step": 2388
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4605640498559052,
      "learning_rate": 2.6995248322938765e-05,
      "loss": 0.5176,
      "step": 2389
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9574691895743233,
      "learning_rate": 2.6973148310695086e-05,
      "loss": 0.5402,
      "step": 2390
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9402825957290692,
      "learning_rate": 2.6951049957827395e-05,
      "loss": 0.6061,
      "step": 2391
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.7536830049364627,
      "learning_rate": 2.6928953276450423e-05,
      "loss": 0.5902,
      "step": 2392
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.754274055758335,
      "learning_rate": 2.6906858278677976e-05,
      "loss": 0.5904,
      "step": 2393
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.430888751441061,
      "learning_rate": 2.6884764976622966e-05,
      "loss": 0.4993,
      "step": 2394
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.8475592547735944,
      "learning_rate": 2.686267338239732e-05,
      "loss": 0.6518,
      "step": 2395
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.7264345357550352,
      "learning_rate": 2.6840583508112065e-05,
      "loss": 0.6069,
      "step": 2396
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.8541958103193652,
      "learning_rate": 2.6818495365877268e-05,
      "loss": 0.5496,
      "step": 2397
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.7524875464085055,
      "learning_rate": 2.6796408967802097e-05,
      "loss": 0.6078,
      "step": 2398
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.9085413517667686,
      "learning_rate": 2.6774324325994678e-05,
      "loss": 0.6233,
      "step": 2399
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4583495444235126,
      "learning_rate": 2.675224145256223e-05,
      "loss": 0.5752,
      "step": 2400
    },
    {
      "epoch": 2.74,
      "eval_blimp_filtered_avg": 0.7083582089552238,
      "eval_blimp_filtered_std": 0.005046569771334871,
      "step": 2400
    },
    {
      "epoch": 2.74,
      "eval_blimp_supplement_avg": 0.8362068965517241,
      "eval_blimp_supplement_std": 0.016715895292665435,
      "step": 2400
    },
    {
      "epoch": 2.74,
      "eval_vqa_filtered_avg": 0.54,
      "eval_vqa_filtered_std": 0.05009082659620333,
      "step": 2400
    },
    {
      "epoch": 2.74,
      "eval_winoground_filtered_avg": 0.61,
      "eval_winoground_filtered_std": 0.04902071300001975,
      "step": 2400
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5890023974806549,
      "learning_rate": 2.6730160359611007e-05,
      "loss": 0.5446,
      "step": 2401
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5855769815431624,
      "learning_rate": 2.6708081059246276e-05,
      "loss": 0.591,
      "step": 2402
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.499338966675462,
      "learning_rate": 2.668600356357229e-05,
      "loss": 0.4841,
      "step": 2403
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.9631296289153768,
      "learning_rate": 2.6663927884692352e-05,
      "loss": 0.6528,
      "step": 2404
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.9536289579089834,
      "learning_rate": 2.6641854034708765e-05,
      "loss": 0.6296,
      "step": 2405
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6207037420396577,
      "learning_rate": 2.6619782025722815e-05,
      "loss": 0.5608,
      "step": 2406
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5768095966770275,
      "learning_rate": 2.6597711869834798e-05,
      "loss": 0.5529,
      "step": 2407
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.7497613652345405,
      "learning_rate": 2.6575643579143952e-05,
      "loss": 0.5619,
      "step": 2408
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5347792520248746,
      "learning_rate": 2.6553577165748542e-05,
      "loss": 0.5764,
      "step": 2409
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3260045451401616,
      "learning_rate": 2.6531512641745782e-05,
      "loss": 0.4473,
      "step": 2410
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.768131172473859,
      "learning_rate": 2.6509450019231853e-05,
      "loss": 0.5429,
      "step": 2411
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.9068567561057217,
      "learning_rate": 2.6487389310301887e-05,
      "loss": 0.6248,
      "step": 2412
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4238819923168322,
      "learning_rate": 2.646533052704997e-05,
      "loss": 0.4916,
      "step": 2413
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.650314508817564,
      "learning_rate": 2.644327368156914e-05,
      "loss": 0.5124,
      "step": 2414
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4534604862944422,
      "learning_rate": 2.642121878595137e-05,
      "loss": 0.5056,
      "step": 2415
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5116465315077083,
      "learning_rate": 2.6399165852287547e-05,
      "loss": 0.5055,
      "step": 2416
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.7145482470611844,
      "learning_rate": 2.6377114892667508e-05,
      "loss": 0.5955,
      "step": 2417
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3249766780375434,
      "learning_rate": 2.6355065919179997e-05,
      "loss": 0.4806,
      "step": 2418
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6100552371136918,
      "learning_rate": 2.633301894391267e-05,
      "loss": 0.5661,
      "step": 2419
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4156629554921591,
      "learning_rate": 2.6310973978952073e-05,
      "loss": 0.5016,
      "step": 2420
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6944254617127148,
      "learning_rate": 2.6288931036383673e-05,
      "loss": 0.5604,
      "step": 2421
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5249502196708427,
      "learning_rate": 2.6266890128291817e-05,
      "loss": 0.5218,
      "step": 2422
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4603944993360072,
      "learning_rate": 2.6244851266759734e-05,
      "loss": 0.5856,
      "step": 2423
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.463095173523544,
      "learning_rate": 2.6222814463869547e-05,
      "loss": 0.5505,
      "step": 2424
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5583605754522811,
      "learning_rate": 2.6200779731702226e-05,
      "loss": 0.5388,
      "step": 2425
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.7384914817737895,
      "learning_rate": 2.617874708233762e-05,
      "loss": 0.5291,
      "step": 2426
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6322823776681425,
      "learning_rate": 2.6156716527854433e-05,
      "loss": 0.5,
      "step": 2427
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4172197746176067,
      "learning_rate": 2.6134688080330233e-05,
      "loss": 0.5234,
      "step": 2428
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.7452846407669764,
      "learning_rate": 2.611266175184141e-05,
      "loss": 0.5359,
      "step": 2429
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6710581410847418,
      "learning_rate": 2.6090637554463205e-05,
      "loss": 0.5409,
      "step": 2430
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.803129127776329,
      "learning_rate": 2.6068615500269694e-05,
      "loss": 0.5842,
      "step": 2431
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.8722108085406193,
      "learning_rate": 2.6046595601333777e-05,
      "loss": 0.5989,
      "step": 2432
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.002621268575036,
      "learning_rate": 2.6024577869727157e-05,
      "loss": 0.6539,
      "step": 2433
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6326759414652372,
      "learning_rate": 2.6002562317520373e-05,
      "loss": 0.5581,
      "step": 2434
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.8633208734305997,
      "learning_rate": 2.598054895678275e-05,
      "loss": 0.6445,
      "step": 2435
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.7746671157191858,
      "learning_rate": 2.595853779958243e-05,
      "loss": 0.5633,
      "step": 2436
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4755252835898276,
      "learning_rate": 2.5936528857986318e-05,
      "loss": 0.5029,
      "step": 2437
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6599842088486934,
      "learning_rate": 2.5914522144060136e-05,
      "loss": 0.6532,
      "step": 2438
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.614379277599864,
      "learning_rate": 2.5892517669868366e-05,
      "loss": 0.5921,
      "step": 2439
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.378573456328833,
      "learning_rate": 2.5870515447474262e-05,
      "loss": 0.5215,
      "step": 2440
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.380286276611997,
      "learning_rate": 2.584851548893987e-05,
      "loss": 0.523,
      "step": 2441
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6267546442763068,
      "learning_rate": 2.582651780632595e-05,
      "loss": 0.5798,
      "step": 2442
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.794271119984938,
      "learning_rate": 2.5804522411692043e-05,
      "loss": 0.5456,
      "step": 2443
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.9930498240851715,
      "learning_rate": 2.5782529317096435e-05,
      "loss": 0.6412,
      "step": 2444
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6723596010274124,
      "learning_rate": 2.576053853459616e-05,
      "loss": 0.5999,
      "step": 2445
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.648129388183208,
      "learning_rate": 2.5738550076246942e-05,
      "loss": 0.4963,
      "step": 2446
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7495647552836808,
      "learning_rate": 2.571656395410327e-05,
      "loss": 0.5943,
      "step": 2447
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7691762077797675,
      "learning_rate": 2.5694580180218353e-05,
      "loss": 0.5716,
      "step": 2448
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5384369171310488,
      "learning_rate": 2.567259876664411e-05,
      "loss": 0.5656,
      "step": 2449
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5420366921673927,
      "learning_rate": 2.565061972543113e-05,
      "loss": 0.541,
      "step": 2450
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.8624024345189993,
      "learning_rate": 2.5628643068628732e-05,
      "loss": 0.6027,
      "step": 2451
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.586489454128573,
      "learning_rate": 2.5606668808284946e-05,
      "loss": 0.5744,
      "step": 2452
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.646907192146188,
      "learning_rate": 2.5584696956446468e-05,
      "loss": 0.5643,
      "step": 2453
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6429844307303405,
      "learning_rate": 2.5562727525158654e-05,
      "loss": 0.6033,
      "step": 2454
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6472729045412704,
      "learning_rate": 2.554076052646555e-05,
      "loss": 0.5808,
      "step": 2455
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4187267930724674,
      "learning_rate": 2.5518795972409893e-05,
      "loss": 0.5391,
      "step": 2456
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.7659687570041256,
      "learning_rate": 2.549683387503305e-05,
      "loss": 0.562,
      "step": 2457
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6785680280500632,
      "learning_rate": 2.5474874246375024e-05,
      "loss": 0.6047,
      "step": 2458
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.780751983216869,
      "learning_rate": 2.545291709847452e-05,
      "loss": 0.5593,
      "step": 2459
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3976992435505207,
      "learning_rate": 2.543096244336884e-05,
      "loss": 0.4786,
      "step": 2460
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5894652544801404,
      "learning_rate": 2.5409010293093926e-05,
      "loss": 0.5087,
      "step": 2461
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6405232025384233,
      "learning_rate": 2.5387060659684367e-05,
      "loss": 0.5267,
      "step": 2462
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.8685042025115508,
      "learning_rate": 2.5365113555173343e-05,
      "loss": 0.6079,
      "step": 2463
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9492361151384117,
      "learning_rate": 2.5343168991592666e-05,
      "loss": 0.5993,
      "step": 2464
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3417495468568617,
      "learning_rate": 2.5321226980972755e-05,
      "loss": 0.468,
      "step": 2465
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.614616263807012,
      "learning_rate": 2.529928753534263e-05,
      "loss": 0.5812,
      "step": 2466
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5160354705382533,
      "learning_rate": 2.5277350666729878e-05,
      "loss": 0.487,
      "step": 2467
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.7622683948397793,
      "learning_rate": 2.5255416387160718e-05,
      "loss": 0.5797,
      "step": 2468
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.63102370536342,
      "learning_rate": 2.5233484708659915e-05,
      "loss": 0.5326,
      "step": 2469
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.830794699048526,
      "learning_rate": 2.5211555643250835e-05,
      "loss": 0.6042,
      "step": 2470
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6065603955896028,
      "learning_rate": 2.518962920295538e-05,
      "loss": 0.5593,
      "step": 2471
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6093970099865766,
      "learning_rate": 2.5167705399794035e-05,
      "loss": 0.5617,
      "step": 2472
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5363070121943423,
      "learning_rate": 2.5145784245785836e-05,
      "loss": 0.5338,
      "step": 2473
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.7070207464332203,
      "learning_rate": 2.5123865752948366e-05,
      "loss": 0.6365,
      "step": 2474
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.760797309941051,
      "learning_rate": 2.5101949933297742e-05,
      "loss": 0.5526,
      "step": 2475
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6053345065983649,
      "learning_rate": 2.5080036798848623e-05,
      "loss": 0.5945,
      "step": 2476
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.701705537105333,
      "learning_rate": 2.5058126361614196e-05,
      "loss": 0.5682,
      "step": 2477
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4799589760921794,
      "learning_rate": 2.5036218633606162e-05,
      "loss": 0.4972,
      "step": 2478
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6325366220423247,
      "learning_rate": 2.501431362683476e-05,
      "loss": 0.5228,
      "step": 2479
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3708996523601025,
      "learning_rate": 2.4992411353308692e-05,
      "loss": 0.4758,
      "step": 2480
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.9565325528638586,
      "learning_rate": 2.4970511825035204e-05,
      "loss": 0.6441,
      "step": 2481
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6735243792803307,
      "learning_rate": 2.4948615054020026e-05,
      "loss": 0.587,
      "step": 2482
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5509557657716617,
      "learning_rate": 2.4926721052267368e-05,
      "loss": 0.505,
      "step": 2483
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.658008526233024,
      "learning_rate": 2.4904829831779922e-05,
      "loss": 0.5227,
      "step": 2484
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6662078333520949,
      "learning_rate": 2.4882941404558857e-05,
      "loss": 0.6058,
      "step": 2485
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.8362524665492055,
      "learning_rate": 2.4861055782603818e-05,
      "loss": 0.5948,
      "step": 2486
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4774112763971732,
      "learning_rate": 2.483917297791292e-05,
      "loss": 0.504,
      "step": 2487
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4036194367805415,
      "learning_rate": 2.48172930024827e-05,
      "loss": 0.4575,
      "step": 2488
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6288794041769565,
      "learning_rate": 2.479541586830817e-05,
      "loss": 0.5408,
      "step": 2489
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6526532229339757,
      "learning_rate": 2.4773541587382774e-05,
      "loss": 0.5089,
      "step": 2490
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.554718035570261,
      "learning_rate": 2.4751670171698424e-05,
      "loss": 0.5415,
      "step": 2491
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5586240546393613,
      "learning_rate": 2.4729801633245407e-05,
      "loss": 0.5703,
      "step": 2492
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4349246350567475,
      "learning_rate": 2.470793598401246e-05,
      "loss": 0.4576,
      "step": 2493
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4072994979809836,
      "learning_rate": 2.468607323598674e-05,
      "loss": 0.4844,
      "step": 2494
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5215708574262183,
      "learning_rate": 2.4664213401153817e-05,
      "loss": 0.5681,
      "step": 2495
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5803750227028286,
      "learning_rate": 2.4642356491497667e-05,
      "loss": 0.5791,
      "step": 2496
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5669052903716503,
      "learning_rate": 2.462050251900062e-05,
      "loss": 0.5631,
      "step": 2497
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.7635380281823394,
      "learning_rate": 2.4598651495643435e-05,
      "loss": 0.6302,
      "step": 2498
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.511265834071214,
      "learning_rate": 2.4576803433405265e-05,
      "loss": 0.5517,
      "step": 2499
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.511931973102632,
      "learning_rate": 2.4554958344263625e-05,
      "loss": 0.5076,
      "step": 2500
    },
    {
      "epoch": 2.86,
      "eval_blimp_filtered_avg": 0.7077611940298507,
      "eval_blimp_filtered_std": 0.00503768892626433,
      "step": 2500
    },
    {
      "epoch": 2.86,
      "eval_blimp_supplement_avg": 0.8297413793103449,
      "eval_blimp_supplement_std": 0.016954357231514238,
      "step": 2500
    },
    {
      "epoch": 2.86,
      "eval_vqa_filtered_avg": 0.53,
      "eval_vqa_filtered_std": 0.0501613558046592,
      "step": 2500
    },
    {
      "epoch": 2.86,
      "eval_winoground_filtered_avg": 0.61,
      "eval_winoground_filtered_std": 0.04902071300001975,
      "step": 2500
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5286621468871062,
      "learning_rate": 2.4533116240194357e-05,
      "loss": 0.5833,
      "step": 2501
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.8668613120594766,
      "learning_rate": 2.451127713317174e-05,
      "loss": 0.6164,
      "step": 2502
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6054552185841249,
      "learning_rate": 2.4489441035168366e-05,
      "loss": 0.5919,
      "step": 2503
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6611851980805905,
      "learning_rate": 2.44676079581552e-05,
      "loss": 0.6113,
      "step": 2504
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.8339552813962121,
      "learning_rate": 2.4445777914101513e-05,
      "loss": 0.5932,
      "step": 2505
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6716300160882853,
      "learning_rate": 2.4423950914974953e-05,
      "loss": 0.5966,
      "step": 2506
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5912951043128012,
      "learning_rate": 2.4402126972741486e-05,
      "loss": 0.4877,
      "step": 2507
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.7154882765792576,
      "learning_rate": 2.4380306099365405e-05,
      "loss": 0.5849,
      "step": 2508
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.714650291660585,
      "learning_rate": 2.435848830680929e-05,
      "loss": 0.5706,
      "step": 2509
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4631447280610446,
      "learning_rate": 2.433667360703408e-05,
      "loss": 0.468,
      "step": 2510
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.7934473086404177,
      "learning_rate": 2.431486201199899e-05,
      "loss": 0.5753,
      "step": 2511
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.7303633542211778,
      "learning_rate": 2.4293053533661546e-05,
      "loss": 0.5329,
      "step": 2512
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6896023237240552,
      "learning_rate": 2.427124818397756e-05,
      "loss": 0.5792,
      "step": 2513
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6936888088340405,
      "learning_rate": 2.424944597490111e-05,
      "loss": 0.5102,
      "step": 2514
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.707092378178774,
      "learning_rate": 2.4227646918384578e-05,
      "loss": 0.5576,
      "step": 2515
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.047641396109776,
      "learning_rate": 2.4205851026378616e-05,
      "loss": 0.6459,
      "step": 2516
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.921952353497433,
      "learning_rate": 2.4184058310832136e-05,
      "loss": 0.6019,
      "step": 2517
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6698917051009101,
      "learning_rate": 2.4162268783692298e-05,
      "loss": 0.5844,
      "step": 2518
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5876019548615217,
      "learning_rate": 2.4140482456904525e-05,
      "loss": 0.56,
      "step": 2519
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.503578033110074,
      "learning_rate": 2.4118699342412483e-05,
      "loss": 0.5559,
      "step": 2520
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6337188503512856,
      "learning_rate": 2.4096919452158088e-05,
      "loss": 0.5346,
      "step": 2521
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4242859526247236,
      "learning_rate": 2.407514279808146e-05,
      "loss": 0.481,
      "step": 2522
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6318707395577712,
      "learning_rate": 2.4053369392120978e-05,
      "loss": 0.5725,
      "step": 2523
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3919425512166943,
      "learning_rate": 2.4031599246213216e-05,
      "loss": 0.5004,
      "step": 2524
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5760304778183454,
      "learning_rate": 2.4009832372292985e-05,
      "loss": 0.5394,
      "step": 2525
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.7094298169402913,
      "learning_rate": 2.3988068782293268e-05,
      "loss": 0.585,
      "step": 2526
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.736440208476538,
      "learning_rate": 2.3966308488145277e-05,
      "loss": 0.6034,
      "step": 2527
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4983680785018814,
      "learning_rate": 2.39445515017784e-05,
      "loss": 0.4962,
      "step": 2528
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5370376539578916,
      "learning_rate": 2.3922797835120234e-05,
      "loss": 0.5291,
      "step": 2529
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.8426650575043322,
      "learning_rate": 2.390104750009653e-05,
      "loss": 0.5679,
      "step": 2530
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5029315312688272,
      "learning_rate": 2.3879300508631223e-05,
      "loss": 0.4951,
      "step": 2531
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.9566296658024793,
      "learning_rate": 2.3857556872646417e-05,
      "loss": 0.6377,
      "step": 2532
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.7824753881360509,
      "learning_rate": 2.383581660406237e-05,
      "loss": 0.6222,
      "step": 2533
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6854856733395123,
      "learning_rate": 2.3814079714797517e-05,
      "loss": 0.5635,
      "step": 2534
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5650817215150223,
      "learning_rate": 2.3792346216768394e-05,
      "loss": 0.4905,
      "step": 2535
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4776356275009017,
      "learning_rate": 2.3770616121889728e-05,
      "loss": 0.5259,
      "step": 2536
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.148542646620855,
      "learning_rate": 2.374888944207434e-05,
      "loss": 0.6644,
      "step": 2537
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.7514340525900207,
      "learning_rate": 2.372716618923323e-05,
      "loss": 0.5523,
      "step": 2538
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.7002675869036674,
      "learning_rate": 2.370544637527545e-05,
      "loss": 0.5708,
      "step": 2539
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6572783740572865,
      "learning_rate": 2.368373001210822e-05,
      "loss": 0.5352,
      "step": 2540
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5803846331984346,
      "learning_rate": 2.366201711163684e-05,
      "loss": 0.5482,
      "step": 2541
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4762362147502803,
      "learning_rate": 2.3640307685764752e-05,
      "loss": 0.5367,
      "step": 2542
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6727900323404599,
      "learning_rate": 2.361860174639343e-05,
      "loss": 0.5762,
      "step": 2543
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4102544547478615,
      "learning_rate": 2.359689930542249e-05,
      "loss": 0.4604,
      "step": 2544
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6960413848174793,
      "learning_rate": 2.3575200374749596e-05,
      "loss": 0.5754,
      "step": 2545
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3222538263306851,
      "learning_rate": 2.3553504966270532e-05,
      "loss": 0.4526,
      "step": 2546
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4668830654681182,
      "learning_rate": 2.3531813091879093e-05,
      "loss": 0.4848,
      "step": 2547
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6261847779797634,
      "learning_rate": 2.3510124763467163e-05,
      "loss": 0.5672,
      "step": 2548
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.8814336851572064,
      "learning_rate": 2.3488439992924706e-05,
      "loss": 0.5758,
      "step": 2549
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.62873228794386,
      "learning_rate": 2.34667587921397e-05,
      "loss": 0.5606,
      "step": 2550
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6745314108953606,
      "learning_rate": 2.3445081172998197e-05,
      "loss": 0.5573,
      "step": 2551
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6813920513623368,
      "learning_rate": 2.3423407147384237e-05,
      "loss": 0.5682,
      "step": 2552
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.88012486807738,
      "learning_rate": 2.340173672717994e-05,
      "loss": 0.609,
      "step": 2553
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.919606203573852,
      "learning_rate": 2.3380069924265434e-05,
      "loss": 0.6288,
      "step": 2554
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6911207519536529,
      "learning_rate": 2.3358406750518865e-05,
      "loss": 0.551,
      "step": 2555
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3824750825664498,
      "learning_rate": 2.3336747217816357e-05,
      "loss": 0.4561,
      "step": 2556
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.7143022575686937,
      "learning_rate": 2.3315091338032088e-05,
      "loss": 0.5865,
      "step": 2557
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5515708572622944,
      "learning_rate": 2.3293439123038213e-05,
      "loss": 0.5066,
      "step": 2558
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4392909577851323,
      "learning_rate": 2.3271790584704874e-05,
      "loss": 0.4976,
      "step": 2559
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5949542793115068,
      "learning_rate": 2.3250145734900187e-05,
      "loss": 0.4987,
      "step": 2560
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.9185733190419743,
      "learning_rate": 2.322850458549027e-05,
      "loss": 0.576,
      "step": 2561
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5653660705395396,
      "learning_rate": 2.32068671483392e-05,
      "loss": 0.5115,
      "step": 2562
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.761549683465301,
      "learning_rate": 2.3185233435309022e-05,
      "loss": 0.5744,
      "step": 2563
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5325978297977152,
      "learning_rate": 2.3163603458259727e-05,
      "loss": 0.504,
      "step": 2564
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4655820584107524,
      "learning_rate": 2.3141977229049276e-05,
      "loss": 0.4727,
      "step": 2565
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.617732165905792,
      "learning_rate": 2.3120354759533563e-05,
      "loss": 0.5252,
      "step": 2566
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5204375345667147,
      "learning_rate": 2.3098736061566425e-05,
      "loss": 0.5405,
      "step": 2567
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.898260481740341,
      "learning_rate": 2.307712114699964e-05,
      "loss": 0.6155,
      "step": 2568
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4827898043607493,
      "learning_rate": 2.3055510027682894e-05,
      "loss": 0.4841,
      "step": 2569
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.709374023674404,
      "learning_rate": 2.3033902715463806e-05,
      "loss": 0.5495,
      "step": 2570
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.7643758121520166,
      "learning_rate": 2.30122992221879e-05,
      "loss": 0.5723,
      "step": 2571
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5523819331583761,
      "learning_rate": 2.2990699559698616e-05,
      "loss": 0.504,
      "step": 2572
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5839209833601204,
      "learning_rate": 2.2969103739837284e-05,
      "loss": 0.5577,
      "step": 2573
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4482724323037721,
      "learning_rate": 2.294751177444313e-05,
      "loss": 0.5217,
      "step": 2574
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6906444667115583,
      "learning_rate": 2.2925923675353263e-05,
      "loss": 0.5676,
      "step": 2575
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6697282028762062,
      "learning_rate": 2.2904339454402697e-05,
      "loss": 0.5663,
      "step": 2576
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5151733361889286,
      "learning_rate": 2.288275912342428e-05,
      "loss": 0.4979,
      "step": 2577
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.894149307623881,
      "learning_rate": 2.2861182694248757e-05,
      "loss": 0.6322,
      "step": 2578
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6051188701787136,
      "learning_rate": 2.2839610178704722e-05,
      "loss": 0.5358,
      "step": 2579
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5763072332268873,
      "learning_rate": 2.2818041588618638e-05,
      "loss": 0.503,
      "step": 2580
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.8137711929624794,
      "learning_rate": 2.2796476935814786e-05,
      "loss": 0.5951,
      "step": 2581
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4187578078213683,
      "learning_rate": 2.2774916232115313e-05,
      "loss": 0.5161,
      "step": 2582
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.8211408597505352,
      "learning_rate": 2.2753359489340203e-05,
      "loss": 0.5707,
      "step": 2583
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5443226133416879,
      "learning_rate": 2.2731806719307248e-05,
      "loss": 0.5081,
      "step": 2584
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.65916081970866,
      "learning_rate": 2.2710257933832097e-05,
      "loss": 0.506,
      "step": 2585
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5498804214074637,
      "learning_rate": 2.2688713144728167e-05,
      "loss": 0.5197,
      "step": 2586
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4971407613355252,
      "learning_rate": 2.266717236380672e-05,
      "loss": 0.4967,
      "step": 2587
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4450040264298876,
      "learning_rate": 2.2645635602876803e-05,
      "loss": 0.5175,
      "step": 2588
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.455034634001717,
      "learning_rate": 2.2624102873745288e-05,
      "loss": 0.5005,
      "step": 2589
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.606403492327346,
      "learning_rate": 2.260257418821679e-05,
      "loss": 0.5424,
      "step": 2590
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.600496585118297,
      "learning_rate": 2.2581049558093736e-05,
      "loss": 0.5362,
      "step": 2591
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.7292369394685212,
      "learning_rate": 2.2559528995176326e-05,
      "loss": 0.512,
      "step": 2592
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.8732139843176647,
      "learning_rate": 2.2538012511262554e-05,
      "loss": 0.6214,
      "step": 2593
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5830573663301715,
      "learning_rate": 2.2516500118148127e-05,
      "loss": 0.5145,
      "step": 2594
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.8363465021533223,
      "learning_rate": 2.2494991827626534e-05,
      "loss": 0.5975,
      "step": 2595
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.807231343223661,
      "learning_rate": 2.2473487651489034e-05,
      "loss": 0.5624,
      "step": 2596
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6781632864269258,
      "learning_rate": 2.2451987601524617e-05,
      "loss": 0.5831,
      "step": 2597
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5734781894207117,
      "learning_rate": 2.2430491689519988e-05,
      "loss": 0.5307,
      "step": 2598
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.7404856435011737,
      "learning_rate": 2.24089999272596e-05,
      "loss": 0.5955,
      "step": 2599
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6483536276577295,
      "learning_rate": 2.2387512326525656e-05,
      "loss": 0.5666,
      "step": 2600
    },
    {
      "epoch": 2.97,
      "eval_blimp_filtered_avg": 0.7083582089552238,
      "eval_blimp_filtered_std": 0.005038850126596099,
      "step": 2600
    },
    {
      "epoch": 2.97,
      "eval_blimp_supplement_avg": 0.8232758620689655,
      "eval_blimp_supplement_std": 0.017145751255536926,
      "step": 2600
    },
    {
      "epoch": 2.97,
      "eval_vqa_filtered_avg": 0.53,
      "eval_vqa_filtered_std": 0.0501613558046592,
      "step": 2600
    },
    {
      "epoch": 2.97,
      "eval_winoground_filtered_avg": 0.63,
      "eval_winoground_filtered_std": 0.04852365870939099,
      "step": 2600
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.357131424872802,
      "learning_rate": 2.2366028899098042e-05,
      "loss": 0.4412,
      "step": 2601
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.7926587341688218,
      "learning_rate": 2.234454965675439e-05,
      "loss": 0.6014,
      "step": 2602
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6661874495563895,
      "learning_rate": 2.2323074611269982e-05,
      "loss": 0.5753,
      "step": 2603
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6647129990554523,
      "learning_rate": 2.2301603774417858e-05,
      "loss": 0.5668,
      "step": 2604
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.7686665297552053,
      "learning_rate": 2.2280137157968722e-05,
      "loss": 0.6025,
      "step": 2605
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6146654810774528,
      "learning_rate": 2.2258674773690973e-05,
      "loss": 0.5049,
      "step": 2606
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.745922854316909,
      "learning_rate": 2.223721663335068e-05,
      "loss": 0.5485,
      "step": 2607
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6308418414914574,
      "learning_rate": 2.2215762748711597e-05,
      "loss": 0.5901,
      "step": 2608
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6485073120552196,
      "learning_rate": 2.2194313131535134e-05,
      "loss": 0.5176,
      "step": 2609
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5388179104612965,
      "learning_rate": 2.217286779358038e-05,
      "loss": 0.533,
      "step": 2610
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4858393301946013,
      "learning_rate": 2.2151426746604046e-05,
      "loss": 0.5038,
      "step": 2611
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.803762164582456,
      "learning_rate": 2.2129990002360524e-05,
      "loss": 0.593,
      "step": 2612
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.584164409678872,
      "learning_rate": 2.2108557572601834e-05,
      "loss": 0.5191,
      "step": 2613
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.7538749504617561,
      "learning_rate": 2.2087129469077624e-05,
      "loss": 0.5147,
      "step": 2614
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.7049431198685459,
      "learning_rate": 2.2065705703535176e-05,
      "loss": 0.5462,
      "step": 2615
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.734883227071255,
      "learning_rate": 2.2044286287719396e-05,
      "loss": 0.4893,
      "step": 2616
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.895017379129858,
      "learning_rate": 2.20228712333728e-05,
      "loss": 0.5536,
      "step": 2617
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.615669260425703,
      "learning_rate": 2.200146055223552e-05,
      "loss": 0.5015,
      "step": 2618
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.72062890339604,
      "learning_rate": 2.1980054256045294e-05,
      "loss": 0.5528,
      "step": 2619
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.499645997137307,
      "learning_rate": 2.195865235653743e-05,
      "loss": 0.4237,
      "step": 2620
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.7896813919162782,
      "learning_rate": 2.1937254865444857e-05,
      "loss": 0.5663,
      "step": 2621
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6849576109695044,
      "learning_rate": 2.1915861794498075e-05,
      "loss": 0.5851,
      "step": 2622
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5785835153527468,
      "learning_rate": 2.1894473155425167e-05,
      "loss": 0.4721,
      "step": 2623
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.598859473270781,
      "learning_rate": 2.1873088959951763e-05,
      "loss": 0.542,
      "step": 2624
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3416425988665588,
      "learning_rate": 2.185170921980108e-05,
      "loss": 0.4416,
      "step": 2625
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4112844914783742,
      "learning_rate": 2.1830333946693892e-05,
      "loss": 0.4631,
      "step": 2626
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3046042000874916,
      "learning_rate": 2.180896315234852e-05,
      "loss": 0.4017,
      "step": 2627
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3155352234569533,
      "learning_rate": 2.1787596848480818e-05,
      "loss": 0.3509,
      "step": 2628
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.403812242053255,
      "learning_rate": 2.1766235046804195e-05,
      "loss": 0.3835,
      "step": 2629
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.5072021880729263,
      "learning_rate": 2.1744877759029585e-05,
      "loss": 0.3478,
      "step": 2630
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.4349932005499952,
      "learning_rate": 2.172352499686545e-05,
      "loss": 0.369,
      "step": 2631
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.361904308310425,
      "learning_rate": 2.170217677201776e-05,
      "loss": 0.3569,
      "step": 2632
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.7590303706754287,
      "learning_rate": 2.168083309619001e-05,
      "loss": 0.3457,
      "step": 2633
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.7609189105198475,
      "learning_rate": 2.1659493981083205e-05,
      "loss": 0.3863,
      "step": 2634
    },
    {
      "epoch": 3.01,
      "grad_norm": 2.287356914815681,
      "learning_rate": 2.1638159438395834e-05,
      "loss": 0.4067,
      "step": 2635
    },
    {
      "epoch": 3.01,
      "grad_norm": 2.070021945308727,
      "learning_rate": 2.1616829479823892e-05,
      "loss": 0.3585,
      "step": 2636
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.954274431148398,
      "learning_rate": 2.1595504117060846e-05,
      "loss": 0.3501,
      "step": 2637
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.750895265074918,
      "learning_rate": 2.1574183361797656e-05,
      "loss": 0.3625,
      "step": 2638
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.6667826865693518,
      "learning_rate": 2.1552867225722766e-05,
      "loss": 0.3462,
      "step": 2639
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.6927607728081344,
      "learning_rate": 2.153155572052208e-05,
      "loss": 0.3412,
      "step": 2640
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.89617522307,
      "learning_rate": 2.151024885787894e-05,
      "loss": 0.3745,
      "step": 2641
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.5265331129452635,
      "learning_rate": 2.1488946649474153e-05,
      "loss": 0.2995,
      "step": 2642
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.4499918873805218,
      "learning_rate": 2.1467649106986013e-05,
      "loss": 0.3626,
      "step": 2643
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.6208973143205931,
      "learning_rate": 2.144635624209022e-05,
      "loss": 0.316,
      "step": 2644
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.6227396787389867,
      "learning_rate": 2.1425068066459897e-05,
      "loss": 0.3464,
      "step": 2645
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.2930433217959392,
      "learning_rate": 2.140378459176562e-05,
      "loss": 0.2952,
      "step": 2646
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.7140960677974586,
      "learning_rate": 2.138250582967539e-05,
      "loss": 0.4016,
      "step": 2647
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.5877537430434894,
      "learning_rate": 2.1361231791854635e-05,
      "loss": 0.367,
      "step": 2648
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.4878984723998965,
      "learning_rate": 2.133996248996614e-05,
      "loss": 0.3287,
      "step": 2649
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.428256030468463,
      "learning_rate": 2.1318697935670142e-05,
      "loss": 0.325,
      "step": 2650
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.6275220890702329,
      "learning_rate": 2.129743814062427e-05,
      "loss": 0.3741,
      "step": 2651
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.3972614716558973,
      "learning_rate": 2.1276183116483546e-05,
      "loss": 0.3221,
      "step": 2652
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.407038825949832,
      "learning_rate": 2.1254932874900327e-05,
      "loss": 0.332,
      "step": 2653
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.5538780552063625,
      "learning_rate": 2.123368742752443e-05,
      "loss": 0.3778,
      "step": 2654
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.7820852020828666,
      "learning_rate": 2.121244678600298e-05,
      "loss": 0.362,
      "step": 2655
    },
    {
      "epoch": 3.04,
      "grad_norm": 1.7868747137410532,
      "learning_rate": 2.1191210961980497e-05,
      "loss": 0.3789,
      "step": 2656
    },
    {
      "epoch": 3.04,
      "grad_norm": 1.6883367047171058,
      "learning_rate": 2.1169979967098853e-05,
      "loss": 0.3324,
      "step": 2657
    },
    {
      "epoch": 3.04,
      "grad_norm": 1.5144726500779786,
      "learning_rate": 2.1148753812997264e-05,
      "loss": 0.3262,
      "step": 2658
    },
    {
      "epoch": 3.04,
      "grad_norm": 1.766095313970161,
      "learning_rate": 2.11275325113123e-05,
      "loss": 0.3488,
      "step": 2659
    },
    {
      "epoch": 3.04,
      "grad_norm": 1.7419767326501874,
      "learning_rate": 2.110631607367788e-05,
      "loss": 0.404,
      "step": 2660
    },
    {
      "epoch": 3.04,
      "grad_norm": 1.885532535213587,
      "learning_rate": 2.1085104511725245e-05,
      "loss": 0.3797,
      "step": 2661
    },
    {
      "epoch": 3.04,
      "grad_norm": 1.9548518839371425,
      "learning_rate": 2.1063897837082948e-05,
      "loss": 0.4424,
      "step": 2662
    },
    {
      "epoch": 3.04,
      "grad_norm": 1.755467776794688,
      "learning_rate": 2.10426960613769e-05,
      "loss": 0.3913,
      "step": 2663
    },
    {
      "epoch": 3.04,
      "grad_norm": 1.7693928091828273,
      "learning_rate": 2.1021499196230296e-05,
      "loss": 0.3506,
      "step": 2664
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.7656774692210475,
      "learning_rate": 2.1000307253263656e-05,
      "loss": 0.4354,
      "step": 2665
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.578195240241289,
      "learning_rate": 2.097912024409479e-05,
      "loss": 0.334,
      "step": 2666
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.8279863965405545,
      "learning_rate": 2.0957938180338805e-05,
      "loss": 0.353,
      "step": 2667
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.7726169054175807,
      "learning_rate": 2.09367610736081e-05,
      "loss": 0.3902,
      "step": 2668
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.7226823502608852,
      "learning_rate": 2.0915588935512366e-05,
      "loss": 0.3789,
      "step": 2669
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.7652104339280297,
      "learning_rate": 2.0894421777658545e-05,
      "loss": 0.3573,
      "step": 2670
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.7481506864756966,
      "learning_rate": 2.0873259611650877e-05,
      "loss": 0.3897,
      "step": 2671
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.7784514156744655,
      "learning_rate": 2.085210244909084e-05,
      "loss": 0.4073,
      "step": 2672
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.4361804269589065,
      "learning_rate": 2.0830950301577195e-05,
      "loss": 0.3243,
      "step": 2673
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.5803221732052093,
      "learning_rate": 2.080980318070593e-05,
      "loss": 0.3204,
      "step": 2674
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.5780854372894364,
      "learning_rate": 2.078866109807029e-05,
      "loss": 0.3123,
      "step": 2675
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.6453218059354193,
      "learning_rate": 2.076752406526075e-05,
      "loss": 0.3221,
      "step": 2676
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.6997696120686863,
      "learning_rate": 2.0746392093865033e-05,
      "loss": 0.3735,
      "step": 2677
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.6689112413244198,
      "learning_rate": 2.0725265195468076e-05,
      "loss": 0.3659,
      "step": 2678
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.8006695095148453,
      "learning_rate": 2.0704143381652017e-05,
      "loss": 0.379,
      "step": 2679
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.6759238367013933,
      "learning_rate": 2.0683026663996242e-05,
      "loss": 0.3323,
      "step": 2680
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.7988865464491937,
      "learning_rate": 2.0661915054077318e-05,
      "loss": 0.3885,
      "step": 2681
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.530675468607564,
      "learning_rate": 2.0640808563469026e-05,
      "loss": 0.3152,
      "step": 2682
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.3616069132308932,
      "learning_rate": 2.061970720374232e-05,
      "loss": 0.2893,
      "step": 2683
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.7835555134391277,
      "learning_rate": 2.0598610986465372e-05,
      "loss": 0.3707,
      "step": 2684
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.6220486327817834,
      "learning_rate": 2.05775199232035e-05,
      "loss": 0.3169,
      "step": 2685
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.682830399297447,
      "learning_rate": 2.0556434025519244e-05,
      "loss": 0.3215,
      "step": 2686
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.7022182621211923,
      "learning_rate": 2.053535330497225e-05,
      "loss": 0.3313,
      "step": 2687
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.7226412055065694,
      "learning_rate": 2.0514277773119373e-05,
      "loss": 0.3942,
      "step": 2688
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.869212159461291,
      "learning_rate": 2.04932074415146e-05,
      "loss": 0.3651,
      "step": 2689
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.6241006010883736,
      "learning_rate": 2.0472142321709088e-05,
      "loss": 0.3517,
      "step": 2690
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.6749644328117637,
      "learning_rate": 2.045108242525114e-05,
      "loss": 0.3302,
      "step": 2691
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.804799365933856,
      "learning_rate": 2.0430027763686138e-05,
      "loss": 0.3883,
      "step": 2692
    },
    {
      "epoch": 3.08,
      "grad_norm": 2.713806186190843,
      "learning_rate": 2.0408978348556654e-05,
      "loss": 0.391,
      "step": 2693
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.683967300847189,
      "learning_rate": 2.0387934191402383e-05,
      "loss": 0.3227,
      "step": 2694
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.680691883731434,
      "learning_rate": 2.0366895303760112e-05,
      "loss": 0.3722,
      "step": 2695
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.5512992816188689,
      "learning_rate": 2.0345861697163733e-05,
      "loss": 0.3297,
      "step": 2696
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.6478906189654,
      "learning_rate": 2.032483338314427e-05,
      "loss": 0.355,
      "step": 2697
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.4479108829612006,
      "learning_rate": 2.0303810373229833e-05,
      "loss": 0.3191,
      "step": 2698
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.96446354548414,
      "learning_rate": 2.028279267894564e-05,
      "loss": 0.4273,
      "step": 2699
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.6754968446687535,
      "learning_rate": 2.0261780311813938e-05,
      "loss": 0.3706,
      "step": 2700
    },
    {
      "epoch": 3.09,
      "eval_blimp_filtered_avg": 0.7071641791044776,
      "eval_blimp_filtered_std": 0.005053476200204506,
      "step": 2700
    },
    {
      "epoch": 3.09,
      "eval_blimp_supplement_avg": 0.8275862068965517,
      "eval_blimp_supplement_std": 0.016969582176787734,
      "step": 2700
    },
    {
      "epoch": 3.09,
      "eval_vqa_filtered_avg": 0.54,
      "eval_vqa_filtered_std": 0.05009082659620333,
      "step": 2700
    },
    {
      "epoch": 3.09,
      "eval_winoground_filtered_avg": 0.62,
      "eval_winoground_filtered_std": 0.04878317312145633,
      "step": 2700
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.8572749078947344,
      "learning_rate": 2.0240773283354128e-05,
      "loss": 0.3907,
      "step": 2701
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.8048772078846242,
      "learning_rate": 2.0219771605082638e-05,
      "loss": 0.3493,
      "step": 2702
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.6625584324444274,
      "learning_rate": 2.0198775288512993e-05,
      "loss": 0.3647,
      "step": 2703
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.704295156706242,
      "learning_rate": 2.0177784345155733e-05,
      "loss": 0.3434,
      "step": 2704
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.7763638376770787,
      "learning_rate": 2.01567987865185e-05,
      "loss": 0.3536,
      "step": 2705
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.4900234930681169,
      "learning_rate": 2.013581862410596e-05,
      "loss": 0.3327,
      "step": 2706
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.8325520731613423,
      "learning_rate": 2.0114843869419825e-05,
      "loss": 0.3331,
      "step": 2707
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.6079758140434934,
      "learning_rate": 2.0093874533958856e-05,
      "loss": 0.3639,
      "step": 2708
    },
    {
      "epoch": 3.1,
      "grad_norm": 2.005318574950965,
      "learning_rate": 2.007291062921881e-05,
      "loss": 0.4313,
      "step": 2709
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.680905796661775,
      "learning_rate": 2.0051952166692495e-05,
      "loss": 0.3443,
      "step": 2710
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.6178878001733519,
      "learning_rate": 2.003099915786973e-05,
      "loss": 0.3443,
      "step": 2711
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.6449534699435964,
      "learning_rate": 2.0010051614237335e-05,
      "loss": 0.3453,
      "step": 2712
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.705578905082511,
      "learning_rate": 1.9989109547279144e-05,
      "loss": 0.3299,
      "step": 2713
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.2262162009324051,
      "learning_rate": 1.9968172968475987e-05,
      "loss": 0.2884,
      "step": 2714
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.5235572174688607,
      "learning_rate": 1.9947241889305678e-05,
      "loss": 0.3479,
      "step": 2715
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.5091412476483421,
      "learning_rate": 1.992631632124303e-05,
      "loss": 0.3023,
      "step": 2716
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.7206807689414096,
      "learning_rate": 1.9905396275759816e-05,
      "loss": 0.3887,
      "step": 2717
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.4781583422030184,
      "learning_rate": 1.9884481764324795e-05,
      "loss": 0.3063,
      "step": 2718
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.7721608036765988,
      "learning_rate": 1.9863572798403685e-05,
      "loss": 0.3644,
      "step": 2719
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.8505180481064998,
      "learning_rate": 1.9842669389459184e-05,
      "loss": 0.3485,
      "step": 2720
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.6225929287545218,
      "learning_rate": 1.9821771548950904e-05,
      "loss": 0.4065,
      "step": 2721
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.65635868983549,
      "learning_rate": 1.9800879288335434e-05,
      "loss": 0.3413,
      "step": 2722
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.9562161662496131,
      "learning_rate": 1.9779992619066303e-05,
      "loss": 0.4326,
      "step": 2723
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.8244555963746283,
      "learning_rate": 1.9759111552593966e-05,
      "loss": 0.3685,
      "step": 2724
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.4214586457753526,
      "learning_rate": 1.9738236100365817e-05,
      "loss": 0.3183,
      "step": 2725
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.5871768448369212,
      "learning_rate": 1.971736627382615e-05,
      "loss": 0.3137,
      "step": 2726
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.740052525134841,
      "learning_rate": 1.9696502084416194e-05,
      "loss": 0.3301,
      "step": 2727
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.6454042019208077,
      "learning_rate": 1.9675643543574083e-05,
      "loss": 0.3347,
      "step": 2728
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.6693869245887718,
      "learning_rate": 1.9654790662734864e-05,
      "loss": 0.3396,
      "step": 2729
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.5185349938241304,
      "learning_rate": 1.9633943453330464e-05,
      "loss": 0.302,
      "step": 2730
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.7903591911591805,
      "learning_rate": 1.96131019267897e-05,
      "loss": 0.3523,
      "step": 2731
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.8985491744435927,
      "learning_rate": 1.9592266094538285e-05,
      "loss": 0.3512,
      "step": 2732
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.8298734097204397,
      "learning_rate": 1.9571435967998833e-05,
      "loss": 0.3698,
      "step": 2733
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.642300193165578,
      "learning_rate": 1.9550611558590767e-05,
      "loss": 0.3598,
      "step": 2734
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.9156473514067873,
      "learning_rate": 1.952979287773043e-05,
      "loss": 0.4259,
      "step": 2735
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.9262991301403767,
      "learning_rate": 1.9508979936830998e-05,
      "loss": 0.3184,
      "step": 2736
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.3551224910594966,
      "learning_rate": 1.948817274730254e-05,
      "loss": 0.2928,
      "step": 2737
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.4767107958855294,
      "learning_rate": 1.9467371320551904e-05,
      "loss": 0.3007,
      "step": 2738
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.7425887704320129,
      "learning_rate": 1.9446575667982833e-05,
      "loss": 0.3801,
      "step": 2739
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.5756327390752816,
      "learning_rate": 1.942578580099588e-05,
      "loss": 0.3338,
      "step": 2740
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.7004637756964813,
      "learning_rate": 1.9405001730988455e-05,
      "loss": 0.3716,
      "step": 2741
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.510498061767137,
      "learning_rate": 1.9384223469354745e-05,
      "loss": 0.3391,
      "step": 2742
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.6741544119103722,
      "learning_rate": 1.936345102748578e-05,
      "loss": 0.354,
      "step": 2743
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.5083984908309647,
      "learning_rate": 1.934268441676941e-05,
      "loss": 0.3022,
      "step": 2744
    },
    {
      "epoch": 3.14,
      "grad_norm": 2.20864737252675,
      "learning_rate": 1.9321923648590263e-05,
      "loss": 0.4096,
      "step": 2745
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.8362438840785034,
      "learning_rate": 1.9301168734329796e-05,
      "loss": 0.358,
      "step": 2746
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.967977162646533,
      "learning_rate": 1.9280419685366194e-05,
      "loss": 0.4192,
      "step": 2747
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.752960301401521,
      "learning_rate": 1.9259676513074506e-05,
      "loss": 0.3831,
      "step": 2748
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.7039486164318827,
      "learning_rate": 1.9238939228826513e-05,
      "loss": 0.338,
      "step": 2749
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.892120274787313,
      "learning_rate": 1.9218207843990782e-05,
      "loss": 0.3843,
      "step": 2750
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.515728558480076,
      "learning_rate": 1.9197482369932615e-05,
      "loss": 0.2904,
      "step": 2751
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.6891622030623774,
      "learning_rate": 1.917676281801413e-05,
      "loss": 0.3779,
      "step": 2752
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.7366704023736232,
      "learning_rate": 1.9156049199594154e-05,
      "loss": 0.3367,
      "step": 2753
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.6219529259762722,
      "learning_rate": 1.9135341526028282e-05,
      "loss": 0.3181,
      "step": 2754
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.794882824029016,
      "learning_rate": 1.9114639808668834e-05,
      "loss": 0.3533,
      "step": 2755
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.6304321130979411,
      "learning_rate": 1.909394405886488e-05,
      "loss": 0.3293,
      "step": 2756
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.9686275647728504,
      "learning_rate": 1.9073254287962213e-05,
      "loss": 0.3781,
      "step": 2757
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.797995195381755,
      "learning_rate": 1.9052570507303346e-05,
      "loss": 0.3485,
      "step": 2758
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.8464930515407463,
      "learning_rate": 1.9031892728227504e-05,
      "loss": 0.3491,
      "step": 2759
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.998673951930414,
      "learning_rate": 1.9011220962070636e-05,
      "loss": 0.4185,
      "step": 2760
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.647886038840517,
      "learning_rate": 1.899055522016538e-05,
      "loss": 0.3406,
      "step": 2761
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.7469167052699113,
      "learning_rate": 1.8969895513841078e-05,
      "loss": 0.3426,
      "step": 2762
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.6298025039874395,
      "learning_rate": 1.8949241854423774e-05,
      "loss": 0.3631,
      "step": 2763
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.5387566346745785,
      "learning_rate": 1.8928594253236167e-05,
      "loss": 0.3072,
      "step": 2764
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.6652335105524154,
      "learning_rate": 1.8907952721597662e-05,
      "loss": 0.3285,
      "step": 2765
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.9622255863596874,
      "learning_rate": 1.888731727082433e-05,
      "loss": 0.3699,
      "step": 2766
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.6133286017663837,
      "learning_rate": 1.886668791222891e-05,
      "loss": 0.3536,
      "step": 2767
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.6189627292132054,
      "learning_rate": 1.884606465712078e-05,
      "loss": 0.3426,
      "step": 2768
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.5180790830588289,
      "learning_rate": 1.8825447516806005e-05,
      "loss": 0.3355,
      "step": 2769
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.5159006824059091,
      "learning_rate": 1.880483650258728e-05,
      "loss": 0.3344,
      "step": 2770
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.7369545369325752,
      "learning_rate": 1.8784231625763947e-05,
      "loss": 0.3767,
      "step": 2771
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.6925091462179485,
      "learning_rate": 1.8763632897631963e-05,
      "loss": 0.3849,
      "step": 2772
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.4746729581782758,
      "learning_rate": 1.8743040329483947e-05,
      "loss": 0.2861,
      "step": 2773
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.6235143457355996,
      "learning_rate": 1.8722453932609123e-05,
      "loss": 0.3839,
      "step": 2774
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.6203474301465195,
      "learning_rate": 1.8701873718293335e-05,
      "loss": 0.3475,
      "step": 2775
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.5434739161831863,
      "learning_rate": 1.868129969781903e-05,
      "loss": 0.3309,
      "step": 2776
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.7086895659122565,
      "learning_rate": 1.866073188246527e-05,
      "loss": 0.3421,
      "step": 2777
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.7505042390740861,
      "learning_rate": 1.8640170283507707e-05,
      "loss": 0.3459,
      "step": 2778
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.734090255889686,
      "learning_rate": 1.8619614912218586e-05,
      "loss": 0.3693,
      "step": 2779
    },
    {
      "epoch": 3.18,
      "grad_norm": 2.0217194407931895,
      "learning_rate": 1.8599065779866772e-05,
      "loss": 0.3885,
      "step": 2780
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.912105018080852,
      "learning_rate": 1.857852289771764e-05,
      "loss": 0.3912,
      "step": 2781
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.7067593130749932,
      "learning_rate": 1.8557986277033194e-05,
      "loss": 0.3617,
      "step": 2782
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.8156488256279486,
      "learning_rate": 1.8537455929071978e-05,
      "loss": 0.3359,
      "step": 2783
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.9163468123915892,
      "learning_rate": 1.8516931865089135e-05,
      "loss": 0.3508,
      "step": 2784
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.9372602451523557,
      "learning_rate": 1.849641409633631e-05,
      "loss": 0.3827,
      "step": 2785
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.744013465943843,
      "learning_rate": 1.8475902634061725e-05,
      "loss": 0.3405,
      "step": 2786
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.5882459072916195,
      "learning_rate": 1.8455397489510147e-05,
      "loss": 0.3523,
      "step": 2787
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.8072048498894613,
      "learning_rate": 1.8434898673922892e-05,
      "loss": 0.3828,
      "step": 2788
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.8029869683430813,
      "learning_rate": 1.8414406198537764e-05,
      "loss": 0.3945,
      "step": 2789
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.6253126025399085,
      "learning_rate": 1.8393920074589116e-05,
      "loss": 0.3199,
      "step": 2790
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.728512560228188,
      "learning_rate": 1.8373440313307836e-05,
      "loss": 0.3623,
      "step": 2791
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.5429357046356371,
      "learning_rate": 1.8352966925921316e-05,
      "loss": 0.3259,
      "step": 2792
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.5886421655919192,
      "learning_rate": 1.8332499923653418e-05,
      "loss": 0.3347,
      "step": 2793
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.7631259064636553,
      "learning_rate": 1.8312039317724535e-05,
      "loss": 0.3639,
      "step": 2794
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.5601328864581159,
      "learning_rate": 1.8291585119351562e-05,
      "loss": 0.2856,
      "step": 2795
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.7283926512614753,
      "learning_rate": 1.827113733974786e-05,
      "loss": 0.3484,
      "step": 2796
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.5685569311415188,
      "learning_rate": 1.8250695990123298e-05,
      "loss": 0.3326,
      "step": 2797
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.9360501876758043,
      "learning_rate": 1.823026108168417e-05,
      "loss": 0.3913,
      "step": 2798
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.6518436056974197,
      "learning_rate": 1.8209832625633287e-05,
      "loss": 0.3334,
      "step": 2799
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.6101614265209427,
      "learning_rate": 1.8189410633169908e-05,
      "loss": 0.3103,
      "step": 2800
    },
    {
      "epoch": 3.2,
      "eval_blimp_filtered_avg": 0.7076119402985075,
      "eval_blimp_filtered_std": 0.005048910103662564,
      "step": 2800
    },
    {
      "epoch": 3.2,
      "eval_blimp_supplement_avg": 0.8297413793103449,
      "eval_blimp_supplement_std": 0.016871070217749664,
      "step": 2800
    },
    {
      "epoch": 3.2,
      "eval_vqa_filtered_avg": 0.54,
      "eval_vqa_filtered_std": 0.05009082659620333,
      "step": 2800
    },
    {
      "epoch": 3.2,
      "eval_winoground_filtered_avg": 0.63,
      "eval_winoground_filtered_std": 0.04852365870939099,
      "step": 2800
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.846839749247897,
      "learning_rate": 1.816899511548975e-05,
      "loss": 0.3604,
      "step": 2801
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.9713530327042428,
      "learning_rate": 1.8148586083784958e-05,
      "loss": 0.4307,
      "step": 2802
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.6228788560551952,
      "learning_rate": 1.8128183549244156e-05,
      "loss": 0.3364,
      "step": 2803
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.4797086216781092,
      "learning_rate": 1.8107787523052387e-05,
      "loss": 0.3127,
      "step": 2804
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.8072729377916872,
      "learning_rate": 1.808739801639113e-05,
      "loss": 0.3849,
      "step": 2805
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.5744540972539967,
      "learning_rate": 1.806701504043829e-05,
      "loss": 0.3226,
      "step": 2806
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.7190326488175334,
      "learning_rate": 1.8046638606368185e-05,
      "loss": 0.3852,
      "step": 2807
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.7368314932419142,
      "learning_rate": 1.802626872535155e-05,
      "loss": 0.352,
      "step": 2808
    },
    {
      "epoch": 3.21,
      "grad_norm": 2.027756241816126,
      "learning_rate": 1.800590540855555e-05,
      "loss": 0.4047,
      "step": 2809
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.5772283368583697,
      "learning_rate": 1.798554866714371e-05,
      "loss": 0.3226,
      "step": 2810
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.7078127474269367,
      "learning_rate": 1.796519851227598e-05,
      "loss": 0.3443,
      "step": 2811
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.4800914859604148,
      "learning_rate": 1.7944854955108685e-05,
      "loss": 0.2952,
      "step": 2812
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.610443493807102,
      "learning_rate": 1.7924518006794543e-05,
      "loss": 0.3305,
      "step": 2813
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.7980340282862703,
      "learning_rate": 1.7904187678482655e-05,
      "loss": 0.3155,
      "step": 2814
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.9193797862167743,
      "learning_rate": 1.7883863981318463e-05,
      "loss": 0.3247,
      "step": 2815
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.7224618682699597,
      "learning_rate": 1.7863546926443806e-05,
      "loss": 0.3417,
      "step": 2816
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.7224297261418282,
      "learning_rate": 1.7843236524996863e-05,
      "loss": 0.3577,
      "step": 2817
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.7649870536799317,
      "learning_rate": 1.782293278811218e-05,
      "loss": 0.3667,
      "step": 2818
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.9146476356885729,
      "learning_rate": 1.780263572692063e-05,
      "loss": 0.3958,
      "step": 2819
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.770374667269612,
      "learning_rate": 1.778234535254944e-05,
      "loss": 0.3917,
      "step": 2820
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.6596909889376479,
      "learning_rate": 1.776206167612217e-05,
      "loss": 0.3669,
      "step": 2821
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.7922388662304016,
      "learning_rate": 1.774178470875871e-05,
      "loss": 0.3401,
      "step": 2822
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.7756944863909183,
      "learning_rate": 1.7721514461575254e-05,
      "loss": 0.3566,
      "step": 2823
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.8359171470661515,
      "learning_rate": 1.770125094568434e-05,
      "loss": 0.4445,
      "step": 2824
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.2342330994439954,
      "learning_rate": 1.7680994172194794e-05,
      "loss": 0.2732,
      "step": 2825
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.6680456833351873,
      "learning_rate": 1.766074415221176e-05,
      "loss": 0.333,
      "step": 2826
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.6563918466978282,
      "learning_rate": 1.7640500896836663e-05,
      "loss": 0.3268,
      "step": 2827
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.8200713858684565,
      "learning_rate": 1.7620264417167236e-05,
      "loss": 0.3566,
      "step": 2828
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.7466814820835563,
      "learning_rate": 1.760003472429749e-05,
      "loss": 0.3414,
      "step": 2829
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.8395970659737075,
      "learning_rate": 1.7579811829317725e-05,
      "loss": 0.3329,
      "step": 2830
    },
    {
      "epoch": 3.24,
      "grad_norm": 2.0181025246842896,
      "learning_rate": 1.7559595743314485e-05,
      "loss": 0.3609,
      "step": 2831
    },
    {
      "epoch": 3.24,
      "grad_norm": 2.0282852093443173,
      "learning_rate": 1.753938647737062e-05,
      "loss": 0.3544,
      "step": 2832
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.9356418811635834,
      "learning_rate": 1.7519184042565204e-05,
      "loss": 0.3683,
      "step": 2833
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.690599424592348,
      "learning_rate": 1.7498988449973616e-05,
      "loss": 0.3318,
      "step": 2834
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.8308555712008865,
      "learning_rate": 1.7478799710667422e-05,
      "loss": 0.3385,
      "step": 2835
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.9289694475343488,
      "learning_rate": 1.745861783571448e-05,
      "loss": 0.3829,
      "step": 2836
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.5432042982693668,
      "learning_rate": 1.743844283617885e-05,
      "loss": 0.3077,
      "step": 2837
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.818312371873108,
      "learning_rate": 1.741827472312086e-05,
      "loss": 0.3566,
      "step": 2838
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.9663521976298275,
      "learning_rate": 1.7398113507597037e-05,
      "loss": 0.3745,
      "step": 2839
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.8468783125726533,
      "learning_rate": 1.737795920066012e-05,
      "loss": 0.3815,
      "step": 2840
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.7597927881566682,
      "learning_rate": 1.7357811813359076e-05,
      "loss": 0.3391,
      "step": 2841
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.7681781480430105,
      "learning_rate": 1.733767135673908e-05,
      "loss": 0.3547,
      "step": 2842
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.507220607854943,
      "learning_rate": 1.7317537841841495e-05,
      "loss": 0.3084,
      "step": 2843
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.7235977552438764,
      "learning_rate": 1.729741127970389e-05,
      "loss": 0.3477,
      "step": 2844
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.7311730252108701,
      "learning_rate": 1.727729168136002e-05,
      "loss": 0.3479,
      "step": 2845
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.3471514384543821,
      "learning_rate": 1.7257179057839825e-05,
      "loss": 0.3009,
      "step": 2846
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.7361304532070418,
      "learning_rate": 1.7237073420169417e-05,
      "loss": 0.3825,
      "step": 2847
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.9127954133610097,
      "learning_rate": 1.7216974779371058e-05,
      "loss": 0.3783,
      "step": 2848
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.6523710826349411,
      "learning_rate": 1.7196883146463213e-05,
      "loss": 0.3284,
      "step": 2849
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.6658028876866993,
      "learning_rate": 1.7176798532460475e-05,
      "loss": 0.3564,
      "step": 2850
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.5969757283575292,
      "learning_rate": 1.7156720948373598e-05,
      "loss": 0.2937,
      "step": 2851
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.7888333172166773,
      "learning_rate": 1.7136650405209515e-05,
      "loss": 0.3653,
      "step": 2852
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.8407979072693352,
      "learning_rate": 1.7116586913971237e-05,
      "loss": 0.355,
      "step": 2853
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.6789850567724438,
      "learning_rate": 1.7096530485657945e-05,
      "loss": 0.3249,
      "step": 2854
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.928489429827395,
      "learning_rate": 1.707648113126495e-05,
      "loss": 0.377,
      "step": 2855
    },
    {
      "epoch": 3.26,
      "grad_norm": 2.019676516637998,
      "learning_rate": 1.7056438861783686e-05,
      "loss": 0.4085,
      "step": 2856
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.9071953028013326,
      "learning_rate": 1.703640368820168e-05,
      "loss": 0.3692,
      "step": 2857
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.9141998880671587,
      "learning_rate": 1.7016375621502585e-05,
      "loss": 0.3421,
      "step": 2858
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.9025302113975384,
      "learning_rate": 1.699635467266615e-05,
      "loss": 0.339,
      "step": 2859
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.852810285183148,
      "learning_rate": 1.6976340852668262e-05,
      "loss": 0.3571,
      "step": 2860
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.5641459447295256,
      "learning_rate": 1.695633417248084e-05,
      "loss": 0.3257,
      "step": 2861
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.4481265531826815,
      "learning_rate": 1.6936334643071915e-05,
      "loss": 0.3227,
      "step": 2862
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.9645563585354051,
      "learning_rate": 1.6916342275405608e-05,
      "loss": 0.3541,
      "step": 2863
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.5870662406697462,
      "learning_rate": 1.6896357080442108e-05,
      "loss": 0.3185,
      "step": 2864
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.8082855987867876,
      "learning_rate": 1.6876379069137654e-05,
      "loss": 0.3653,
      "step": 2865
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.8777538259000188,
      "learning_rate": 1.6856408252444548e-05,
      "loss": 0.3839,
      "step": 2866
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.651662976315101,
      "learning_rate": 1.6836444641311196e-05,
      "loss": 0.3611,
      "step": 2867
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.8497276663948348,
      "learning_rate": 1.6816488246682006e-05,
      "loss": 0.4144,
      "step": 2868
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.6049176606324167,
      "learning_rate": 1.6796539079497445e-05,
      "loss": 0.3466,
      "step": 2869
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.7902060719136006,
      "learning_rate": 1.6776597150693998e-05,
      "loss": 0.3892,
      "step": 2870
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.7793788222749927,
      "learning_rate": 1.6756662471204206e-05,
      "loss": 0.3463,
      "step": 2871
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.760203190652717,
      "learning_rate": 1.6736735051956633e-05,
      "loss": 0.3399,
      "step": 2872
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.6850272097905614,
      "learning_rate": 1.6716814903875855e-05,
      "loss": 0.3651,
      "step": 2873
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.7706160828095343,
      "learning_rate": 1.669690203788247e-05,
      "loss": 0.3448,
      "step": 2874
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.619502692646987,
      "learning_rate": 1.6676996464893076e-05,
      "loss": 0.3077,
      "step": 2875
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.727607467950793,
      "learning_rate": 1.6657098195820273e-05,
      "loss": 0.3733,
      "step": 2876
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.7214508694777664,
      "learning_rate": 1.663720724157267e-05,
      "loss": 0.3058,
      "step": 2877
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.7881783731478662,
      "learning_rate": 1.661732361305484e-05,
      "loss": 0.3462,
      "step": 2878
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.7137558957218586,
      "learning_rate": 1.6597447321167365e-05,
      "loss": 0.378,
      "step": 2879
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.7848299071201086,
      "learning_rate": 1.657757837680679e-05,
      "loss": 0.3382,
      "step": 2880
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.7119345179940404,
      "learning_rate": 1.6557716790865636e-05,
      "loss": 0.3392,
      "step": 2881
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.6910575181791345,
      "learning_rate": 1.65378625742324e-05,
      "loss": 0.3501,
      "step": 2882
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.8671498949067138,
      "learning_rate": 1.6518015737791523e-05,
      "loss": 0.4002,
      "step": 2883
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.6174548847677663,
      "learning_rate": 1.6498176292423416e-05,
      "loss": 0.3294,
      "step": 2884
    },
    {
      "epoch": 3.3,
      "grad_norm": 2.0362226731299553,
      "learning_rate": 1.6478344249004415e-05,
      "loss": 0.3455,
      "step": 2885
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.5763105476350487,
      "learning_rate": 1.6458519618406836e-05,
      "loss": 0.3361,
      "step": 2886
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.6942091848610714,
      "learning_rate": 1.643870241149888e-05,
      "loss": 0.3494,
      "step": 2887
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.511444217573424,
      "learning_rate": 1.6418892639144717e-05,
      "loss": 0.3058,
      "step": 2888
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.8327412905343003,
      "learning_rate": 1.6399090312204435e-05,
      "loss": 0.3541,
      "step": 2889
    },
    {
      "epoch": 3.3,
      "grad_norm": 2.041300058419874,
      "learning_rate": 1.6379295441534034e-05,
      "loss": 0.3457,
      "step": 2890
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.7000095569867044,
      "learning_rate": 1.6359508037985416e-05,
      "loss": 0.3517,
      "step": 2891
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.8821112953054164,
      "learning_rate": 1.6339728112406417e-05,
      "loss": 0.3341,
      "step": 2892
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.4558260162165442,
      "learning_rate": 1.631995567564075e-05,
      "loss": 0.3505,
      "step": 2893
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.596772890430169,
      "learning_rate": 1.6300190738528043e-05,
      "loss": 0.3093,
      "step": 2894
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.4339015954411656,
      "learning_rate": 1.628043331190377e-05,
      "loss": 0.2882,
      "step": 2895
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.8485615735226422,
      "learning_rate": 1.626068340659934e-05,
      "loss": 0.3803,
      "step": 2896
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.8894958845802008,
      "learning_rate": 1.6240941033442007e-05,
      "loss": 0.3723,
      "step": 2897
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.8734038381776297,
      "learning_rate": 1.622120620325491e-05,
      "loss": 0.3327,
      "step": 2898
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.9165707125816056,
      "learning_rate": 1.6201478926857048e-05,
      "loss": 0.3617,
      "step": 2899
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.6976284933264005,
      "learning_rate": 1.6181759215063274e-05,
      "loss": 0.3153,
      "step": 2900
    },
    {
      "epoch": 3.31,
      "eval_blimp_filtered_avg": 0.7098507462686567,
      "eval_blimp_filtered_std": 0.005044785932262036,
      "step": 2900
    },
    {
      "epoch": 3.31,
      "eval_blimp_supplement_avg": 0.8383620689655172,
      "eval_blimp_supplement_std": 0.016549250449612328,
      "step": 2900
    },
    {
      "epoch": 3.31,
      "eval_vqa_filtered_avg": 0.53,
      "eval_vqa_filtered_std": 0.0501613558046592,
      "step": 2900
    },
    {
      "epoch": 3.31,
      "eval_winoground_filtered_avg": 0.63,
      "eval_winoground_filtered_std": 0.04852365870939099,
      "step": 2900
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.855472796069688,
      "learning_rate": 1.6162047078684296e-05,
      "loss": 0.3858,
      "step": 2901
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.671798101109258,
      "learning_rate": 1.6142342528526676e-05,
      "loss": 0.3248,
      "step": 2902
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.990494428229662,
      "learning_rate": 1.6122645575392825e-05,
      "loss": 0.4294,
      "step": 2903
    },
    {
      "epoch": 3.32,
      "grad_norm": 2.0234760781962615,
      "learning_rate": 1.6102956230080957e-05,
      "loss": 0.4252,
      "step": 2904
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.4830476669793302,
      "learning_rate": 1.608327450338513e-05,
      "loss": 0.2899,
      "step": 2905
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.921391971595966,
      "learning_rate": 1.6063600406095235e-05,
      "loss": 0.3733,
      "step": 2906
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.7608103358574814,
      "learning_rate": 1.6043933948997002e-05,
      "loss": 0.3542,
      "step": 2907
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.8257739089615002,
      "learning_rate": 1.6024275142871914e-05,
      "loss": 0.3294,
      "step": 2908
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.6388662360933361,
      "learning_rate": 1.6004623998497304e-05,
      "loss": 0.3115,
      "step": 2909
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.8432968772081824,
      "learning_rate": 1.5984980526646288e-05,
      "loss": 0.364,
      "step": 2910
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.6963516284634215,
      "learning_rate": 1.5965344738087798e-05,
      "loss": 0.3312,
      "step": 2911
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.9701989873445196,
      "learning_rate": 1.5945716643586505e-05,
      "loss": 0.3971,
      "step": 2912
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.5862576825185524,
      "learning_rate": 1.59260962539029e-05,
      "loss": 0.344,
      "step": 2913
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.6448582274930588,
      "learning_rate": 1.5906483579793264e-05,
      "loss": 0.2993,
      "step": 2914
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.94482167702806,
      "learning_rate": 1.5886878632009624e-05,
      "loss": 0.3499,
      "step": 2915
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.9030441027680192,
      "learning_rate": 1.5867281421299756e-05,
      "loss": 0.3843,
      "step": 2916
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.6572790500126,
      "learning_rate": 1.5847691958407222e-05,
      "loss": 0.329,
      "step": 2917
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.6778813263215042,
      "learning_rate": 1.582811025407133e-05,
      "loss": 0.3224,
      "step": 2918
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.958017173048513,
      "learning_rate": 1.5808536319027138e-05,
      "loss": 0.3573,
      "step": 2919
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.9840326163408428,
      "learning_rate": 1.5788970164005405e-05,
      "loss": 0.3661,
      "step": 2920
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.7233528641145186,
      "learning_rate": 1.5769411799732698e-05,
      "loss": 0.3527,
      "step": 2921
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.6325396263748353,
      "learning_rate": 1.5749861236931263e-05,
      "loss": 0.3443,
      "step": 2922
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.7337882897818788,
      "learning_rate": 1.573031848631907e-05,
      "loss": 0.3413,
      "step": 2923
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.7007896373310991,
      "learning_rate": 1.5710783558609835e-05,
      "loss": 0.3233,
      "step": 2924
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.5721704132447487,
      "learning_rate": 1.5691256464512942e-05,
      "loss": 0.3254,
      "step": 2925
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.6644642201717317,
      "learning_rate": 1.5671737214733514e-05,
      "loss": 0.3634,
      "step": 2926
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.6625500785091931,
      "learning_rate": 1.5652225819972366e-05,
      "loss": 0.3905,
      "step": 2927
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.6662741850450593,
      "learning_rate": 1.5632722290926006e-05,
      "loss": 0.3134,
      "step": 2928
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.604663858504697,
      "learning_rate": 1.5613226638286627e-05,
      "loss": 0.3376,
      "step": 2929
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.7024114510231054,
      "learning_rate": 1.5593738872742105e-05,
      "loss": 0.3482,
      "step": 2930
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.7359119670714909,
      "learning_rate": 1.5574259004975995e-05,
      "loss": 0.3177,
      "step": 2931
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.8676636467477452,
      "learning_rate": 1.555478704566753e-05,
      "loss": 0.3696,
      "step": 2932
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.8680620007856736,
      "learning_rate": 1.5535323005491582e-05,
      "loss": 0.3608,
      "step": 2933
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.4328347654958125,
      "learning_rate": 1.5515866895118695e-05,
      "loss": 0.2946,
      "step": 2934
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.8172449793911727,
      "learning_rate": 1.549641872521508e-05,
      "loss": 0.3738,
      "step": 2935
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.688272221626524,
      "learning_rate": 1.5476978506442576e-05,
      "loss": 0.2926,
      "step": 2936
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.6271532639076185,
      "learning_rate": 1.5457546249458672e-05,
      "loss": 0.2838,
      "step": 2937
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.5800477895526606,
      "learning_rate": 1.5438121964916492e-05,
      "loss": 0.3113,
      "step": 2938
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.7175068951083134,
      "learning_rate": 1.5418705663464786e-05,
      "loss": 0.3507,
      "step": 2939
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.8159139749068232,
      "learning_rate": 1.5399297355747927e-05,
      "loss": 0.3449,
      "step": 2940
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.8301088133599228,
      "learning_rate": 1.5379897052405925e-05,
      "loss": 0.3925,
      "step": 2941
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.946611998243458,
      "learning_rate": 1.5360504764074358e-05,
      "loss": 0.3572,
      "step": 2942
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.657953518645956,
      "learning_rate": 1.5341120501384445e-05,
      "loss": 0.3283,
      "step": 2943
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.7297669154406161,
      "learning_rate": 1.5321744274963e-05,
      "loss": 0.3502,
      "step": 2944
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.7366857126493394,
      "learning_rate": 1.530237609543243e-05,
      "loss": 0.3515,
      "step": 2945
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.7521779825015775,
      "learning_rate": 1.528301597341073e-05,
      "loss": 0.3208,
      "step": 2946
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.6060872221779385,
      "learning_rate": 1.526366391951147e-05,
      "loss": 0.3311,
      "step": 2947
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.624874976095571,
      "learning_rate": 1.5244319944343805e-05,
      "loss": 0.3572,
      "step": 2948
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.9799003433481304,
      "learning_rate": 1.5224984058512472e-05,
      "loss": 0.4491,
      "step": 2949
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.6234210466118209,
      "learning_rate": 1.5205656272617741e-05,
      "loss": 0.2894,
      "step": 2950
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.555546798883995,
      "learning_rate": 1.5186336597255468e-05,
      "loss": 0.2871,
      "step": 2951
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.6161649818792334,
      "learning_rate": 1.5167025043017062e-05,
      "loss": 0.3196,
      "step": 2952
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.68356030099316,
      "learning_rate": 1.5147721620489464e-05,
      "loss": 0.3397,
      "step": 2953
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.6575723198731642,
      "learning_rate": 1.5128426340255176e-05,
      "loss": 0.3572,
      "step": 2954
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.9740136866180402,
      "learning_rate": 1.5109139212892222e-05,
      "loss": 0.4123,
      "step": 2955
    },
    {
      "epoch": 3.38,
      "grad_norm": 2.040551387110036,
      "learning_rate": 1.5089860248974159e-05,
      "loss": 0.3403,
      "step": 2956
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.9755278198036634,
      "learning_rate": 1.5070589459070081e-05,
      "loss": 0.4083,
      "step": 2957
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.8261808268960327,
      "learning_rate": 1.5051326853744591e-05,
      "loss": 0.3561,
      "step": 2958
    },
    {
      "epoch": 3.38,
      "grad_norm": 2.045178536098856,
      "learning_rate": 1.503207244355779e-05,
      "loss": 0.3698,
      "step": 2959
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.6637222066973374,
      "learning_rate": 1.5012826239065297e-05,
      "loss": 0.3884,
      "step": 2960
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.6283101457992561,
      "learning_rate": 1.499358825081826e-05,
      "loss": 0.318,
      "step": 2961
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.7181469648532854,
      "learning_rate": 1.4974358489363304e-05,
      "loss": 0.3347,
      "step": 2962
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.7724374971188146,
      "learning_rate": 1.4955136965242513e-05,
      "loss": 0.3435,
      "step": 2963
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.6030872515069117,
      "learning_rate": 1.493592368899349e-05,
      "loss": 0.345,
      "step": 2964
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.558127293378082,
      "learning_rate": 1.491671867114931e-05,
      "loss": 0.3011,
      "step": 2965
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.753976447563429,
      "learning_rate": 1.4897521922238533e-05,
      "loss": 0.3342,
      "step": 2966
    },
    {
      "epoch": 3.39,
      "grad_norm": 2.0850637360944,
      "learning_rate": 1.4878333452785134e-05,
      "loss": 0.3877,
      "step": 2967
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.8577807219792772,
      "learning_rate": 1.4859153273308624e-05,
      "loss": 0.3269,
      "step": 2968
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.8682414751876746,
      "learning_rate": 1.4839981394323915e-05,
      "loss": 0.3214,
      "step": 2969
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.8229861225462627,
      "learning_rate": 1.4820817826341398e-05,
      "loss": 0.3735,
      "step": 2970
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.7199806592057292,
      "learning_rate": 1.4801662579866876e-05,
      "loss": 0.3441,
      "step": 2971
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.5620592419465489,
      "learning_rate": 1.4782515665401613e-05,
      "loss": 0.2771,
      "step": 2972
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.7698554696930324,
      "learning_rate": 1.4763377093442309e-05,
      "loss": 0.3674,
      "step": 2973
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.7996773003009492,
      "learning_rate": 1.474424687448108e-05,
      "loss": 0.322,
      "step": 2974
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.4286684671132572,
      "learning_rate": 1.4725125019005462e-05,
      "loss": 0.2668,
      "step": 2975
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.7712267605478331,
      "learning_rate": 1.4706011537498412e-05,
      "loss": 0.356,
      "step": 2976
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.9933364354345846,
      "learning_rate": 1.4686906440438298e-05,
      "loss": 0.4033,
      "step": 2977
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.9033692318796294,
      "learning_rate": 1.4667809738298876e-05,
      "loss": 0.3656,
      "step": 2978
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.7233866701077907,
      "learning_rate": 1.4648721441549332e-05,
      "loss": 0.3283,
      "step": 2979
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.7880265830871005,
      "learning_rate": 1.4629641560654196e-05,
      "loss": 0.3247,
      "step": 2980
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.7977983956971981,
      "learning_rate": 1.4610570106073423e-05,
      "loss": 0.3763,
      "step": 2981
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.6571473669273542,
      "learning_rate": 1.4591507088262336e-05,
      "loss": 0.3421,
      "step": 2982
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.7730201798225829,
      "learning_rate": 1.4572452517671635e-05,
      "loss": 0.3917,
      "step": 2983
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.5429435460413903,
      "learning_rate": 1.4553406404747387e-05,
      "loss": 0.3468,
      "step": 2984
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.9629492804140025,
      "learning_rate": 1.4534368759931023e-05,
      "loss": 0.3691,
      "step": 2985
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.5569253501164435,
      "learning_rate": 1.4515339593659331e-05,
      "loss": 0.3277,
      "step": 2986
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.616634267201562,
      "learning_rate": 1.4496318916364462e-05,
      "loss": 0.3266,
      "step": 2987
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.5011808791155468,
      "learning_rate": 1.4477306738473878e-05,
      "loss": 0.3298,
      "step": 2988
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.7159429159622113,
      "learning_rate": 1.4458303070410425e-05,
      "loss": 0.2795,
      "step": 2989
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.6632500094060527,
      "learning_rate": 1.4439307922592257e-05,
      "loss": 0.3371,
      "step": 2990
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.7028979060443121,
      "learning_rate": 1.4420321305432871e-05,
      "loss": 0.3674,
      "step": 2991
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.6318766092917967,
      "learning_rate": 1.4401343229341072e-05,
      "loss": 0.3236,
      "step": 2992
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.6741143660570865,
      "learning_rate": 1.4382373704721003e-05,
      "loss": 0.3267,
      "step": 2993
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.879409243279543,
      "learning_rate": 1.43634127419721e-05,
      "loss": 0.3385,
      "step": 2994
    },
    {
      "epoch": 3.42,
      "grad_norm": 2.0197508010089327,
      "learning_rate": 1.4344460351489115e-05,
      "loss": 0.3737,
      "step": 2995
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.5829418539173687,
      "learning_rate": 1.4325516543662111e-05,
      "loss": 0.321,
      "step": 2996
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.7431327690515066,
      "learning_rate": 1.4306581328876411e-05,
      "loss": 0.3705,
      "step": 2997
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.3886751845134782,
      "learning_rate": 1.4287654717512658e-05,
      "loss": 0.2779,
      "step": 2998
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.5251146369417643,
      "learning_rate": 1.426873671994676e-05,
      "loss": 0.3263,
      "step": 2999
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.6659596192520043,
      "learning_rate": 1.4249827346549947e-05,
      "loss": 0.3319,
      "step": 3000
    },
    {
      "epoch": 3.43,
      "eval_blimp_filtered_avg": 0.7070149253731344,
      "eval_blimp_filtered_std": 0.005049400321561112,
      "step": 3000
    },
    {
      "epoch": 3.43,
      "eval_blimp_supplement_avg": 0.8254310344827587,
      "eval_blimp_supplement_std": 0.017147120451087555,
      "step": 3000
    },
    {
      "epoch": 3.43,
      "eval_vqa_filtered_avg": 0.55,
      "eval_vqa_filtered_std": 0.049999999999999996,
      "step": 3000
    },
    {
      "epoch": 3.43,
      "eval_winoground_filtered_avg": 0.63,
      "eval_winoground_filtered_std": 0.04852365870939099,
      "step": 3000
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.6487038918104202,
      "learning_rate": 1.4230926607688654e-05,
      "loss": 0.3397,
      "step": 3001
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.7706764684003955,
      "learning_rate": 1.4212034513724622e-05,
      "loss": 0.3931,
      "step": 3002
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.5121172515551724,
      "learning_rate": 1.4193151075014851e-05,
      "loss": 0.3001,
      "step": 3003
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.7637532957148088,
      "learning_rate": 1.4174276301911603e-05,
      "loss": 0.3615,
      "step": 3004
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.6697840333103966,
      "learning_rate": 1.4155410204762345e-05,
      "loss": 0.3194,
      "step": 3005
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.5371723055657451,
      "learning_rate": 1.4136552793909843e-05,
      "loss": 0.2967,
      "step": 3006
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.7075288649386442,
      "learning_rate": 1.4117704079692056e-05,
      "loss": 0.3369,
      "step": 3007
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.638572399438955,
      "learning_rate": 1.4098864072442243e-05,
      "loss": 0.3056,
      "step": 3008
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.7659252560919043,
      "learning_rate": 1.40800327824888e-05,
      "loss": 0.348,
      "step": 3009
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.8949626360776999,
      "learning_rate": 1.4061210220155404e-05,
      "loss": 0.3137,
      "step": 3010
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.9376135639660192,
      "learning_rate": 1.4042396395760929e-05,
      "loss": 0.3895,
      "step": 3011
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.6489569915563285,
      "learning_rate": 1.4023591319619461e-05,
      "loss": 0.3553,
      "step": 3012
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.7343888235033202,
      "learning_rate": 1.4004795002040302e-05,
      "loss": 0.3149,
      "step": 3013
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.7142794651346227,
      "learning_rate": 1.398600745332791e-05,
      "loss": 0.3809,
      "step": 3014
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.7252538495476057,
      "learning_rate": 1.3967228683781996e-05,
      "loss": 0.3235,
      "step": 3015
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.709878132329718,
      "learning_rate": 1.3948458703697416e-05,
      "loss": 0.3032,
      "step": 3016
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.778740522550717,
      "learning_rate": 1.3929697523364236e-05,
      "loss": 0.3433,
      "step": 3017
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.852443303117038,
      "learning_rate": 1.3910945153067655e-05,
      "loss": 0.3957,
      "step": 3018
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.7610414703767516,
      "learning_rate": 1.389220160308808e-05,
      "loss": 0.335,
      "step": 3019
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.6302327781381354,
      "learning_rate": 1.3873466883701072e-05,
      "loss": 0.3072,
      "step": 3020
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.75157520640694,
      "learning_rate": 1.3854741005177362e-05,
      "loss": 0.3676,
      "step": 3021
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.728389852711567,
      "learning_rate": 1.3836023977782788e-05,
      "loss": 0.323,
      "step": 3022
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.6601891807215463,
      "learning_rate": 1.3817315811778406e-05,
      "loss": 0.3253,
      "step": 3023
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.6942680177413953,
      "learning_rate": 1.3798616517420365e-05,
      "loss": 0.3744,
      "step": 3024
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.6913231933083042,
      "learning_rate": 1.3779926104959972e-05,
      "loss": 0.3427,
      "step": 3025
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.7736203095539391,
      "learning_rate": 1.3761244584643638e-05,
      "loss": 0.3625,
      "step": 3026
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.818249610109039,
      "learning_rate": 1.3742571966712928e-05,
      "loss": 0.3405,
      "step": 3027
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.8901268008442258,
      "learning_rate": 1.3723908261404514e-05,
      "loss": 0.3335,
      "step": 3028
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.7969664458246266,
      "learning_rate": 1.3705253478950186e-05,
      "loss": 0.3488,
      "step": 3029
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.5820662196347837,
      "learning_rate": 1.3686607629576843e-05,
      "loss": 0.307,
      "step": 3030
    },
    {
      "epoch": 3.46,
      "grad_norm": 2.054394984485385,
      "learning_rate": 1.3667970723506475e-05,
      "loss": 0.3574,
      "step": 3031
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.7648302077044742,
      "learning_rate": 1.3649342770956188e-05,
      "loss": 0.3249,
      "step": 3032
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.4917697783323047,
      "learning_rate": 1.3630723782138161e-05,
      "loss": 0.2975,
      "step": 3033
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.9224348441009182,
      "learning_rate": 1.3612113767259686e-05,
      "loss": 0.3671,
      "step": 3034
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.7445653386996811,
      "learning_rate": 1.3593512736523092e-05,
      "loss": 0.3502,
      "step": 3035
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.9752916084397736,
      "learning_rate": 1.3574920700125817e-05,
      "loss": 0.3752,
      "step": 3036
    },
    {
      "epoch": 3.47,
      "grad_norm": 2.3797031546522596,
      "learning_rate": 1.355633766826036e-05,
      "loss": 0.3947,
      "step": 3037
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.8315844092809528,
      "learning_rate": 1.3537763651114282e-05,
      "loss": 0.342,
      "step": 3038
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.7539520213831874,
      "learning_rate": 1.3519198658870201e-05,
      "loss": 0.3433,
      "step": 3039
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.6734448778850202,
      "learning_rate": 1.350064270170579e-05,
      "loss": 0.3082,
      "step": 3040
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.6325730014267292,
      "learning_rate": 1.3482095789793771e-05,
      "loss": 0.3212,
      "step": 3041
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.603289296504397,
      "learning_rate": 1.346355793330191e-05,
      "loss": 0.2909,
      "step": 3042
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.6437141704805867,
      "learning_rate": 1.3445029142392981e-05,
      "loss": 0.3659,
      "step": 3043
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.7270936542367163,
      "learning_rate": 1.3426509427224818e-05,
      "loss": 0.3372,
      "step": 3044
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.7215332968523593,
      "learning_rate": 1.3407998797950275e-05,
      "loss": 0.3184,
      "step": 3045
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.4654734531853169,
      "learning_rate": 1.3389497264717211e-05,
      "loss": 0.289,
      "step": 3046
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.7303407885179485,
      "learning_rate": 1.337100483766854e-05,
      "loss": 0.3575,
      "step": 3047
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.454991637530655,
      "learning_rate": 1.335252152694211e-05,
      "loss": 0.2869,
      "step": 3048
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.6534214293925706,
      "learning_rate": 1.3334047342670841e-05,
      "loss": 0.3336,
      "step": 3049
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.9259632312673205,
      "learning_rate": 1.3315582294982608e-05,
      "loss": 0.3909,
      "step": 3050
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.4729378289097672,
      "learning_rate": 1.3297126394000307e-05,
      "loss": 0.2791,
      "step": 3051
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.845258721827043,
      "learning_rate": 1.3278679649841783e-05,
      "loss": 0.3683,
      "step": 3052
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.3647525872701112,
      "learning_rate": 1.326024207261989e-05,
      "loss": 0.2956,
      "step": 3053
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.6456345278846,
      "learning_rate": 1.3241813672442445e-05,
      "loss": 0.311,
      "step": 3054
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.8815846319617848,
      "learning_rate": 1.3223394459412261e-05,
      "loss": 0.3329,
      "step": 3055
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.8575993070197596,
      "learning_rate": 1.3204984443627058e-05,
      "loss": 0.3465,
      "step": 3056
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.779311716251278,
      "learning_rate": 1.3186583635179565e-05,
      "loss": 0.3558,
      "step": 3057
    },
    {
      "epoch": 3.49,
      "grad_norm": 2.0926741156129336,
      "learning_rate": 1.3168192044157439e-05,
      "loss": 0.4065,
      "step": 3058
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.9433645499927505,
      "learning_rate": 1.3149809680643303e-05,
      "loss": 0.341,
      "step": 3059
    },
    {
      "epoch": 3.5,
      "grad_norm": 2.124825708609613,
      "learning_rate": 1.3131436554714687e-05,
      "loss": 0.3933,
      "step": 3060
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.7833299665170277,
      "learning_rate": 1.3113072676444078e-05,
      "loss": 0.3151,
      "step": 3061
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.9032221229576078,
      "learning_rate": 1.309471805589891e-05,
      "loss": 0.3591,
      "step": 3062
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.694104033581344,
      "learning_rate": 1.3076372703141522e-05,
      "loss": 0.3167,
      "step": 3063
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.652909967169597,
      "learning_rate": 1.3058036628229181e-05,
      "loss": 0.3026,
      "step": 3064
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.77370510247556,
      "learning_rate": 1.3039709841214042e-05,
      "loss": 0.3209,
      "step": 3065
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.5924321257715062,
      "learning_rate": 1.3021392352143196e-05,
      "loss": 0.299,
      "step": 3066
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.6370960550819704,
      "learning_rate": 1.3003084171058633e-05,
      "loss": 0.2875,
      "step": 3067
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.8015464035595548,
      "learning_rate": 1.2984785307997233e-05,
      "loss": 0.314,
      "step": 3068
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.7156979355439286,
      "learning_rate": 1.2966495772990775e-05,
      "loss": 0.3253,
      "step": 3069
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.618199474233205,
      "learning_rate": 1.2948215576065912e-05,
      "loss": 0.3129,
      "step": 3070
    },
    {
      "epoch": 3.51,
      "grad_norm": 2.100581745521604,
      "learning_rate": 1.2929944727244193e-05,
      "loss": 0.4044,
      "step": 3071
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.6752602557195877,
      "learning_rate": 1.2911683236542044e-05,
      "loss": 0.3151,
      "step": 3072
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.7906710299113704,
      "learning_rate": 1.2893431113970725e-05,
      "loss": 0.3644,
      "step": 3073
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.7107522494388911,
      "learning_rate": 1.2875188369536404e-05,
      "loss": 0.3203,
      "step": 3074
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.59608874207288,
      "learning_rate": 1.2856955013240086e-05,
      "loss": 0.3258,
      "step": 3075
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.471059122162096,
      "learning_rate": 1.2838731055077636e-05,
      "loss": 0.2816,
      "step": 3076
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.556756199506015,
      "learning_rate": 1.2820516505039762e-05,
      "loss": 0.3468,
      "step": 3077
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.8253062503504618,
      "learning_rate": 1.280231137311202e-05,
      "loss": 0.3464,
      "step": 3078
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.5349930998186094,
      "learning_rate": 1.2784115669274797e-05,
      "loss": 0.306,
      "step": 3079
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.3902261975627537,
      "learning_rate": 1.2765929403503315e-05,
      "loss": 0.2618,
      "step": 3080
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.756944469693273,
      "learning_rate": 1.274775258576763e-05,
      "loss": 0.3457,
      "step": 3081
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.6788730696001357,
      "learning_rate": 1.2729585226032592e-05,
      "loss": 0.2879,
      "step": 3082
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.5312871047148184,
      "learning_rate": 1.2711427334257889e-05,
      "loss": 0.2755,
      "step": 3083
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.6780001342962048,
      "learning_rate": 1.2693278920398016e-05,
      "loss": 0.3169,
      "step": 3084
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.848900017237091,
      "learning_rate": 1.2675139994402267e-05,
      "loss": 0.343,
      "step": 3085
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.643053046734982,
      "learning_rate": 1.2657010566214735e-05,
      "loss": 0.2958,
      "step": 3086
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.7739176813165556,
      "learning_rate": 1.2638890645774308e-05,
      "loss": 0.2839,
      "step": 3087
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.76014399340268,
      "learning_rate": 1.2620780243014664e-05,
      "loss": 0.3222,
      "step": 3088
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.872543625173132,
      "learning_rate": 1.2602679367864266e-05,
      "loss": 0.3585,
      "step": 3089
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.4180739829191722,
      "learning_rate": 1.2584588030246332e-05,
      "loss": 0.2742,
      "step": 3090
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.678373546141313,
      "learning_rate": 1.2566506240078875e-05,
      "loss": 0.2917,
      "step": 3091
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.9724428622549384,
      "learning_rate": 1.2548434007274666e-05,
      "loss": 0.32,
      "step": 3092
    },
    {
      "epoch": 3.53,
      "grad_norm": 2.0196336666401806,
      "learning_rate": 1.253037134174124e-05,
      "loss": 0.37,
      "step": 3093
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.853205118958177,
      "learning_rate": 1.2512318253380876e-05,
      "loss": 0.322,
      "step": 3094
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.5759658167625505,
      "learning_rate": 1.2494274752090619e-05,
      "loss": 0.2927,
      "step": 3095
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.743662195575402,
      "learning_rate": 1.2476240847762246e-05,
      "loss": 0.3232,
      "step": 3096
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.9981274469255619,
      "learning_rate": 1.245821655028228e-05,
      "loss": 0.3789,
      "step": 3097
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.61617293126209,
      "learning_rate": 1.2440201869531976e-05,
      "loss": 0.2768,
      "step": 3098
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.6446278745201628,
      "learning_rate": 1.2422196815387302e-05,
      "loss": 0.2931,
      "step": 3099
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.914434652545396,
      "learning_rate": 1.2404201397718969e-05,
      "loss": 0.3093,
      "step": 3100
    },
    {
      "epoch": 3.54,
      "eval_blimp_filtered_avg": 0.7098507462686567,
      "eval_blimp_filtered_std": 0.005049912771209966,
      "step": 3100
    },
    {
      "epoch": 3.54,
      "eval_blimp_supplement_avg": 0.8254310344827587,
      "eval_blimp_supplement_std": 0.017147120451087555,
      "step": 3100
    },
    {
      "epoch": 3.54,
      "eval_vqa_filtered_avg": 0.54,
      "eval_vqa_filtered_std": 0.05009082659620333,
      "step": 3100
    },
    {
      "epoch": 3.54,
      "eval_winoground_filtered_avg": 0.62,
      "eval_winoground_filtered_std": 0.048783173121456316,
      "step": 3100
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.7685506357559746,
      "learning_rate": 1.2386215626392387e-05,
      "loss": 0.3338,
      "step": 3101
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.8228251989090838,
      "learning_rate": 1.2368239511267716e-05,
      "loss": 0.3674,
      "step": 3102
    },
    {
      "epoch": 3.55,
      "grad_norm": 2.261720908845574,
      "learning_rate": 1.2350273062199768e-05,
      "loss": 0.3766,
      "step": 3103
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.9433533020826268,
      "learning_rate": 1.2332316289038088e-05,
      "loss": 0.3864,
      "step": 3104
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.7048571851150447,
      "learning_rate": 1.2314369201626913e-05,
      "loss": 0.3387,
      "step": 3105
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.9682988796336496,
      "learning_rate": 1.2296431809805185e-05,
      "loss": 0.3781,
      "step": 3106
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.451947049340914,
      "learning_rate": 1.227850412340648e-05,
      "loss": 0.303,
      "step": 3107
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.5739877424689739,
      "learning_rate": 1.2260586152259093e-05,
      "loss": 0.3111,
      "step": 3108
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.6888581888525003,
      "learning_rate": 1.2242677906186007e-05,
      "loss": 0.3466,
      "step": 3109
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.7017003632839864,
      "learning_rate": 1.2224779395004854e-05,
      "loss": 0.3353,
      "step": 3110
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.5511523889257761,
      "learning_rate": 1.2206890628527906e-05,
      "loss": 0.2842,
      "step": 3111
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.6268814189838856,
      "learning_rate": 1.218901161656213e-05,
      "loss": 0.3365,
      "step": 3112
    },
    {
      "epoch": 3.56,
      "grad_norm": 2.0371211944697554,
      "learning_rate": 1.2171142368909124e-05,
      "loss": 0.3816,
      "step": 3113
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.8140629940481283,
      "learning_rate": 1.2153282895365154e-05,
      "loss": 0.3527,
      "step": 3114
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.835893121093516,
      "learning_rate": 1.2135433205721078e-05,
      "loss": 0.3572,
      "step": 3115
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.9714285759122068,
      "learning_rate": 1.2117593309762464e-05,
      "loss": 0.438,
      "step": 3116
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.6066233095947327,
      "learning_rate": 1.2099763217269458e-05,
      "loss": 0.2985,
      "step": 3117
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.690971980994285,
      "learning_rate": 1.2081942938016844e-05,
      "loss": 0.3189,
      "step": 3118
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.6176213547013993,
      "learning_rate": 1.2064132481774042e-05,
      "loss": 0.3033,
      "step": 3119
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.7215887701688355,
      "learning_rate": 1.2046331858305048e-05,
      "loss": 0.3182,
      "step": 3120
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.786283551261612,
      "learning_rate": 1.2028541077368508e-05,
      "loss": 0.3516,
      "step": 3121
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.7005848922934785,
      "learning_rate": 1.2010760148717655e-05,
      "loss": 0.2938,
      "step": 3122
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.8524477641719026,
      "learning_rate": 1.199298908210032e-05,
      "loss": 0.3744,
      "step": 3123
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.744156659006687,
      "learning_rate": 1.1975227887258934e-05,
      "loss": 0.2997,
      "step": 3124
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.8094596548246216,
      "learning_rate": 1.1957476573930509e-05,
      "loss": 0.3342,
      "step": 3125
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.762121142624815,
      "learning_rate": 1.1939735151846642e-05,
      "loss": 0.368,
      "step": 3126
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.6612876553444829,
      "learning_rate": 1.1922003630733523e-05,
      "loss": 0.3183,
      "step": 3127
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.7165872497607941,
      "learning_rate": 1.1904282020311872e-05,
      "loss": 0.3098,
      "step": 3128
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.5560508865108225,
      "learning_rate": 1.188657033029702e-05,
      "loss": 0.2704,
      "step": 3129
    },
    {
      "epoch": 3.58,
      "grad_norm": 2.0450408520205827,
      "learning_rate": 1.1868868570398839e-05,
      "loss": 0.3769,
      "step": 3130
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.535688532442569,
      "learning_rate": 1.1851176750321763e-05,
      "loss": 0.2947,
      "step": 3131
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.7779533608235647,
      "learning_rate": 1.183349487976477e-05,
      "loss": 0.3404,
      "step": 3132
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.7775448300993049,
      "learning_rate": 1.1815822968421395e-05,
      "loss": 0.3367,
      "step": 3133
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.6842321187542395,
      "learning_rate": 1.17981610259797e-05,
      "loss": 0.3237,
      "step": 3134
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.683310802122298,
      "learning_rate": 1.178050906212229e-05,
      "loss": 0.304,
      "step": 3135
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.7535692187550078,
      "learning_rate": 1.176286708652631e-05,
      "loss": 0.3504,
      "step": 3136
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.8286250782951312,
      "learning_rate": 1.1745235108863393e-05,
      "loss": 0.3305,
      "step": 3137
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.7006750846405496,
      "learning_rate": 1.1727613138799726e-05,
      "loss": 0.3223,
      "step": 3138
    },
    {
      "epoch": 3.59,
      "grad_norm": 2.0018585121386665,
      "learning_rate": 1.1710001185996002e-05,
      "loss": 0.3639,
      "step": 3139
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.725413282753975,
      "learning_rate": 1.169239926010741e-05,
      "loss": 0.3049,
      "step": 3140
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.4198847586251475,
      "learning_rate": 1.1674807370783663e-05,
      "loss": 0.2855,
      "step": 3141
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.8336687597290684,
      "learning_rate": 1.1657225527668948e-05,
      "loss": 0.323,
      "step": 3142
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.8125601590869638,
      "learning_rate": 1.1639653740401966e-05,
      "loss": 0.3383,
      "step": 3143
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.914531229550508,
      "learning_rate": 1.1622092018615897e-05,
      "loss": 0.3655,
      "step": 3144
    },
    {
      "epoch": 3.59,
      "grad_norm": 2.0345544773110067,
      "learning_rate": 1.1604540371938388e-05,
      "loss": 0.3379,
      "step": 3145
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.928188113062439,
      "learning_rate": 1.1586998809991582e-05,
      "loss": 0.3446,
      "step": 3146
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.9036125167376081,
      "learning_rate": 1.1569467342392084e-05,
      "loss": 0.3635,
      "step": 3147
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.5834036153835496,
      "learning_rate": 1.1551945978750977e-05,
      "loss": 0.2782,
      "step": 3148
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.9299324513842362,
      "learning_rate": 1.1534434728673788e-05,
      "loss": 0.2991,
      "step": 3149
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.7218821763051888,
      "learning_rate": 1.1516933601760514e-05,
      "loss": 0.324,
      "step": 3150
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.6253873332626856,
      "learning_rate": 1.1499442607605588e-05,
      "loss": 0.2712,
      "step": 3151
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.722361288611786,
      "learning_rate": 1.1481961755797907e-05,
      "loss": 0.3293,
      "step": 3152
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.6929164893017825,
      "learning_rate": 1.1464491055920799e-05,
      "loss": 0.3047,
      "step": 3153
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.91957171601289,
      "learning_rate": 1.1447030517552007e-05,
      "loss": 0.3198,
      "step": 3154
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.94350238715806,
      "learning_rate": 1.1429580150263716e-05,
      "loss": 0.3707,
      "step": 3155
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.68918214251136,
      "learning_rate": 1.1412139963622567e-05,
      "loss": 0.2767,
      "step": 3156
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.6106692861637166,
      "learning_rate": 1.139470996718959e-05,
      "loss": 0.3019,
      "step": 3157
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.7957865613450776,
      "learning_rate": 1.1377290170520209e-05,
      "loss": 0.3063,
      "step": 3158
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.6453976228385057,
      "learning_rate": 1.1359880583164288e-05,
      "loss": 0.2935,
      "step": 3159
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.747599700409099,
      "learning_rate": 1.1342481214666086e-05,
      "loss": 0.2924,
      "step": 3160
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.970509718057189,
      "learning_rate": 1.1325092074564265e-05,
      "loss": 0.3337,
      "step": 3161
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.981064074497533,
      "learning_rate": 1.130771317239185e-05,
      "loss": 0.3548,
      "step": 3162
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.7793828644203094,
      "learning_rate": 1.1290344517676296e-05,
      "loss": 0.3022,
      "step": 3163
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.8706012395138774,
      "learning_rate": 1.127298611993942e-05,
      "loss": 0.3246,
      "step": 3164
    },
    {
      "epoch": 3.62,
      "grad_norm": 2.0356096155160976,
      "learning_rate": 1.1255637988697421e-05,
      "loss": 0.3559,
      "step": 3165
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.6757282198159555,
      "learning_rate": 1.1238300133460846e-05,
      "loss": 0.2955,
      "step": 3166
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.9562470587837217,
      "learning_rate": 1.1220972563734635e-05,
      "loss": 0.3571,
      "step": 3167
    },
    {
      "epoch": 3.62,
      "grad_norm": 2.0305306635132627,
      "learning_rate": 1.1203655289018087e-05,
      "loss": 0.3919,
      "step": 3168
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.770924477189803,
      "learning_rate": 1.1186348318804846e-05,
      "loss": 0.3449,
      "step": 3169
    },
    {
      "epoch": 3.62,
      "grad_norm": 2.0193135527685326,
      "learning_rate": 1.1169051662582918e-05,
      "loss": 0.3539,
      "step": 3170
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.7258764319625994,
      "learning_rate": 1.1151765329834648e-05,
      "loss": 0.3209,
      "step": 3171
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.999129348730187,
      "learning_rate": 1.113448933003672e-05,
      "loss": 0.3867,
      "step": 3172
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.5533157941655678,
      "learning_rate": 1.1117223672660158e-05,
      "loss": 0.2908,
      "step": 3173
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.5330718953044564,
      "learning_rate": 1.1099968367170324e-05,
      "loss": 0.3447,
      "step": 3174
    },
    {
      "epoch": 3.63,
      "grad_norm": 2.0421385469853313,
      "learning_rate": 1.1082723423026874e-05,
      "loss": 0.3659,
      "step": 3175
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.7856043838931894,
      "learning_rate": 1.106548884968382e-05,
      "loss": 0.3272,
      "step": 3176
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.5312559452498011,
      "learning_rate": 1.1048264656589467e-05,
      "loss": 0.2841,
      "step": 3177
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.880956078769587,
      "learning_rate": 1.103105085318644e-05,
      "loss": 0.3798,
      "step": 3178
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.66802743430195,
      "learning_rate": 1.1013847448911661e-05,
      "loss": 0.3236,
      "step": 3179
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.4714756258651984,
      "learning_rate": 1.0996654453196357e-05,
      "loss": 0.2932,
      "step": 3180
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.6696491197436587,
      "learning_rate": 1.0979471875466042e-05,
      "loss": 0.2753,
      "step": 3181
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.8870449810252428,
      "learning_rate": 1.0962299725140536e-05,
      "loss": 0.343,
      "step": 3182
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.6051815940509764,
      "learning_rate": 1.0945138011633905e-05,
      "loss": 0.2916,
      "step": 3183
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.8967986632257379,
      "learning_rate": 1.0927986744354534e-05,
      "loss": 0.3271,
      "step": 3184
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.6749945335587793,
      "learning_rate": 1.0910845932705057e-05,
      "loss": 0.3107,
      "step": 3185
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.741143583536521,
      "learning_rate": 1.089371558608239e-05,
      "loss": 0.2866,
      "step": 3186
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.779821576344915,
      "learning_rate": 1.0876595713877699e-05,
      "loss": 0.3195,
      "step": 3187
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.8391988151883165,
      "learning_rate": 1.0859486325476416e-05,
      "loss": 0.3196,
      "step": 3188
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.9662418641127082,
      "learning_rate": 1.0842387430258227e-05,
      "loss": 0.3216,
      "step": 3189
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.9753262645627967,
      "learning_rate": 1.0825299037597062e-05,
      "loss": 0.2974,
      "step": 3190
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.8023690770140015,
      "learning_rate": 1.08082211568611e-05,
      "loss": 0.3342,
      "step": 3191
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.997699633196374,
      "learning_rate": 1.0791153797412733e-05,
      "loss": 0.358,
      "step": 3192
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.4510131879649206,
      "learning_rate": 1.077409696860861e-05,
      "loss": 0.2503,
      "step": 3193
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.9157371432197379,
      "learning_rate": 1.0757050679799591e-05,
      "loss": 0.3139,
      "step": 3194
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.8237478733852026,
      "learning_rate": 1.07400149403308e-05,
      "loss": 0.3422,
      "step": 3195
    },
    {
      "epoch": 3.65,
      "grad_norm": 2.063867860065222,
      "learning_rate": 1.0722989759541507e-05,
      "loss": 0.3564,
      "step": 3196
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.989892710497351,
      "learning_rate": 1.0705975146765247e-05,
      "loss": 0.3435,
      "step": 3197
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.694505597615198,
      "learning_rate": 1.0688971111329743e-05,
      "loss": 0.3231,
      "step": 3198
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.6775834060244599,
      "learning_rate": 1.067197766255693e-05,
      "loss": 0.3168,
      "step": 3199
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.9110297245788337,
      "learning_rate": 1.0654994809762909e-05,
      "loss": 0.3239,
      "step": 3200
    },
    {
      "epoch": 3.66,
      "eval_blimp_filtered_avg": 0.7083582089552238,
      "eval_blimp_filtered_std": 0.005044652118903126,
      "step": 3200
    },
    {
      "epoch": 3.66,
      "eval_blimp_supplement_avg": 0.8297413793103449,
      "eval_blimp_supplement_std": 0.01692664042823066,
      "step": 3200
    },
    {
      "epoch": 3.66,
      "eval_vqa_filtered_avg": 0.55,
      "eval_vqa_filtered_std": 0.049999999999999996,
      "step": 3200
    },
    {
      "epoch": 3.66,
      "eval_winoground_filtered_avg": 0.62,
      "eval_winoground_filtered_std": 0.048783173121456316,
      "step": 3200
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.8331856992185267,
      "learning_rate": 1.0638022562258006e-05,
      "loss": 0.3495,
      "step": 3201
    },
    {
      "epoch": 3.66,
      "grad_norm": 2.0471979889035685,
      "learning_rate": 1.062106092934671e-05,
      "loss": 0.3632,
      "step": 3202
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.6317056423550587,
      "learning_rate": 1.0604109920327732e-05,
      "loss": 0.3282,
      "step": 3203
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.5319454718563028,
      "learning_rate": 1.0587169544493898e-05,
      "loss": 0.2692,
      "step": 3204
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.7178439999291004,
      "learning_rate": 1.0570239811132244e-05,
      "loss": 0.3262,
      "step": 3205
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.8399791580365275,
      "learning_rate": 1.0553320729523964e-05,
      "loss": 0.3329,
      "step": 3206
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.854066951971062,
      "learning_rate": 1.053641230894441e-05,
      "loss": 0.375,
      "step": 3207
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.678597156121599,
      "learning_rate": 1.05195145586631e-05,
      "loss": 0.3073,
      "step": 3208
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.9492656998004563,
      "learning_rate": 1.0502627487943662e-05,
      "loss": 0.3374,
      "step": 3209
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.8850158765034448,
      "learning_rate": 1.0485751106043932e-05,
      "loss": 0.3058,
      "step": 3210
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.7155528357216145,
      "learning_rate": 1.0468885422215846e-05,
      "loss": 0.3186,
      "step": 3211
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.8840864422043984,
      "learning_rate": 1.0452030445705489e-05,
      "loss": 0.3693,
      "step": 3212
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.8566329638040096,
      "learning_rate": 1.0435186185753051e-05,
      "loss": 0.354,
      "step": 3213
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.8243655621345662,
      "learning_rate": 1.041835265159288e-05,
      "loss": 0.3358,
      "step": 3214
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.895834760102567,
      "learning_rate": 1.0401529852453426e-05,
      "loss": 0.3582,
      "step": 3215
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.883855156218635,
      "learning_rate": 1.0384717797557272e-05,
      "loss": 0.3488,
      "step": 3216
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.8753385172035024,
      "learning_rate": 1.0367916496121066e-05,
      "loss": 0.3564,
      "step": 3217
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.5581239097231536,
      "learning_rate": 1.0351125957355619e-05,
      "loss": 0.2929,
      "step": 3218
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.6712497567200275,
      "learning_rate": 1.0334346190465808e-05,
      "loss": 0.3176,
      "step": 3219
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.9841056741883214,
      "learning_rate": 1.0317577204650619e-05,
      "loss": 0.352,
      "step": 3220
    },
    {
      "epoch": 3.68,
      "grad_norm": 2.277745459834375,
      "learning_rate": 1.0300819009103099e-05,
      "loss": 0.3337,
      "step": 3221
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.7091811518707942,
      "learning_rate": 1.0284071613010411e-05,
      "loss": 0.3094,
      "step": 3222
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.5437048480473454,
      "learning_rate": 1.0267335025553791e-05,
      "loss": 0.2785,
      "step": 3223
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.4740095056069515,
      "learning_rate": 1.0250609255908541e-05,
      "loss": 0.2472,
      "step": 3224
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.8443298883050658,
      "learning_rate": 1.0233894313244035e-05,
      "loss": 0.377,
      "step": 3225
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.9049423886351762,
      "learning_rate": 1.0217190206723716e-05,
      "loss": 0.3948,
      "step": 3226
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.5565407875237012,
      "learning_rate": 1.0200496945505085e-05,
      "loss": 0.3111,
      "step": 3227
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.8634186939314208,
      "learning_rate": 1.0183814538739696e-05,
      "loss": 0.3617,
      "step": 3228
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.8725958676122814,
      "learning_rate": 1.0167142995573156e-05,
      "loss": 0.3186,
      "step": 3229
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.6512325529283156,
      "learning_rate": 1.0150482325145102e-05,
      "loss": 0.3072,
      "step": 3230
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.755744132242807,
      "learning_rate": 1.0133832536589225e-05,
      "loss": 0.3123,
      "step": 3231
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.8594090122073734,
      "learning_rate": 1.011719363903325e-05,
      "loss": 0.3232,
      "step": 3232
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.7943586675953913,
      "learning_rate": 1.0100565641598927e-05,
      "loss": 0.3085,
      "step": 3233
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.4177633952307087,
      "learning_rate": 1.008394855340203e-05,
      "loss": 0.2539,
      "step": 3234
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.7922484554834204,
      "learning_rate": 1.0067342383552357e-05,
      "loss": 0.3426,
      "step": 3235
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.9923380361743905,
      "learning_rate": 1.0050747141153711e-05,
      "loss": 0.3498,
      "step": 3236
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.6547338045205031,
      "learning_rate": 1.0034162835303929e-05,
      "loss": 0.321,
      "step": 3237
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.6455944465720236,
      "learning_rate": 1.0017589475094802e-05,
      "loss": 0.3033,
      "step": 3238
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.839029796939605,
      "learning_rate": 1.0001027069612173e-05,
      "loss": 0.3401,
      "step": 3239
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.692549470487343,
      "learning_rate": 9.984475627935853e-06,
      "loss": 0.3496,
      "step": 3240
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.5801293975463875,
      "learning_rate": 9.967935159139641e-06,
      "loss": 0.2839,
      "step": 3241
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.7638184144413835,
      "learning_rate": 9.951405672291356e-06,
      "loss": 0.3044,
      "step": 3242
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.789553206869906,
      "learning_rate": 9.934887176452738e-06,
      "loss": 0.3229,
      "step": 3243
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.6099079330659272,
      "learning_rate": 9.91837968067954e-06,
      "loss": 0.338,
      "step": 3244
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.9214665474174653,
      "learning_rate": 9.90188319402148e-06,
      "loss": 0.3211,
      "step": 3245
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.984817250367123,
      "learning_rate": 9.885397725522241e-06,
      "loss": 0.3407,
      "step": 3246
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.7260492628041721,
      "learning_rate": 9.868923284219445e-06,
      "loss": 0.2874,
      "step": 3247
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.6792347650158006,
      "learning_rate": 9.852459879144694e-06,
      "loss": 0.3092,
      "step": 3248
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.7235056717253587,
      "learning_rate": 9.836007519323523e-06,
      "loss": 0.3014,
      "step": 3249
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.7881623973782552,
      "learning_rate": 9.819566213775444e-06,
      "loss": 0.2733,
      "step": 3250
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.5293450706414344,
      "learning_rate": 9.803135971513861e-06,
      "loss": 0.2722,
      "step": 3251
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.9187274362821638,
      "learning_rate": 9.786716801546144e-06,
      "loss": 0.3465,
      "step": 3252
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.6796549274277277,
      "learning_rate": 9.770308712873581e-06,
      "loss": 0.3008,
      "step": 3253
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.7981420347283326,
      "learning_rate": 9.753911714491405e-06,
      "loss": 0.3111,
      "step": 3254
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.9775276857130313,
      "learning_rate": 9.737525815388731e-06,
      "loss": 0.3248,
      "step": 3255
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.8502394828142499,
      "learning_rate": 9.721151024548613e-06,
      "loss": 0.3342,
      "step": 3256
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.7967109555963228,
      "learning_rate": 9.70478735094803e-06,
      "loss": 0.3086,
      "step": 3257
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.8723825793622173,
      "learning_rate": 9.688434803557841e-06,
      "loss": 0.3136,
      "step": 3258
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.8369264412914914,
      "learning_rate": 9.672093391342826e-06,
      "loss": 0.3745,
      "step": 3259
    },
    {
      "epoch": 3.73,
      "grad_norm": 2.047048236646556,
      "learning_rate": 9.655763123261632e-06,
      "loss": 0.3668,
      "step": 3260
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.8414233284113417,
      "learning_rate": 9.639444008266818e-06,
      "loss": 0.3514,
      "step": 3261
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.8859118650406699,
      "learning_rate": 9.623136055304829e-06,
      "loss": 0.3539,
      "step": 3262
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.6177122732475666,
      "learning_rate": 9.60683927331598e-06,
      "loss": 0.2938,
      "step": 3263
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.7923717358316071,
      "learning_rate": 9.590553671234473e-06,
      "loss": 0.3271,
      "step": 3264
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.4722256348724987,
      "learning_rate": 9.574279257988375e-06,
      "loss": 0.3065,
      "step": 3265
    },
    {
      "epoch": 3.73,
      "grad_norm": 2.528122471235067,
      "learning_rate": 9.558016042499618e-06,
      "loss": 0.3992,
      "step": 3266
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.6193624434893126,
      "learning_rate": 9.541764033684009e-06,
      "loss": 0.3104,
      "step": 3267
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.9315117408661655,
      "learning_rate": 9.525523240451176e-06,
      "loss": 0.3646,
      "step": 3268
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.8456647472953083,
      "learning_rate": 9.50929367170464e-06,
      "loss": 0.324,
      "step": 3269
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.621827145596258,
      "learning_rate": 9.493075336341739e-06,
      "loss": 0.3195,
      "step": 3270
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.4601736027717929,
      "learning_rate": 9.476868243253673e-06,
      "loss": 0.2584,
      "step": 3271
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.5471014887907844,
      "learning_rate": 9.460672401325467e-06,
      "loss": 0.3157,
      "step": 3272
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.7691086063417354,
      "learning_rate": 9.444487819435979e-06,
      "loss": 0.3289,
      "step": 3273
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.5848702178386238,
      "learning_rate": 9.428314506457895e-06,
      "loss": 0.2883,
      "step": 3274
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.9392359523350995,
      "learning_rate": 9.412152471257729e-06,
      "loss": 0.3514,
      "step": 3275
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.7542827031201322,
      "learning_rate": 9.396001722695811e-06,
      "loss": 0.3075,
      "step": 3276
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.5154417509067974,
      "learning_rate": 9.379862269626263e-06,
      "loss": 0.2996,
      "step": 3277
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.87231573291652,
      "learning_rate": 9.363734120897044e-06,
      "loss": 0.3382,
      "step": 3278
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.7423924916220017,
      "learning_rate": 9.347617285349896e-06,
      "loss": 0.3179,
      "step": 3279
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.813311382798986,
      "learning_rate": 9.331511771820371e-06,
      "loss": 0.3527,
      "step": 3280
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.818371353676802,
      "learning_rate": 9.315417589137805e-06,
      "loss": 0.2813,
      "step": 3281
    },
    {
      "epoch": 3.75,
      "grad_norm": 2.055502876518803,
      "learning_rate": 9.299334746125333e-06,
      "loss": 0.3543,
      "step": 3282
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.9544375256614137,
      "learning_rate": 9.283263251599863e-06,
      "loss": 0.3198,
      "step": 3283
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.8942639767768696,
      "learning_rate": 9.26720311437209e-06,
      "loss": 0.2967,
      "step": 3284
    },
    {
      "epoch": 3.75,
      "grad_norm": 2.0677455288445135,
      "learning_rate": 9.251154343246464e-06,
      "loss": 0.3388,
      "step": 3285
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.5601137076964275,
      "learning_rate": 9.235116947021225e-06,
      "loss": 0.2784,
      "step": 3286
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.7674832339223003,
      "learning_rate": 9.219090934488373e-06,
      "loss": 0.2995,
      "step": 3287
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.641561140869782,
      "learning_rate": 9.20307631443366e-06,
      "loss": 0.3115,
      "step": 3288
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.816478001779519,
      "learning_rate": 9.187073095636601e-06,
      "loss": 0.3109,
      "step": 3289
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.708422379718636,
      "learning_rate": 9.171081286870455e-06,
      "loss": 0.3005,
      "step": 3290
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.7402412025763432,
      "learning_rate": 9.15510089690222e-06,
      "loss": 0.3482,
      "step": 3291
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.8904842103777224,
      "learning_rate": 9.13913193449266e-06,
      "loss": 0.3365,
      "step": 3292
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.647057906657686,
      "learning_rate": 9.12317440839623e-06,
      "loss": 0.2553,
      "step": 3293
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.8017862630251873,
      "learning_rate": 9.107228327361151e-06,
      "loss": 0.3628,
      "step": 3294
    },
    {
      "epoch": 3.77,
      "grad_norm": 2.08061430291548,
      "learning_rate": 9.091293700129362e-06,
      "loss": 0.3558,
      "step": 3295
    },
    {
      "epoch": 3.77,
      "grad_norm": 2.2119605927637007,
      "learning_rate": 9.075370535436506e-06,
      "loss": 0.3619,
      "step": 3296
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.787614389755362,
      "learning_rate": 9.05945884201199e-06,
      "loss": 0.2818,
      "step": 3297
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.9785023301311158,
      "learning_rate": 9.043558628578865e-06,
      "loss": 0.3417,
      "step": 3298
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.886855393160379,
      "learning_rate": 9.027669903853932e-06,
      "loss": 0.2996,
      "step": 3299
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.8637981227175542,
      "learning_rate": 9.011792676547684e-06,
      "loss": 0.2793,
      "step": 3300
    },
    {
      "epoch": 3.77,
      "eval_blimp_filtered_avg": 0.7092537313432836,
      "eval_blimp_filtered_std": 0.005056992622238261,
      "step": 3300
    },
    {
      "epoch": 3.77,
      "eval_blimp_supplement_avg": 0.8275862068965517,
      "eval_blimp_supplement_std": 0.017052388046731506,
      "step": 3300
    },
    {
      "epoch": 3.77,
      "eval_vqa_filtered_avg": 0.54,
      "eval_vqa_filtered_std": 0.05009082659620333,
      "step": 3300
    },
    {
      "epoch": 3.77,
      "eval_winoground_filtered_avg": 0.63,
      "eval_winoground_filtered_std": 0.04852365870939099,
      "step": 3300
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.6495857840064685,
      "learning_rate": 8.995926955364322e-06,
      "loss": 0.3149,
      "step": 3301
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.939119727354639,
      "learning_rate": 8.980072749001707e-06,
      "loss": 0.3267,
      "step": 3302
    },
    {
      "epoch": 3.77,
      "grad_norm": 2.140761475807706,
      "learning_rate": 8.964230066151401e-06,
      "loss": 0.3459,
      "step": 3303
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.7856493807426588,
      "learning_rate": 8.948398915498688e-06,
      "loss": 0.3282,
      "step": 3304
    },
    {
      "epoch": 3.78,
      "grad_norm": 2.204707880516655,
      "learning_rate": 8.932579305722484e-06,
      "loss": 0.353,
      "step": 3305
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.8815934955482945,
      "learning_rate": 8.91677124549538e-06,
      "loss": 0.343,
      "step": 3306
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.9241259356038334,
      "learning_rate": 8.900974743483654e-06,
      "loss": 0.3572,
      "step": 3307
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.7558692108877554,
      "learning_rate": 8.885189808347237e-06,
      "loss": 0.3135,
      "step": 3308
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.728513353860636,
      "learning_rate": 8.869416448739738e-06,
      "loss": 0.35,
      "step": 3309
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.7511006693218687,
      "learning_rate": 8.853654673308369e-06,
      "loss": 0.3298,
      "step": 3310
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.9419981529880552,
      "learning_rate": 8.837904490694058e-06,
      "loss": 0.3713,
      "step": 3311
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.8675675072337479,
      "learning_rate": 8.822165909531339e-06,
      "loss": 0.3267,
      "step": 3312
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.7160989227589454,
      "learning_rate": 8.806438938448388e-06,
      "loss": 0.3243,
      "step": 3313
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.9647972831151246,
      "learning_rate": 8.790723586067036e-06,
      "loss": 0.3145,
      "step": 3314
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.9148949294814661,
      "learning_rate": 8.775019861002703e-06,
      "loss": 0.3661,
      "step": 3315
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.439701057519792,
      "learning_rate": 8.759327771864475e-06,
      "loss": 0.2685,
      "step": 3316
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.7482887416405477,
      "learning_rate": 8.743647327255046e-06,
      "loss": 0.332,
      "step": 3317
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.3844612574257027,
      "learning_rate": 8.727978535770724e-06,
      "loss": 0.2557,
      "step": 3318
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.7935910408683542,
      "learning_rate": 8.71232140600143e-06,
      "loss": 0.3174,
      "step": 3319
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.5289781833358091,
      "learning_rate": 8.69667594653069e-06,
      "loss": 0.2676,
      "step": 3320
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.7028109130207243,
      "learning_rate": 8.68104216593564e-06,
      "loss": 0.2673,
      "step": 3321
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.9174336464035668,
      "learning_rate": 8.665420072787006e-06,
      "loss": 0.3171,
      "step": 3322
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.7161868038352026,
      "learning_rate": 8.649809675649101e-06,
      "loss": 0.2984,
      "step": 3323
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.9057473498999278,
      "learning_rate": 8.634210983079832e-06,
      "loss": 0.3462,
      "step": 3324
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.7215320440416964,
      "learning_rate": 8.618624003630698e-06,
      "loss": 0.2986,
      "step": 3325
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.9644824658857407,
      "learning_rate": 8.603048745846764e-06,
      "loss": 0.312,
      "step": 3326
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.706403257836996,
      "learning_rate": 8.58748521826668e-06,
      "loss": 0.2635,
      "step": 3327
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.6373016225617312,
      "learning_rate": 8.571933429422652e-06,
      "loss": 0.309,
      "step": 3328
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.6536959559596802,
      "learning_rate": 8.55639338784046e-06,
      "loss": 0.2563,
      "step": 3329
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.8006627478038464,
      "learning_rate": 8.540865102039446e-06,
      "loss": 0.3223,
      "step": 3330
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.5743388070666295,
      "learning_rate": 8.525348580532507e-06,
      "loss": 0.2691,
      "step": 3331
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.763759660128755,
      "learning_rate": 8.509843831826074e-06,
      "loss": 0.3184,
      "step": 3332
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.7954828370735978,
      "learning_rate": 8.494350864420138e-06,
      "loss": 0.3228,
      "step": 3333
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.8661265560530769,
      "learning_rate": 8.478869686808235e-06,
      "loss": 0.3155,
      "step": 3334
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.9187469258495158,
      "learning_rate": 8.463400307477433e-06,
      "loss": 0.3107,
      "step": 3335
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.8347169785415125,
      "learning_rate": 8.447942734908327e-06,
      "loss": 0.3611,
      "step": 3336
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.5544187487840204,
      "learning_rate": 8.432496977575051e-06,
      "loss": 0.2688,
      "step": 3337
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.7731190225441797,
      "learning_rate": 8.417063043945247e-06,
      "loss": 0.2897,
      "step": 3338
    },
    {
      "epoch": 3.82,
      "grad_norm": 2.0245925177190394,
      "learning_rate": 8.401640942480098e-06,
      "loss": 0.365,
      "step": 3339
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.7154405520034934,
      "learning_rate": 8.38623068163426e-06,
      "loss": 0.3386,
      "step": 3340
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.7877388731096566,
      "learning_rate": 8.370832269855937e-06,
      "loss": 0.3049,
      "step": 3341
    },
    {
      "epoch": 3.82,
      "grad_norm": 2.033744803446432,
      "learning_rate": 8.355445715586821e-06,
      "loss": 0.3528,
      "step": 3342
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.7875128155822464,
      "learning_rate": 8.340071027262107e-06,
      "loss": 0.3418,
      "step": 3343
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.6385196393986419,
      "learning_rate": 8.324708213310476e-06,
      "loss": 0.2869,
      "step": 3344
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.8604911597402665,
      "learning_rate": 8.309357282154113e-06,
      "loss": 0.3033,
      "step": 3345
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.667526591644457,
      "learning_rate": 8.294018242208683e-06,
      "loss": 0.3357,
      "step": 3346
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.7077882900033745,
      "learning_rate": 8.278691101883328e-06,
      "loss": 0.2984,
      "step": 3347
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.6150443950671687,
      "learning_rate": 8.263375869580676e-06,
      "loss": 0.2874,
      "step": 3348
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.6531434748797083,
      "learning_rate": 8.248072553696808e-06,
      "loss": 0.2891,
      "step": 3349
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.952644951857894,
      "learning_rate": 8.232781162621281e-06,
      "loss": 0.3252,
      "step": 3350
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.6406123609875942,
      "learning_rate": 8.217501704737136e-06,
      "loss": 0.2819,
      "step": 3351
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.6263558288896058,
      "learning_rate": 8.202234188420855e-06,
      "loss": 0.2884,
      "step": 3352
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.5825244473390558,
      "learning_rate": 8.186978622042354e-06,
      "loss": 0.2771,
      "step": 3353
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.711973177038978,
      "learning_rate": 8.171735013965021e-06,
      "loss": 0.3443,
      "step": 3354
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.7019945922489752,
      "learning_rate": 8.156503372545694e-06,
      "loss": 0.2738,
      "step": 3355
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.6166323461697025,
      "learning_rate": 8.141283706134639e-06,
      "loss": 0.2884,
      "step": 3356
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.5290359027319178,
      "learning_rate": 8.126076023075537e-06,
      "loss": 0.2965,
      "step": 3357
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.7339465406968944,
      "learning_rate": 8.110880331705542e-06,
      "loss": 0.2931,
      "step": 3358
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.8500526120012915,
      "learning_rate": 8.09569664035521e-06,
      "loss": 0.3398,
      "step": 3359
    },
    {
      "epoch": 3.84,
      "grad_norm": 2.1702025230142743,
      "learning_rate": 8.080524957348523e-06,
      "loss": 0.3957,
      "step": 3360
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.7484076341313342,
      "learning_rate": 8.06536529100287e-06,
      "loss": 0.2847,
      "step": 3361
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.6489149537449646,
      "learning_rate": 8.050217649629059e-06,
      "loss": 0.3236,
      "step": 3362
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.6254384750893296,
      "learning_rate": 8.035082041531312e-06,
      "loss": 0.2908,
      "step": 3363
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.6305850165184994,
      "learning_rate": 8.019958475007252e-06,
      "loss": 0.2764,
      "step": 3364
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.9916629421741612,
      "learning_rate": 8.00484695834789e-06,
      "loss": 0.3763,
      "step": 3365
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.6762577750137426,
      "learning_rate": 7.989747499837641e-06,
      "loss": 0.3156,
      "step": 3366
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.7960649256629853,
      "learning_rate": 7.97466010775431e-06,
      "loss": 0.2925,
      "step": 3367
    },
    {
      "epoch": 3.85,
      "grad_norm": 2.266307556638757,
      "learning_rate": 7.95958479036908e-06,
      "loss": 0.3819,
      "step": 3368
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.6356817832917125,
      "learning_rate": 7.944521555946525e-06,
      "loss": 0.2893,
      "step": 3369
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.70561682902458,
      "learning_rate": 7.92947041274458e-06,
      "loss": 0.2688,
      "step": 3370
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.5876081967276376,
      "learning_rate": 7.914431369014564e-06,
      "loss": 0.2883,
      "step": 3371
    },
    {
      "epoch": 3.85,
      "grad_norm": 2.0950903536767793,
      "learning_rate": 7.899404433001156e-06,
      "loss": 0.3774,
      "step": 3372
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.8250395984914607,
      "learning_rate": 7.884389612942399e-06,
      "loss": 0.3292,
      "step": 3373
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.841852931771991,
      "learning_rate": 7.8693869170697e-06,
      "loss": 0.3358,
      "step": 3374
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.8758736086950833,
      "learning_rate": 7.854396353607812e-06,
      "loss": 0.3286,
      "step": 3375
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.8311256216390883,
      "learning_rate": 7.839417930774837e-06,
      "loss": 0.3047,
      "step": 3376
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.8750632231291036,
      "learning_rate": 7.824451656782233e-06,
      "loss": 0.3274,
      "step": 3377
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.8811922627086772,
      "learning_rate": 7.809497539834767e-06,
      "loss": 0.3384,
      "step": 3378
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.7863556881836633,
      "learning_rate": 7.794555588130576e-06,
      "loss": 0.3169,
      "step": 3379
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.8503588985509487,
      "learning_rate": 7.779625809861116e-06,
      "loss": 0.3436,
      "step": 3380
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.97503966655127,
      "learning_rate": 7.764708213211156e-06,
      "loss": 0.2863,
      "step": 3381
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.9033967489083723,
      "learning_rate": 7.749802806358807e-06,
      "loss": 0.2977,
      "step": 3382
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.9105915367793516,
      "learning_rate": 7.734909597475488e-06,
      "loss": 0.3413,
      "step": 3383
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.878396529653984,
      "learning_rate": 7.72002859472593e-06,
      "loss": 0.294,
      "step": 3384
    },
    {
      "epoch": 3.87,
      "grad_norm": 2.0015262375848404,
      "learning_rate": 7.70515980626817e-06,
      "loss": 0.3666,
      "step": 3385
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.9056655053562435,
      "learning_rate": 7.690303240253563e-06,
      "loss": 0.3319,
      "step": 3386
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.9094300974369116,
      "learning_rate": 7.67545890482674e-06,
      "loss": 0.3309,
      "step": 3387
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.933872164825433,
      "learning_rate": 7.660626808125643e-06,
      "loss": 0.3079,
      "step": 3388
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.7083761642108133,
      "learning_rate": 7.645806958281498e-06,
      "loss": 0.3121,
      "step": 3389
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.5884443815651115,
      "learning_rate": 7.630999363418838e-06,
      "loss": 0.2716,
      "step": 3390
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.7179296047333583,
      "learning_rate": 7.616204031655442e-06,
      "loss": 0.3338,
      "step": 3391
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.8647101141265858,
      "learning_rate": 7.60142097110239e-06,
      "loss": 0.3003,
      "step": 3392
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.7782833006087928,
      "learning_rate": 7.58665018986403e-06,
      "loss": 0.2977,
      "step": 3393
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.911143213895411,
      "learning_rate": 7.57189169603798e-06,
      "loss": 0.3233,
      "step": 3394
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.7900772830765816,
      "learning_rate": 7.55714549771511e-06,
      "loss": 0.3486,
      "step": 3395
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.8784229038858753,
      "learning_rate": 7.542411602979558e-06,
      "loss": 0.3431,
      "step": 3396
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.7234450720737096,
      "learning_rate": 7.527690019908718e-06,
      "loss": 0.3236,
      "step": 3397
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.7971786541386872,
      "learning_rate": 7.512980756573253e-06,
      "loss": 0.3106,
      "step": 3398
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.6507364136333686,
      "learning_rate": 7.498283821037033e-06,
      "loss": 0.2765,
      "step": 3399
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.8994906630753163,
      "learning_rate": 7.483599221357189e-06,
      "loss": 0.3717,
      "step": 3400
    },
    {
      "epoch": 3.89,
      "eval_blimp_filtered_avg": 0.7094029850746268,
      "eval_blimp_filtered_std": 0.005048776399611175,
      "step": 3400
    },
    {
      "epoch": 3.89,
      "eval_blimp_supplement_avg": 0.8318965517241379,
      "eval_blimp_supplement_std": 0.016785971458432806,
      "step": 3400
    },
    {
      "epoch": 3.89,
      "eval_vqa_filtered_avg": 0.55,
      "eval_vqa_filtered_std": 0.049999999999999996,
      "step": 3400
    },
    {
      "epoch": 3.89,
      "eval_winoground_filtered_avg": 0.63,
      "eval_winoground_filtered_std": 0.04852365870939099,
      "step": 3400
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.7618515291313033,
      "learning_rate": 7.468926965584099e-06,
      "loss": 0.3039,
      "step": 3401
    },
    {
      "epoch": 3.89,
      "grad_norm": 2.092130558627998,
      "learning_rate": 7.454267061761357e-06,
      "loss": 0.3718,
      "step": 3402
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.8026742238561475,
      "learning_rate": 7.4396195179258064e-06,
      "loss": 0.3175,
      "step": 3403
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.785898134687311,
      "learning_rate": 7.424984342107473e-06,
      "loss": 0.3177,
      "step": 3404
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.9713520429724871,
      "learning_rate": 7.4103615423296575e-06,
      "loss": 0.3103,
      "step": 3405
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.8137329934963284,
      "learning_rate": 7.3957511266088426e-06,
      "loss": 0.298,
      "step": 3406
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.9572263319617353,
      "learning_rate": 7.381153102954731e-06,
      "loss": 0.3874,
      "step": 3407
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.7750282976173533,
      "learning_rate": 7.366567479370216e-06,
      "loss": 0.3043,
      "step": 3408
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.6995435336675986,
      "learning_rate": 7.351994263851414e-06,
      "loss": 0.2956,
      "step": 3409
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.8629206080294956,
      "learning_rate": 7.337433464387632e-06,
      "loss": 0.3127,
      "step": 3410
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.6704195264085224,
      "learning_rate": 7.322885088961372e-06,
      "loss": 0.2733,
      "step": 3411
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.8762146765861518,
      "learning_rate": 7.308349145548301e-06,
      "loss": 0.3139,
      "step": 3412
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.592001018647559,
      "learning_rate": 7.2938256421173175e-06,
      "loss": 0.2935,
      "step": 3413
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.4756048403822417,
      "learning_rate": 7.279314586630464e-06,
      "loss": 0.2735,
      "step": 3414
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.6291676444406142,
      "learning_rate": 7.264815987042981e-06,
      "loss": 0.2553,
      "step": 3415
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.8886939660434943,
      "learning_rate": 7.250329851303245e-06,
      "loss": 0.343,
      "step": 3416
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.7876627724060412,
      "learning_rate": 7.235856187352838e-06,
      "loss": 0.2937,
      "step": 3417
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.4868815527888508,
      "learning_rate": 7.221395003126489e-06,
      "loss": 0.256,
      "step": 3418
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.8281293575303834,
      "learning_rate": 7.206946306552081e-06,
      "loss": 0.3537,
      "step": 3419
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.6637085324316998,
      "learning_rate": 7.192510105550664e-06,
      "loss": 0.3019,
      "step": 3420
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.9128650664706421,
      "learning_rate": 7.178086408036421e-06,
      "loss": 0.3406,
      "step": 3421
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.7505293495439256,
      "learning_rate": 7.1636752219167e-06,
      "loss": 0.3275,
      "step": 3422
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.979186586578943,
      "learning_rate": 7.149276555091973e-06,
      "loss": 0.3254,
      "step": 3423
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.8492187613295716,
      "learning_rate": 7.134890415455861e-06,
      "loss": 0.3008,
      "step": 3424
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.9402288585330325,
      "learning_rate": 7.120516810895102e-06,
      "loss": 0.3501,
      "step": 3425
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.8959172838257747,
      "learning_rate": 7.106155749289573e-06,
      "loss": 0.3263,
      "step": 3426
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.7898080101988312,
      "learning_rate": 7.091807238512277e-06,
      "loss": 0.3039,
      "step": 3427
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.8203516753950575,
      "learning_rate": 7.077471286429329e-06,
      "loss": 0.3266,
      "step": 3428
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.7732832821184341,
      "learning_rate": 7.063147900899969e-06,
      "loss": 0.2972,
      "step": 3429
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.8520698096691632,
      "learning_rate": 7.048837089776536e-06,
      "loss": 0.3598,
      "step": 3430
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.7712115722175286,
      "learning_rate": 7.034538860904482e-06,
      "loss": 0.3108,
      "step": 3431
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.8825957283626626,
      "learning_rate": 7.02025322212237e-06,
      "loss": 0.2935,
      "step": 3432
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.784924334909121,
      "learning_rate": 7.005980181261838e-06,
      "loss": 0.306,
      "step": 3433
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.5986633815920146,
      "learning_rate": 6.991719746147628e-06,
      "loss": 0.2984,
      "step": 3434
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.4637769136956478,
      "learning_rate": 6.977471924597585e-06,
      "loss": 0.2172,
      "step": 3435
    },
    {
      "epoch": 3.93,
      "grad_norm": 2.2607112977943213,
      "learning_rate": 6.9632367244226105e-06,
      "loss": 0.3888,
      "step": 3436
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.7980602588827554,
      "learning_rate": 6.9490141534267385e-06,
      "loss": 0.314,
      "step": 3437
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.6344524573713302,
      "learning_rate": 6.9348042194070114e-06,
      "loss": 0.2884,
      "step": 3438
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.9569577141386953,
      "learning_rate": 6.920606930153592e-06,
      "loss": 0.3321,
      "step": 3439
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.937711853444309,
      "learning_rate": 6.906422293449691e-06,
      "loss": 0.3525,
      "step": 3440
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.6214704495424461,
      "learning_rate": 6.892250317071604e-06,
      "loss": 0.2918,
      "step": 3441
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.6163076872190076,
      "learning_rate": 6.878091008788648e-06,
      "loss": 0.3082,
      "step": 3442
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.8781528169572232,
      "learning_rate": 6.8639443763632235e-06,
      "loss": 0.3125,
      "step": 3443
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.7431956471913117,
      "learning_rate": 6.849810427550774e-06,
      "loss": 0.3385,
      "step": 3444
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.7737074958766668,
      "learning_rate": 6.835689170099812e-06,
      "loss": 0.2994,
      "step": 3445
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.7137669817501675,
      "learning_rate": 6.821580611751849e-06,
      "loss": 0.2904,
      "step": 3446
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.8815783943163773,
      "learning_rate": 6.807484760241461e-06,
      "loss": 0.2941,
      "step": 3447
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.8787214961985061,
      "learning_rate": 6.793401623296258e-06,
      "loss": 0.2936,
      "step": 3448
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.9672291642814772,
      "learning_rate": 6.779331208636883e-06,
      "loss": 0.3537,
      "step": 3449
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.7504420490060781,
      "learning_rate": 6.765273523976982e-06,
      "loss": 0.3458,
      "step": 3450
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.8582479330550994,
      "learning_rate": 6.751228577023239e-06,
      "loss": 0.2926,
      "step": 3451
    },
    {
      "epoch": 3.95,
      "grad_norm": 1.6118511377326696,
      "learning_rate": 6.737196375475362e-06,
      "loss": 0.3084,
      "step": 3452
    },
    {
      "epoch": 3.95,
      "grad_norm": 1.9935370681764006,
      "learning_rate": 6.7231769270260625e-06,
      "loss": 0.3019,
      "step": 3453
    },
    {
      "epoch": 3.95,
      "grad_norm": 1.7753847884572385,
      "learning_rate": 6.709170239361066e-06,
      "loss": 0.2949,
      "step": 3454
    },
    {
      "epoch": 3.95,
      "grad_norm": 1.8277893397148324,
      "learning_rate": 6.6951763201590825e-06,
      "loss": 0.3228,
      "step": 3455
    },
    {
      "epoch": 3.95,
      "grad_norm": 1.6213430181477921,
      "learning_rate": 6.681195177091842e-06,
      "loss": 0.2636,
      "step": 3456
    },
    {
      "epoch": 3.95,
      "grad_norm": 1.7283844810111206,
      "learning_rate": 6.667226817824068e-06,
      "loss": 0.2924,
      "step": 3457
    },
    {
      "epoch": 3.95,
      "grad_norm": 1.6756504296209052,
      "learning_rate": 6.653271250013474e-06,
      "loss": 0.2593,
      "step": 3458
    },
    {
      "epoch": 3.95,
      "grad_norm": 1.9883111382342826,
      "learning_rate": 6.63932848131076e-06,
      "loss": 0.3189,
      "step": 3459
    },
    {
      "epoch": 3.95,
      "grad_norm": 1.8931177035656557,
      "learning_rate": 6.625398519359606e-06,
      "loss": 0.3397,
      "step": 3460
    },
    {
      "epoch": 3.96,
      "grad_norm": 1.6364210876388272,
      "learning_rate": 6.611481371796675e-06,
      "loss": 0.2726,
      "step": 3461
    },
    {
      "epoch": 3.96,
      "grad_norm": 1.937854459605809,
      "learning_rate": 6.597577046251618e-06,
      "loss": 0.3595,
      "step": 3462
    },
    {
      "epoch": 3.96,
      "grad_norm": 1.8338003291955232,
      "learning_rate": 6.583685550347018e-06,
      "loss": 0.2923,
      "step": 3463
    },
    {
      "epoch": 3.96,
      "grad_norm": 1.9241650068452867,
      "learning_rate": 6.569806891698467e-06,
      "loss": 0.336,
      "step": 3464
    },
    {
      "epoch": 3.96,
      "grad_norm": 2.055210161676895,
      "learning_rate": 6.555941077914493e-06,
      "loss": 0.3773,
      "step": 3465
    },
    {
      "epoch": 3.96,
      "grad_norm": 1.5166084930593196,
      "learning_rate": 6.5420881165965984e-06,
      "loss": 0.2523,
      "step": 3466
    },
    {
      "epoch": 3.96,
      "grad_norm": 1.8396661100677558,
      "learning_rate": 6.528248015339226e-06,
      "loss": 0.3117,
      "step": 3467
    },
    {
      "epoch": 3.96,
      "grad_norm": 1.6100396095559388,
      "learning_rate": 6.514420781729782e-06,
      "loss": 0.2554,
      "step": 3468
    },
    {
      "epoch": 3.96,
      "grad_norm": 1.7977774683851462,
      "learning_rate": 6.500606423348604e-06,
      "loss": 0.3275,
      "step": 3469
    },
    {
      "epoch": 3.97,
      "grad_norm": 1.7685125318761408,
      "learning_rate": 6.486804947768985e-06,
      "loss": 0.2841,
      "step": 3470
    },
    {
      "epoch": 3.97,
      "grad_norm": 1.8446781725131896,
      "learning_rate": 6.473016362557156e-06,
      "loss": 0.3467,
      "step": 3471
    },
    {
      "epoch": 3.97,
      "grad_norm": 1.7959760845581787,
      "learning_rate": 6.4592406752722535e-06,
      "loss": 0.3181,
      "step": 3472
    },
    {
      "epoch": 3.97,
      "grad_norm": 1.9130164025794356,
      "learning_rate": 6.445477893466373e-06,
      "loss": 0.3341,
      "step": 3473
    },
    {
      "epoch": 3.97,
      "grad_norm": 1.9356076705281944,
      "learning_rate": 6.431728024684525e-06,
      "loss": 0.2896,
      "step": 3474
    },
    {
      "epoch": 3.97,
      "grad_norm": 1.546134425891044,
      "learning_rate": 6.417991076464649e-06,
      "loss": 0.2593,
      "step": 3475
    },
    {
      "epoch": 3.97,
      "grad_norm": 1.960927045817758,
      "learning_rate": 6.404267056337585e-06,
      "loss": 0.3452,
      "step": 3476
    },
    {
      "epoch": 3.97,
      "grad_norm": 1.9633250549699468,
      "learning_rate": 6.390555971827095e-06,
      "loss": 0.3663,
      "step": 3477
    },
    {
      "epoch": 3.97,
      "grad_norm": 2.0037394725715987,
      "learning_rate": 6.3768578304498546e-06,
      "loss": 0.3313,
      "step": 3478
    },
    {
      "epoch": 3.98,
      "grad_norm": 1.7054557755125508,
      "learning_rate": 6.363172639715442e-06,
      "loss": 0.2733,
      "step": 3479
    },
    {
      "epoch": 3.98,
      "grad_norm": 1.8737314662220688,
      "learning_rate": 6.349500407126313e-06,
      "loss": 0.3223,
      "step": 3480
    },
    {
      "epoch": 3.98,
      "grad_norm": 2.0425939525670853,
      "learning_rate": 6.335841140177852e-06,
      "loss": 0.3968,
      "step": 3481
    },
    {
      "epoch": 3.98,
      "grad_norm": 1.9586147183362377,
      "learning_rate": 6.322194846358318e-06,
      "loss": 0.3688,
      "step": 3482
    },
    {
      "epoch": 3.98,
      "grad_norm": 1.892454556404934,
      "learning_rate": 6.308561533148861e-06,
      "loss": 0.3196,
      "step": 3483
    },
    {
      "epoch": 3.98,
      "grad_norm": 1.9569765052860406,
      "learning_rate": 6.294941208023515e-06,
      "loss": 0.3129,
      "step": 3484
    },
    {
      "epoch": 3.98,
      "grad_norm": 1.587947449232117,
      "learning_rate": 6.281333878449192e-06,
      "loss": 0.2867,
      "step": 3485
    },
    {
      "epoch": 3.98,
      "grad_norm": 1.743887267985296,
      "learning_rate": 6.2677395518856874e-06,
      "loss": 0.304,
      "step": 3486
    },
    {
      "epoch": 3.99,
      "grad_norm": 1.9497636285484239,
      "learning_rate": 6.254158235785668e-06,
      "loss": 0.3341,
      "step": 3487
    },
    {
      "epoch": 3.99,
      "grad_norm": 1.6893767726440567,
      "learning_rate": 6.240589937594641e-06,
      "loss": 0.3119,
      "step": 3488
    },
    {
      "epoch": 3.99,
      "grad_norm": 1.9040879146333112,
      "learning_rate": 6.22703466475101e-06,
      "loss": 0.3083,
      "step": 3489
    },
    {
      "epoch": 3.99,
      "grad_norm": 1.9044715460422965,
      "learning_rate": 6.213492424686029e-06,
      "loss": 0.2901,
      "step": 3490
    },
    {
      "epoch": 3.99,
      "grad_norm": 1.972884174249025,
      "learning_rate": 6.199963224823789e-06,
      "loss": 0.3566,
      "step": 3491
    },
    {
      "epoch": 3.99,
      "grad_norm": 1.9315055880246064,
      "learning_rate": 6.1864470725812775e-06,
      "loss": 0.3491,
      "step": 3492
    },
    {
      "epoch": 3.99,
      "grad_norm": 1.940132208605843,
      "learning_rate": 6.172943975368269e-06,
      "loss": 0.3572,
      "step": 3493
    },
    {
      "epoch": 3.99,
      "grad_norm": 1.8008944154374111,
      "learning_rate": 6.1594539405874235e-06,
      "loss": 0.3501,
      "step": 3494
    },
    {
      "epoch": 3.99,
      "grad_norm": 1.8488065707788055,
      "learning_rate": 6.145976975634229e-06,
      "loss": 0.3048,
      "step": 3495
    },
    {
      "epoch": 4.0,
      "grad_norm": 1.690093186572164,
      "learning_rate": 6.132513087897007e-06,
      "loss": 0.2683,
      "step": 3496
    },
    {
      "epoch": 4.0,
      "grad_norm": 1.6421620646761677,
      "learning_rate": 6.119062284756898e-06,
      "loss": 0.2788,
      "step": 3497
    },
    {
      "epoch": 4.0,
      "grad_norm": 1.7249829329413884,
      "learning_rate": 6.105624573587882e-06,
      "loss": 0.3121,
      "step": 3498
    },
    {
      "epoch": 4.0,
      "grad_norm": 1.87503118048506,
      "learning_rate": 6.092199961756775e-06,
      "loss": 0.3432,
      "step": 3499
    },
    {
      "epoch": 4.0,
      "grad_norm": 1.413388030268998,
      "learning_rate": 6.0787884566231975e-06,
      "loss": 0.2699,
      "step": 3500
    },
    {
      "epoch": 4.0,
      "eval_blimp_filtered_avg": 0.7101492537313433,
      "eval_blimp_filtered_std": 0.005044875139196119,
      "step": 3500
    },
    {
      "epoch": 4.0,
      "eval_blimp_supplement_avg": 0.8318965517241379,
      "eval_blimp_supplement_std": 0.0167900793369906,
      "step": 3500
    },
    {
      "epoch": 4.0,
      "eval_vqa_filtered_avg": 0.55,
      "eval_vqa_filtered_std": 0.049999999999999996,
      "step": 3500
    },
    {
      "epoch": 4.0,
      "eval_winoground_filtered_avg": 0.63,
      "eval_winoground_filtered_std": 0.04852365870939099,
      "step": 3500
    },
    {
      "epoch": 4.0,
      "eval_blimp_filtered_avg": 0.7101492537313433,
      "eval_blimp_filtered_std": 0.005044875139196119,
      "step": 3500
    },
    {
      "epoch": 4.0,
      "eval_blimp_supplement_avg": 0.8318965517241379,
      "eval_blimp_supplement_std": 0.0167900793369906,
      "step": 3500
    },
    {
      "epoch": 4.0,
      "eval_vqa_filtered_avg": 0.55,
      "eval_vqa_filtered_std": 0.049999999999999996,
      "step": 3500
    },
    {
      "epoch": 4.0,
      "eval_winoground_filtered_avg": 0.63,
      "eval_winoground_filtered_std": 0.04852365870939099,
      "step": 3500
    },
    {
      "epoch": 4.0,
      "grad_norm": 1.4094871818955625,
      "learning_rate": 6.065390065539567e-06,
      "loss": 0.2454,
      "step": 3501
    },
    {
      "epoch": 4.0,
      "grad_norm": 1.3258415116262614,
      "learning_rate": 6.052004795851136e-06,
      "loss": 0.1787,
      "step": 3502
    },
    {
      "epoch": 4.0,
      "grad_norm": 1.0662886750334313,
      "learning_rate": 6.0386326548959594e-06,
      "loss": 0.1778,
      "step": 3503
    },
    {
      "epoch": 4.0,
      "grad_norm": 1.2924931015052112,
      "learning_rate": 6.025273650004899e-06,
      "loss": 0.1964,
      "step": 3504
    },
    {
      "epoch": 4.01,
      "grad_norm": 1.2581087572832568,
      "learning_rate": 6.011927788501584e-06,
      "loss": 0.2008,
      "step": 3505
    },
    {
      "epoch": 4.01,
      "grad_norm": 1.321300919929928,
      "learning_rate": 5.998595077702484e-06,
      "loss": 0.1868,
      "step": 3506
    },
    {
      "epoch": 4.01,
      "grad_norm": 1.404726818109712,
      "learning_rate": 5.9852755249168325e-06,
      "loss": 0.235,
      "step": 3507
    },
    {
      "epoch": 4.01,
      "grad_norm": 1.3183089667778218,
      "learning_rate": 5.971969137446652e-06,
      "loss": 0.1759,
      "step": 3508
    },
    {
      "epoch": 4.01,
      "grad_norm": 1.3314471289772236,
      "learning_rate": 5.958675922586762e-06,
      "loss": 0.2025,
      "step": 3509
    },
    {
      "epoch": 4.01,
      "grad_norm": 1.4706631547933326,
      "learning_rate": 5.945395887624724e-06,
      "loss": 0.1784,
      "step": 3510
    },
    {
      "epoch": 4.01,
      "grad_norm": 1.4926656372423883,
      "learning_rate": 5.932129039840916e-06,
      "loss": 0.2264,
      "step": 3511
    },
    {
      "epoch": 4.01,
      "grad_norm": 1.428557767021507,
      "learning_rate": 5.9188753865084625e-06,
      "loss": 0.173,
      "step": 3512
    },
    {
      "epoch": 4.01,
      "grad_norm": 1.3487067561127695,
      "learning_rate": 5.9056349348932646e-06,
      "loss": 0.1773,
      "step": 3513
    },
    {
      "epoch": 4.02,
      "grad_norm": 1.5544468839529828,
      "learning_rate": 5.89240769225398e-06,
      "loss": 0.2169,
      "step": 3514
    },
    {
      "epoch": 4.02,
      "grad_norm": 1.3346235146277545,
      "learning_rate": 5.8791936658420284e-06,
      "loss": 0.1467,
      "step": 3515
    },
    {
      "epoch": 4.02,
      "grad_norm": 1.6153252717956137,
      "learning_rate": 5.865992862901581e-06,
      "loss": 0.1799,
      "step": 3516
    },
    {
      "epoch": 4.02,
      "grad_norm": 1.5092750512405573,
      "learning_rate": 5.852805290669576e-06,
      "loss": 0.1699,
      "step": 3517
    },
    {
      "epoch": 4.02,
      "grad_norm": 1.7312704373976617,
      "learning_rate": 5.83963095637566e-06,
      "loss": 0.1937,
      "step": 3518
    },
    {
      "epoch": 4.02,
      "grad_norm": 1.6227404780825543,
      "learning_rate": 5.8264698672422625e-06,
      "loss": 0.1899,
      "step": 3519
    },
    {
      "epoch": 4.02,
      "grad_norm": 1.6339053485481863,
      "learning_rate": 5.813322030484528e-06,
      "loss": 0.1843,
      "step": 3520
    },
    {
      "epoch": 4.02,
      "grad_norm": 1.8638782275529633,
      "learning_rate": 5.800187453310348e-06,
      "loss": 0.2038,
      "step": 3521
    },
    {
      "epoch": 4.03,
      "grad_norm": 1.618093227185989,
      "learning_rate": 5.78706614292034e-06,
      "loss": 0.1684,
      "step": 3522
    },
    {
      "epoch": 4.03,
      "grad_norm": 1.6279628000408566,
      "learning_rate": 5.773958106507845e-06,
      "loss": 0.1646,
      "step": 3523
    },
    {
      "epoch": 4.03,
      "grad_norm": 1.7863415913162264,
      "learning_rate": 5.760863351258936e-06,
      "loss": 0.2063,
      "step": 3524
    },
    {
      "epoch": 4.03,
      "grad_norm": 1.8782038275663961,
      "learning_rate": 5.747781884352396e-06,
      "loss": 0.2063,
      "step": 3525
    },
    {
      "epoch": 4.03,
      "grad_norm": 1.6260967011671466,
      "learning_rate": 5.734713712959737e-06,
      "loss": 0.1915,
      "step": 3526
    },
    {
      "epoch": 4.03,
      "grad_norm": 1.620435875137579,
      "learning_rate": 5.721658844245153e-06,
      "loss": 0.1833,
      "step": 3527
    },
    {
      "epoch": 4.03,
      "grad_norm": 1.7271857919241373,
      "learning_rate": 5.708617285365574e-06,
      "loss": 0.2044,
      "step": 3528
    },
    {
      "epoch": 4.03,
      "grad_norm": 1.879995862075944,
      "learning_rate": 5.695589043470626e-06,
      "loss": 0.2137,
      "step": 3529
    },
    {
      "epoch": 4.03,
      "grad_norm": 1.7001134010105838,
      "learning_rate": 5.682574125702626e-06,
      "loss": 0.1745,
      "step": 3530
    },
    {
      "epoch": 4.04,
      "grad_norm": 1.7852181031287104,
      "learning_rate": 5.669572539196598e-06,
      "loss": 0.2042,
      "step": 3531
    },
    {
      "epoch": 4.04,
      "grad_norm": 1.5736982573033498,
      "learning_rate": 5.656584291080249e-06,
      "loss": 0.1761,
      "step": 3532
    },
    {
      "epoch": 4.04,
      "grad_norm": 1.26835017016526,
      "learning_rate": 5.643609388473974e-06,
      "loss": 0.1373,
      "step": 3533
    },
    {
      "epoch": 4.04,
      "grad_norm": 1.6398262885750539,
      "learning_rate": 5.630647838490869e-06,
      "loss": 0.1955,
      "step": 3534
    },
    {
      "epoch": 4.04,
      "grad_norm": 1.5210868086938862,
      "learning_rate": 5.61769964823667e-06,
      "loss": 0.184,
      "step": 3535
    },
    {
      "epoch": 4.04,
      "grad_norm": 1.448907203017577,
      "learning_rate": 5.604764824809831e-06,
      "loss": 0.1573,
      "step": 3536
    },
    {
      "epoch": 4.04,
      "grad_norm": 1.8169383115784847,
      "learning_rate": 5.591843375301456e-06,
      "loss": 0.1931,
      "step": 3537
    },
    {
      "epoch": 4.04,
      "grad_norm": 1.2899954969996323,
      "learning_rate": 5.578935306795325e-06,
      "loss": 0.1463,
      "step": 3538
    },
    {
      "epoch": 4.04,
      "grad_norm": 1.7755598313883996,
      "learning_rate": 5.566040626367876e-06,
      "loss": 0.2102,
      "step": 3539
    },
    {
      "epoch": 4.05,
      "grad_norm": 1.5727927201698568,
      "learning_rate": 5.553159341088212e-06,
      "loss": 0.1527,
      "step": 3540
    },
    {
      "epoch": 4.05,
      "grad_norm": 1.6852800589827022,
      "learning_rate": 5.540291458018096e-06,
      "loss": 0.1959,
      "step": 3541
    },
    {
      "epoch": 4.05,
      "grad_norm": 1.742035213962156,
      "learning_rate": 5.527436984211933e-06,
      "loss": 0.2172,
      "step": 3542
    },
    {
      "epoch": 4.05,
      "grad_norm": 1.6335180773242286,
      "learning_rate": 5.514595926716795e-06,
      "loss": 0.2186,
      "step": 3543
    },
    {
      "epoch": 4.05,
      "grad_norm": 1.5285596445847476,
      "learning_rate": 5.501768292572369e-06,
      "loss": 0.1721,
      "step": 3544
    },
    {
      "epoch": 4.05,
      "grad_norm": 1.4354258026612206,
      "learning_rate": 5.488954088811002e-06,
      "loss": 0.1605,
      "step": 3545
    },
    {
      "epoch": 4.05,
      "grad_norm": 1.4117859885443333,
      "learning_rate": 5.47615332245769e-06,
      "loss": 0.1552,
      "step": 3546
    },
    {
      "epoch": 4.05,
      "grad_norm": 1.8025505903460413,
      "learning_rate": 5.463366000530054e-06,
      "loss": 0.1912,
      "step": 3547
    },
    {
      "epoch": 4.05,
      "grad_norm": 1.4582679089918311,
      "learning_rate": 5.450592130038321e-06,
      "loss": 0.1717,
      "step": 3548
    },
    {
      "epoch": 4.06,
      "grad_norm": 1.7858765035965334,
      "learning_rate": 5.437831717985374e-06,
      "loss": 0.1937,
      "step": 3549
    },
    {
      "epoch": 4.06,
      "grad_norm": 1.6041094027473113,
      "learning_rate": 5.425084771366702e-06,
      "loss": 0.1909,
      "step": 3550
    },
    {
      "epoch": 4.06,
      "grad_norm": 1.5401748335549221,
      "learning_rate": 5.412351297170427e-06,
      "loss": 0.1774,
      "step": 3551
    },
    {
      "epoch": 4.06,
      "grad_norm": 1.5641555309017636,
      "learning_rate": 5.3996313023772526e-06,
      "loss": 0.2051,
      "step": 3552
    },
    {
      "epoch": 4.06,
      "grad_norm": 1.7219742167726084,
      "learning_rate": 5.386924793960533e-06,
      "loss": 0.186,
      "step": 3553
    },
    {
      "epoch": 4.06,
      "grad_norm": 1.4822687858844275,
      "learning_rate": 5.374231778886212e-06,
      "loss": 0.1707,
      "step": 3554
    },
    {
      "epoch": 4.06,
      "grad_norm": 1.633999871324436,
      "learning_rate": 5.361552264112833e-06,
      "loss": 0.1905,
      "step": 3555
    },
    {
      "epoch": 4.06,
      "grad_norm": 1.9160259717578692,
      "learning_rate": 5.348886256591532e-06,
      "loss": 0.1915,
      "step": 3556
    },
    {
      "epoch": 4.07,
      "grad_norm": 1.5980298921172533,
      "learning_rate": 5.336233763266053e-06,
      "loss": 0.1825,
      "step": 3557
    },
    {
      "epoch": 4.07,
      "grad_norm": 1.579930785637522,
      "learning_rate": 5.323594791072724e-06,
      "loss": 0.1925,
      "step": 3558
    },
    {
      "epoch": 4.07,
      "grad_norm": 1.7693025926522976,
      "learning_rate": 5.3109693469404675e-06,
      "loss": 0.2084,
      "step": 3559
    },
    {
      "epoch": 4.07,
      "grad_norm": 1.3677667914406104,
      "learning_rate": 5.298357437790782e-06,
      "loss": 0.1386,
      "step": 3560
    },
    {
      "epoch": 4.07,
      "grad_norm": 1.7085956148278254,
      "learning_rate": 5.28575907053775e-06,
      "loss": 0.1936,
      "step": 3561
    },
    {
      "epoch": 4.07,
      "grad_norm": 1.6945284511994316,
      "learning_rate": 5.273174252088031e-06,
      "loss": 0.203,
      "step": 3562
    },
    {
      "epoch": 4.07,
      "grad_norm": 1.3937994594296597,
      "learning_rate": 5.260602989340849e-06,
      "loss": 0.1516,
      "step": 3563
    },
    {
      "epoch": 4.07,
      "grad_norm": 1.9243314128859699,
      "learning_rate": 5.248045289188018e-06,
      "loss": 0.1906,
      "step": 3564
    },
    {
      "epoch": 4.07,
      "grad_norm": 1.337048595697058,
      "learning_rate": 5.235501158513882e-06,
      "loss": 0.136,
      "step": 3565
    },
    {
      "epoch": 4.08,
      "grad_norm": 1.6355684645390103,
      "learning_rate": 5.222970604195371e-06,
      "loss": 0.1844,
      "step": 3566
    },
    {
      "epoch": 4.08,
      "grad_norm": 1.7333959458075274,
      "learning_rate": 5.210453633101968e-06,
      "loss": 0.172,
      "step": 3567
    },
    {
      "epoch": 4.08,
      "grad_norm": 1.8851461457124004,
      "learning_rate": 5.197950252095711e-06,
      "loss": 0.2363,
      "step": 3568
    },
    {
      "epoch": 4.08,
      "grad_norm": 2.0628337677867057,
      "learning_rate": 5.185460468031177e-06,
      "loss": 0.1886,
      "step": 3569
    },
    {
      "epoch": 4.08,
      "grad_norm": 1.7036147309390606,
      "learning_rate": 5.172984287755505e-06,
      "loss": 0.1648,
      "step": 3570
    },
    {
      "epoch": 4.08,
      "grad_norm": 1.7733361927029907,
      "learning_rate": 5.1605217181083594e-06,
      "loss": 0.203,
      "step": 3571
    },
    {
      "epoch": 4.08,
      "grad_norm": 1.597242555028738,
      "learning_rate": 5.148072765921961e-06,
      "loss": 0.1894,
      "step": 3572
    },
    {
      "epoch": 4.08,
      "grad_norm": 1.7859524971301792,
      "learning_rate": 5.135637438021045e-06,
      "loss": 0.1991,
      "step": 3573
    },
    {
      "epoch": 4.08,
      "grad_norm": 1.5723384771294824,
      "learning_rate": 5.123215741222889e-06,
      "loss": 0.1777,
      "step": 3574
    },
    {
      "epoch": 4.09,
      "grad_norm": 1.8757076897224898,
      "learning_rate": 5.110807682337302e-06,
      "loss": 0.2226,
      "step": 3575
    },
    {
      "epoch": 4.09,
      "grad_norm": 1.6817326323274155,
      "learning_rate": 5.098413268166606e-06,
      "loss": 0.175,
      "step": 3576
    },
    {
      "epoch": 4.09,
      "grad_norm": 1.9038180438633192,
      "learning_rate": 5.08603250550565e-06,
      "loss": 0.1823,
      "step": 3577
    },
    {
      "epoch": 4.09,
      "grad_norm": 1.8590776403321487,
      "learning_rate": 5.073665401141797e-06,
      "loss": 0.161,
      "step": 3578
    },
    {
      "epoch": 4.09,
      "grad_norm": 1.5601872605914089,
      "learning_rate": 5.0613119618549246e-06,
      "loss": 0.1954,
      "step": 3579
    },
    {
      "epoch": 4.09,
      "grad_norm": 1.507114107178623,
      "learning_rate": 5.04897219441741e-06,
      "loss": 0.1694,
      "step": 3580
    },
    {
      "epoch": 4.09,
      "grad_norm": 1.1817179179387935,
      "learning_rate": 5.036646105594155e-06,
      "loss": 0.1372,
      "step": 3581
    },
    {
      "epoch": 4.09,
      "grad_norm": 1.794102072145093,
      "learning_rate": 5.024333702142532e-06,
      "loss": 0.2089,
      "step": 3582
    },
    {
      "epoch": 4.09,
      "grad_norm": 1.5816938906785,
      "learning_rate": 5.012034990812432e-06,
      "loss": 0.1849,
      "step": 3583
    },
    {
      "epoch": 4.1,
      "grad_norm": 1.4861820057026012,
      "learning_rate": 4.999749978346235e-06,
      "loss": 0.1968,
      "step": 3584
    },
    {
      "epoch": 4.1,
      "grad_norm": 1.5866374901927756,
      "learning_rate": 4.9874786714788315e-06,
      "loss": 0.1711,
      "step": 3585
    },
    {
      "epoch": 4.1,
      "grad_norm": 1.3639445903575593,
      "learning_rate": 4.975221076937552e-06,
      "loss": 0.1488,
      "step": 3586
    },
    {
      "epoch": 4.1,
      "grad_norm": 1.6507500039543592,
      "learning_rate": 4.962977201442248e-06,
      "loss": 0.1829,
      "step": 3587
    },
    {
      "epoch": 4.1,
      "grad_norm": 1.4267077711987637,
      "learning_rate": 4.9507470517052375e-06,
      "loss": 0.1588,
      "step": 3588
    },
    {
      "epoch": 4.1,
      "grad_norm": 1.8123956030105113,
      "learning_rate": 4.938530634431321e-06,
      "loss": 0.2166,
      "step": 3589
    },
    {
      "epoch": 4.1,
      "grad_norm": 1.6848885824116249,
      "learning_rate": 4.92632795631775e-06,
      "loss": 0.1825,
      "step": 3590
    },
    {
      "epoch": 4.1,
      "grad_norm": 1.5301445371883051,
      "learning_rate": 4.914139024054267e-06,
      "loss": 0.1754,
      "step": 3591
    },
    {
      "epoch": 4.11,
      "grad_norm": 1.4568150805023978,
      "learning_rate": 4.901963844323057e-06,
      "loss": 0.1715,
      "step": 3592
    },
    {
      "epoch": 4.11,
      "grad_norm": 1.4996109015925156,
      "learning_rate": 4.889802423798808e-06,
      "loss": 0.1736,
      "step": 3593
    },
    {
      "epoch": 4.11,
      "grad_norm": 1.5919853091241871,
      "learning_rate": 4.8776547691486065e-06,
      "loss": 0.1833,
      "step": 3594
    },
    {
      "epoch": 4.11,
      "grad_norm": 1.7904404597677657,
      "learning_rate": 4.865520887032034e-06,
      "loss": 0.1983,
      "step": 3595
    },
    {
      "epoch": 4.11,
      "grad_norm": 1.6243120509446152,
      "learning_rate": 4.8534007841011055e-06,
      "loss": 0.1877,
      "step": 3596
    },
    {
      "epoch": 4.11,
      "grad_norm": 1.6210411900325714,
      "learning_rate": 4.841294467000284e-06,
      "loss": 0.1949,
      "step": 3597
    },
    {
      "epoch": 4.11,
      "grad_norm": 1.7338374276828903,
      "learning_rate": 4.829201942366483e-06,
      "loss": 0.1959,
      "step": 3598
    },
    {
      "epoch": 4.11,
      "grad_norm": 1.258629890547153,
      "learning_rate": 4.817123216829032e-06,
      "loss": 0.1461,
      "step": 3599
    },
    {
      "epoch": 4.11,
      "grad_norm": 1.5572531663250824,
      "learning_rate": 4.8050582970097305e-06,
      "loss": 0.1823,
      "step": 3600
    },
    {
      "epoch": 4.11,
      "eval_blimp_filtered_avg": 0.7091044776119403,
      "eval_blimp_filtered_std": 0.005078239417928012,
      "step": 3600
    },
    {
      "epoch": 4.11,
      "eval_blimp_supplement_avg": 0.8297413793103449,
      "eval_blimp_supplement_std": 0.017002666619928053,
      "step": 3600
    },
    {
      "epoch": 4.11,
      "eval_vqa_filtered_avg": 0.54,
      "eval_vqa_filtered_std": 0.05009082659620333,
      "step": 3600
    },
    {
      "epoch": 4.11,
      "eval_winoground_filtered_avg": 0.64,
      "eval_winoground_filtered_std": 0.04824181513244218,
      "step": 3600
    },
    {
      "epoch": 4.12,
      "grad_norm": 1.4123107950506073,
      "learning_rate": 4.793007189522779e-06,
      "loss": 0.1501,
      "step": 3601
    },
    {
      "epoch": 4.12,
      "grad_norm": 1.65026698389643,
      "learning_rate": 4.780969900974831e-06,
      "loss": 0.1617,
      "step": 3602
    },
    {
      "epoch": 4.12,
      "grad_norm": 1.7853922783339735,
      "learning_rate": 4.768946437964931e-06,
      "loss": 0.197,
      "step": 3603
    },
    {
      "epoch": 4.12,
      "grad_norm": 1.5803781590063015,
      "learning_rate": 4.756936807084569e-06,
      "loss": 0.1637,
      "step": 3604
    },
    {
      "epoch": 4.12,
      "grad_norm": 1.8301363507650314,
      "learning_rate": 4.744941014917652e-06,
      "loss": 0.2107,
      "step": 3605
    },
    {
      "epoch": 4.12,
      "grad_norm": 1.434956507492784,
      "learning_rate": 4.7329590680405e-06,
      "loss": 0.1625,
      "step": 3606
    },
    {
      "epoch": 4.12,
      "grad_norm": 1.7654568289478965,
      "learning_rate": 4.720990973021815e-06,
      "loss": 0.1699,
      "step": 3607
    },
    {
      "epoch": 4.12,
      "grad_norm": 1.7592479972998953,
      "learning_rate": 4.709036736422751e-06,
      "loss": 0.1991,
      "step": 3608
    },
    {
      "epoch": 4.12,
      "grad_norm": 1.6014504052513352,
      "learning_rate": 4.697096364796831e-06,
      "loss": 0.1564,
      "step": 3609
    },
    {
      "epoch": 4.13,
      "grad_norm": 1.6435006747491192,
      "learning_rate": 4.685169864689998e-06,
      "loss": 0.2052,
      "step": 3610
    },
    {
      "epoch": 4.13,
      "grad_norm": 1.4115248361875512,
      "learning_rate": 4.673257242640564e-06,
      "loss": 0.1405,
      "step": 3611
    },
    {
      "epoch": 4.13,
      "grad_norm": 1.5514296990111007,
      "learning_rate": 4.661358505179253e-06,
      "loss": 0.1786,
      "step": 3612
    },
    {
      "epoch": 4.13,
      "grad_norm": 1.6255426155308694,
      "learning_rate": 4.649473658829177e-06,
      "loss": 0.1725,
      "step": 3613
    },
    {
      "epoch": 4.13,
      "grad_norm": 1.70637631260215,
      "learning_rate": 4.637602710105826e-06,
      "loss": 0.1933,
      "step": 3614
    },
    {
      "epoch": 4.13,
      "grad_norm": 1.6544008178823197,
      "learning_rate": 4.6257456655170725e-06,
      "loss": 0.1512,
      "step": 3615
    },
    {
      "epoch": 4.13,
      "grad_norm": 1.607028232605771,
      "learning_rate": 4.613902531563167e-06,
      "loss": 0.1641,
      "step": 3616
    },
    {
      "epoch": 4.13,
      "grad_norm": 1.6539753162866468,
      "learning_rate": 4.602073314736738e-06,
      "loss": 0.1846,
      "step": 3617
    },
    {
      "epoch": 4.13,
      "grad_norm": 1.5680546782222142,
      "learning_rate": 4.5902580215227786e-06,
      "loss": 0.162,
      "step": 3618
    },
    {
      "epoch": 4.14,
      "grad_norm": 1.6791550721699193,
      "learning_rate": 4.578456658398654e-06,
      "loss": 0.1795,
      "step": 3619
    },
    {
      "epoch": 4.14,
      "grad_norm": 1.552428639348701,
      "learning_rate": 4.566669231834083e-06,
      "loss": 0.1472,
      "step": 3620
    },
    {
      "epoch": 4.14,
      "grad_norm": 1.5735289129325183,
      "learning_rate": 4.554895748291149e-06,
      "loss": 0.1544,
      "step": 3621
    },
    {
      "epoch": 4.14,
      "grad_norm": 1.9520307203887282,
      "learning_rate": 4.5431362142242994e-06,
      "loss": 0.1948,
      "step": 3622
    },
    {
      "epoch": 4.14,
      "grad_norm": 1.8191828775085097,
      "learning_rate": 4.531390636080327e-06,
      "loss": 0.213,
      "step": 3623
    },
    {
      "epoch": 4.14,
      "grad_norm": 1.6478678764453822,
      "learning_rate": 4.519659020298371e-06,
      "loss": 0.168,
      "step": 3624
    },
    {
      "epoch": 4.14,
      "grad_norm": 1.8661437680136521,
      "learning_rate": 4.507941373309927e-06,
      "loss": 0.1977,
      "step": 3625
    },
    {
      "epoch": 4.14,
      "grad_norm": 1.647690598783053,
      "learning_rate": 4.496237701538817e-06,
      "loss": 0.1712,
      "step": 3626
    },
    {
      "epoch": 4.15,
      "grad_norm": 1.752006370298568,
      "learning_rate": 4.484548011401221e-06,
      "loss": 0.214,
      "step": 3627
    },
    {
      "epoch": 4.15,
      "grad_norm": 1.5726165124498837,
      "learning_rate": 4.472872309305633e-06,
      "loss": 0.1589,
      "step": 3628
    },
    {
      "epoch": 4.15,
      "grad_norm": 1.53962857141423,
      "learning_rate": 4.461210601652885e-06,
      "loss": 0.1615,
      "step": 3629
    },
    {
      "epoch": 4.15,
      "grad_norm": 1.678449878750355,
      "learning_rate": 4.449562894836152e-06,
      "loss": 0.2043,
      "step": 3630
    },
    {
      "epoch": 4.15,
      "grad_norm": 1.6517613710066799,
      "learning_rate": 4.4379291952409085e-06,
      "loss": 0.1791,
      "step": 3631
    },
    {
      "epoch": 4.15,
      "grad_norm": 1.2403424242735617,
      "learning_rate": 4.426309509244986e-06,
      "loss": 0.1596,
      "step": 3632
    },
    {
      "epoch": 4.15,
      "grad_norm": 1.6843365506839507,
      "learning_rate": 4.414703843218491e-06,
      "loss": 0.1638,
      "step": 3633
    },
    {
      "epoch": 4.15,
      "grad_norm": 1.568488850640214,
      "learning_rate": 4.403112203523867e-06,
      "loss": 0.178,
      "step": 3634
    },
    {
      "epoch": 4.15,
      "grad_norm": 1.4125632257248064,
      "learning_rate": 4.391534596515872e-06,
      "loss": 0.1493,
      "step": 3635
    },
    {
      "epoch": 4.16,
      "grad_norm": 1.5438901357065633,
      "learning_rate": 4.379971028541565e-06,
      "loss": 0.154,
      "step": 3636
    },
    {
      "epoch": 4.16,
      "grad_norm": 1.5996936067119456,
      "learning_rate": 4.368421505940293e-06,
      "loss": 0.1701,
      "step": 3637
    },
    {
      "epoch": 4.16,
      "grad_norm": 1.8823633169314742,
      "learning_rate": 4.356886035043728e-06,
      "loss": 0.1921,
      "step": 3638
    },
    {
      "epoch": 4.16,
      "grad_norm": 1.53615527638196,
      "learning_rate": 4.3453646221758234e-06,
      "loss": 0.1513,
      "step": 3639
    },
    {
      "epoch": 4.16,
      "grad_norm": 1.5954419093833805,
      "learning_rate": 4.333857273652845e-06,
      "loss": 0.1792,
      "step": 3640
    },
    {
      "epoch": 4.16,
      "grad_norm": 1.687504061165206,
      "learning_rate": 4.3223639957833185e-06,
      "loss": 0.1688,
      "step": 3641
    },
    {
      "epoch": 4.16,
      "grad_norm": 1.7740885994944953,
      "learning_rate": 4.310884794868071e-06,
      "loss": 0.2014,
      "step": 3642
    },
    {
      "epoch": 4.16,
      "grad_norm": 1.4900979576675595,
      "learning_rate": 4.299419677200222e-06,
      "loss": 0.1676,
      "step": 3643
    },
    {
      "epoch": 4.16,
      "grad_norm": 1.836548333478299,
      "learning_rate": 4.287968649065159e-06,
      "loss": 0.2278,
      "step": 3644
    },
    {
      "epoch": 4.17,
      "grad_norm": 1.4808416028025106,
      "learning_rate": 4.27653171674054e-06,
      "loss": 0.1605,
      "step": 3645
    },
    {
      "epoch": 4.17,
      "grad_norm": 1.863325638957604,
      "learning_rate": 4.2651088864963e-06,
      "loss": 0.1989,
      "step": 3646
    },
    {
      "epoch": 4.17,
      "grad_norm": 1.8394435703432481,
      "learning_rate": 4.2537001645946576e-06,
      "loss": 0.1965,
      "step": 3647
    },
    {
      "epoch": 4.17,
      "grad_norm": 1.5264355081913268,
      "learning_rate": 4.242305557290076e-06,
      "loss": 0.1863,
      "step": 3648
    },
    {
      "epoch": 4.17,
      "grad_norm": 1.7878528601080501,
      "learning_rate": 4.230925070829302e-06,
      "loss": 0.1832,
      "step": 3649
    },
    {
      "epoch": 4.17,
      "grad_norm": 1.7168758194881346,
      "learning_rate": 4.219558711451312e-06,
      "loss": 0.1563,
      "step": 3650
    },
    {
      "epoch": 4.17,
      "grad_norm": 1.788948865352638,
      "learning_rate": 4.208206485387365e-06,
      "loss": 0.2045,
      "step": 3651
    },
    {
      "epoch": 4.17,
      "grad_norm": 1.4070163063696737,
      "learning_rate": 4.196868398860951e-06,
      "loss": 0.1742,
      "step": 3652
    },
    {
      "epoch": 4.17,
      "grad_norm": 1.7935062397205026,
      "learning_rate": 4.18554445808783e-06,
      "loss": 0.1885,
      "step": 3653
    },
    {
      "epoch": 4.18,
      "grad_norm": 1.4066891194080478,
      "learning_rate": 4.174234669275985e-06,
      "loss": 0.1461,
      "step": 3654
    },
    {
      "epoch": 4.18,
      "grad_norm": 1.7359884976773852,
      "learning_rate": 4.162939038625659e-06,
      "loss": 0.1818,
      "step": 3655
    },
    {
      "epoch": 4.18,
      "grad_norm": 1.7266933014088237,
      "learning_rate": 4.151657572329324e-06,
      "loss": 0.1709,
      "step": 3656
    },
    {
      "epoch": 4.18,
      "grad_norm": 1.5304706162458865,
      "learning_rate": 4.1403902765716906e-06,
      "loss": 0.1512,
      "step": 3657
    },
    {
      "epoch": 4.18,
      "grad_norm": 1.3540420751330315,
      "learning_rate": 4.129137157529689e-06,
      "loss": 0.1502,
      "step": 3658
    },
    {
      "epoch": 4.18,
      "grad_norm": 1.541419529595546,
      "learning_rate": 4.1178982213724906e-06,
      "loss": 0.1563,
      "step": 3659
    },
    {
      "epoch": 4.18,
      "grad_norm": 1.333327285317485,
      "learning_rate": 4.106673474261489e-06,
      "loss": 0.1272,
      "step": 3660
    },
    {
      "epoch": 4.18,
      "grad_norm": 1.7970454135853664,
      "learning_rate": 4.0954629223503015e-06,
      "loss": 0.2051,
      "step": 3661
    },
    {
      "epoch": 4.19,
      "grad_norm": 1.763218208254442,
      "learning_rate": 4.084266571784756e-06,
      "loss": 0.1882,
      "step": 3662
    },
    {
      "epoch": 4.19,
      "grad_norm": 1.6740414687469185,
      "learning_rate": 4.0730844287029e-06,
      "loss": 0.2035,
      "step": 3663
    },
    {
      "epoch": 4.19,
      "grad_norm": 1.6089319987805122,
      "learning_rate": 4.061916499234986e-06,
      "loss": 0.1679,
      "step": 3664
    },
    {
      "epoch": 4.19,
      "grad_norm": 1.6698464827319388,
      "learning_rate": 4.050762789503488e-06,
      "loss": 0.1872,
      "step": 3665
    },
    {
      "epoch": 4.19,
      "grad_norm": 1.5268536926450393,
      "learning_rate": 4.039623305623077e-06,
      "loss": 0.1772,
      "step": 3666
    },
    {
      "epoch": 4.19,
      "grad_norm": 1.66062017781184,
      "learning_rate": 4.028498053700608e-06,
      "loss": 0.1656,
      "step": 3667
    },
    {
      "epoch": 4.19,
      "grad_norm": 1.6271471593110698,
      "learning_rate": 4.017387039835162e-06,
      "loss": 0.1735,
      "step": 3668
    },
    {
      "epoch": 4.19,
      "grad_norm": 1.482583993816918,
      "learning_rate": 4.006290270117997e-06,
      "loss": 0.1638,
      "step": 3669
    },
    {
      "epoch": 4.19,
      "grad_norm": 1.8718051682878034,
      "learning_rate": 3.99520775063257e-06,
      "loss": 0.1782,
      "step": 3670
    },
    {
      "epoch": 4.2,
      "grad_norm": 1.6141658213422874,
      "learning_rate": 3.984139487454524e-06,
      "loss": 0.1778,
      "step": 3671
    },
    {
      "epoch": 4.2,
      "grad_norm": 1.8194945682036474,
      "learning_rate": 3.973085486651683e-06,
      "loss": 0.1933,
      "step": 3672
    },
    {
      "epoch": 4.2,
      "grad_norm": 1.5054438271196982,
      "learning_rate": 3.962045754284055e-06,
      "loss": 0.1596,
      "step": 3673
    },
    {
      "epoch": 4.2,
      "grad_norm": 1.7465253597542902,
      "learning_rate": 3.951020296403832e-06,
      "loss": 0.1845,
      "step": 3674
    },
    {
      "epoch": 4.2,
      "grad_norm": 1.6701892801658957,
      "learning_rate": 3.94000911905536e-06,
      "loss": 0.2058,
      "step": 3675
    },
    {
      "epoch": 4.2,
      "grad_norm": 1.4497681988638127,
      "learning_rate": 3.929012228275176e-06,
      "loss": 0.1452,
      "step": 3676
    },
    {
      "epoch": 4.2,
      "grad_norm": 1.8433970744565225,
      "learning_rate": 3.918029630091978e-06,
      "loss": 0.1836,
      "step": 3677
    },
    {
      "epoch": 4.2,
      "grad_norm": 1.6081330903920192,
      "learning_rate": 3.907061330526634e-06,
      "loss": 0.1873,
      "step": 3678
    },
    {
      "epoch": 4.2,
      "grad_norm": 1.781548909863726,
      "learning_rate": 3.8961073355921615e-06,
      "loss": 0.2112,
      "step": 3679
    },
    {
      "epoch": 4.21,
      "grad_norm": 1.6454210260561133,
      "learning_rate": 3.885167651293746e-06,
      "loss": 0.1561,
      "step": 3680
    },
    {
      "epoch": 4.21,
      "grad_norm": 1.5204579812239656,
      "learning_rate": 3.874242283628726e-06,
      "loss": 0.1435,
      "step": 3681
    },
    {
      "epoch": 4.21,
      "grad_norm": 1.703546948235267,
      "learning_rate": 3.863331238586594e-06,
      "loss": 0.2175,
      "step": 3682
    },
    {
      "epoch": 4.21,
      "grad_norm": 1.7610833058787978,
      "learning_rate": 3.852434522148975e-06,
      "loss": 0.183,
      "step": 3683
    },
    {
      "epoch": 4.21,
      "grad_norm": 1.7164209705433557,
      "learning_rate": 3.841552140289656e-06,
      "loss": 0.1894,
      "step": 3684
    },
    {
      "epoch": 4.21,
      "grad_norm": 1.8452260152123887,
      "learning_rate": 3.8306840989745564e-06,
      "loss": 0.2028,
      "step": 3685
    },
    {
      "epoch": 4.21,
      "grad_norm": 1.6357456237600827,
      "learning_rate": 3.819830404161739e-06,
      "loss": 0.1688,
      "step": 3686
    },
    {
      "epoch": 4.21,
      "grad_norm": 1.7858053254355557,
      "learning_rate": 3.808991061801412e-06,
      "loss": 0.1822,
      "step": 3687
    },
    {
      "epoch": 4.21,
      "grad_norm": 1.71794405521022,
      "learning_rate": 3.798166077835885e-06,
      "loss": 0.1545,
      "step": 3688
    },
    {
      "epoch": 4.22,
      "grad_norm": 1.8361786867735206,
      "learning_rate": 3.7873554581996206e-06,
      "loss": 0.2026,
      "step": 3689
    },
    {
      "epoch": 4.22,
      "grad_norm": 1.8160421680007246,
      "learning_rate": 3.7765592088192048e-06,
      "loss": 0.1865,
      "step": 3690
    },
    {
      "epoch": 4.22,
      "grad_norm": 1.885824484361385,
      "learning_rate": 3.7657773356133396e-06,
      "loss": 0.2033,
      "step": 3691
    },
    {
      "epoch": 4.22,
      "grad_norm": 1.6706287233385695,
      "learning_rate": 3.7550098444928404e-06,
      "loss": 0.2035,
      "step": 3692
    },
    {
      "epoch": 4.22,
      "grad_norm": 1.8665411171252195,
      "learning_rate": 3.7442567413606387e-06,
      "loss": 0.1986,
      "step": 3693
    },
    {
      "epoch": 4.22,
      "grad_norm": 1.6800521143435294,
      "learning_rate": 3.7335180321118055e-06,
      "loss": 0.1659,
      "step": 3694
    },
    {
      "epoch": 4.22,
      "grad_norm": 1.599052053696346,
      "learning_rate": 3.7227937226334918e-06,
      "loss": 0.1691,
      "step": 3695
    },
    {
      "epoch": 4.22,
      "grad_norm": 1.5144984118096654,
      "learning_rate": 3.712083818804951e-06,
      "loss": 0.1666,
      "step": 3696
    },
    {
      "epoch": 4.23,
      "grad_norm": 1.6320648556202413,
      "learning_rate": 3.70138832649756e-06,
      "loss": 0.2136,
      "step": 3697
    },
    {
      "epoch": 4.23,
      "grad_norm": 1.8542836738506607,
      "learning_rate": 3.690707251574782e-06,
      "loss": 0.1793,
      "step": 3698
    },
    {
      "epoch": 4.23,
      "grad_norm": 1.675671582118935,
      "learning_rate": 3.6800405998921895e-06,
      "loss": 0.1871,
      "step": 3699
    },
    {
      "epoch": 4.23,
      "grad_norm": 1.4942290157673703,
      "learning_rate": 3.6693883772974145e-06,
      "loss": 0.1512,
      "step": 3700
    },
    {
      "epoch": 4.23,
      "eval_blimp_filtered_avg": 0.7083582089552238,
      "eval_blimp_filtered_std": 0.00507529194566895,
      "step": 3700
    },
    {
      "epoch": 4.23,
      "eval_blimp_supplement_avg": 0.8297413793103449,
      "eval_blimp_supplement_std": 0.016984707085279518,
      "step": 3700
    },
    {
      "epoch": 4.23,
      "eval_vqa_filtered_avg": 0.54,
      "eval_vqa_filtered_std": 0.05009082659620333,
      "step": 3700
    },
    {
      "epoch": 4.23,
      "eval_winoground_filtered_avg": 0.63,
      "eval_winoground_filtered_std": 0.04852365870939099,
      "step": 3700
    },
    {
      "epoch": 4.23,
      "grad_norm": 1.9379166319975445,
      "learning_rate": 3.6587505896302233e-06,
      "loss": 0.1975,
      "step": 3701
    },
    {
      "epoch": 4.23,
      "grad_norm": 1.7599358702900567,
      "learning_rate": 3.64812724272244e-06,
      "loss": 0.1917,
      "step": 3702
    },
    {
      "epoch": 4.23,
      "grad_norm": 1.506788505916425,
      "learning_rate": 3.63751834239798e-06,
      "loss": 0.1716,
      "step": 3703
    },
    {
      "epoch": 4.23,
      "grad_norm": 1.5140111436764256,
      "learning_rate": 3.626923894472847e-06,
      "loss": 0.1759,
      "step": 3704
    },
    {
      "epoch": 4.23,
      "grad_norm": 2.1199666412129727,
      "learning_rate": 3.616343904755095e-06,
      "loss": 0.2243,
      "step": 3705
    },
    {
      "epoch": 4.24,
      "grad_norm": 1.6139545480506223,
      "learning_rate": 3.6057783790448816e-06,
      "loss": 0.1585,
      "step": 3706
    },
    {
      "epoch": 4.24,
      "grad_norm": 1.531290480619055,
      "learning_rate": 3.5952273231344233e-06,
      "loss": 0.1641,
      "step": 3707
    },
    {
      "epoch": 4.24,
      "grad_norm": 1.7333783801472369,
      "learning_rate": 3.5846907428080002e-06,
      "loss": 0.1781,
      "step": 3708
    },
    {
      "epoch": 4.24,
      "grad_norm": 1.7086976844386192,
      "learning_rate": 3.574168643841961e-06,
      "loss": 0.1566,
      "step": 3709
    },
    {
      "epoch": 4.24,
      "grad_norm": 1.7758587595849682,
      "learning_rate": 3.5636610320047145e-06,
      "loss": 0.189,
      "step": 3710
    },
    {
      "epoch": 4.24,
      "grad_norm": 1.571805204871822,
      "learning_rate": 3.5531679130567306e-06,
      "loss": 0.1517,
      "step": 3711
    },
    {
      "epoch": 4.24,
      "grad_norm": 1.7776506239373067,
      "learning_rate": 3.542689292750534e-06,
      "loss": 0.1646,
      "step": 3712
    },
    {
      "epoch": 4.24,
      "grad_norm": 1.5760176938755723,
      "learning_rate": 3.5322251768306857e-06,
      "loss": 0.1706,
      "step": 3713
    },
    {
      "epoch": 4.24,
      "grad_norm": 1.733314579920604,
      "learning_rate": 3.5217755710338173e-06,
      "loss": 0.178,
      "step": 3714
    },
    {
      "epoch": 4.25,
      "grad_norm": 1.7054815607807314,
      "learning_rate": 3.5113404810885894e-06,
      "loss": 0.1677,
      "step": 3715
    },
    {
      "epoch": 4.25,
      "grad_norm": 1.4153728288488856,
      "learning_rate": 3.5009199127157166e-06,
      "loss": 0.168,
      "step": 3716
    },
    {
      "epoch": 4.25,
      "grad_norm": 1.803139903278346,
      "learning_rate": 3.490513871627942e-06,
      "loss": 0.2252,
      "step": 3717
    },
    {
      "epoch": 4.25,
      "grad_norm": 1.4711243075084908,
      "learning_rate": 3.480122363530055e-06,
      "loss": 0.1603,
      "step": 3718
    },
    {
      "epoch": 4.25,
      "grad_norm": 1.6193819731583865,
      "learning_rate": 3.469745394118866e-06,
      "loss": 0.1726,
      "step": 3719
    },
    {
      "epoch": 4.25,
      "grad_norm": 1.6300429607800273,
      "learning_rate": 3.459382969083228e-06,
      "loss": 0.175,
      "step": 3720
    },
    {
      "epoch": 4.25,
      "grad_norm": 1.7750747171559065,
      "learning_rate": 3.4490350941040126e-06,
      "loss": 0.1941,
      "step": 3721
    },
    {
      "epoch": 4.25,
      "grad_norm": 1.8023245559729284,
      "learning_rate": 3.4387017748541096e-06,
      "loss": 0.2142,
      "step": 3722
    },
    {
      "epoch": 4.25,
      "grad_norm": 1.617242603028621,
      "learning_rate": 3.42838301699844e-06,
      "loss": 0.1923,
      "step": 3723
    },
    {
      "epoch": 4.26,
      "grad_norm": 1.6139096941742035,
      "learning_rate": 3.4180788261939388e-06,
      "loss": 0.1622,
      "step": 3724
    },
    {
      "epoch": 4.26,
      "grad_norm": 1.7901455774510333,
      "learning_rate": 3.407789208089548e-06,
      "loss": 0.205,
      "step": 3725
    },
    {
      "epoch": 4.26,
      "grad_norm": 1.4230732957782608,
      "learning_rate": 3.397514168326231e-06,
      "loss": 0.1515,
      "step": 3726
    },
    {
      "epoch": 4.26,
      "grad_norm": 1.4778718547421468,
      "learning_rate": 3.387253712536956e-06,
      "loss": 0.1847,
      "step": 3727
    },
    {
      "epoch": 4.26,
      "grad_norm": 1.7018000802906597,
      "learning_rate": 3.37700784634669e-06,
      "loss": 0.168,
      "step": 3728
    },
    {
      "epoch": 4.26,
      "grad_norm": 1.5281961120305074,
      "learning_rate": 3.3667765753724166e-06,
      "loss": 0.1688,
      "step": 3729
    },
    {
      "epoch": 4.26,
      "grad_norm": 1.544604052215719,
      "learning_rate": 3.3565599052230923e-06,
      "loss": 0.1694,
      "step": 3730
    },
    {
      "epoch": 4.26,
      "grad_norm": 1.6544354079005832,
      "learning_rate": 3.3463578414996865e-06,
      "loss": 0.165,
      "step": 3731
    },
    {
      "epoch": 4.27,
      "grad_norm": 1.8586987562287198,
      "learning_rate": 3.336170389795169e-06,
      "loss": 0.207,
      "step": 3732
    },
    {
      "epoch": 4.27,
      "grad_norm": 1.8521598534679622,
      "learning_rate": 3.3259975556944786e-06,
      "loss": 0.1837,
      "step": 3733
    },
    {
      "epoch": 4.27,
      "grad_norm": 1.716032190409265,
      "learning_rate": 3.3158393447745583e-06,
      "loss": 0.189,
      "step": 3734
    },
    {
      "epoch": 4.27,
      "grad_norm": 1.5900354108970638,
      "learning_rate": 3.305695762604326e-06,
      "loss": 0.1543,
      "step": 3735
    },
    {
      "epoch": 4.27,
      "grad_norm": 1.6072082898948825,
      "learning_rate": 3.295566814744677e-06,
      "loss": 0.1632,
      "step": 3736
    },
    {
      "epoch": 4.27,
      "grad_norm": 1.7034582656615571,
      "learning_rate": 3.2854525067484896e-06,
      "loss": 0.2038,
      "step": 3737
    },
    {
      "epoch": 4.27,
      "grad_norm": 1.881289155989666,
      "learning_rate": 3.2753528441606216e-06,
      "loss": 0.2194,
      "step": 3738
    },
    {
      "epoch": 4.27,
      "grad_norm": 1.685743892853709,
      "learning_rate": 3.2652678325178777e-06,
      "loss": 0.1826,
      "step": 3739
    },
    {
      "epoch": 4.27,
      "grad_norm": 1.5725529781375633,
      "learning_rate": 3.2551974773490545e-06,
      "loss": 0.1758,
      "step": 3740
    },
    {
      "epoch": 4.28,
      "grad_norm": 1.638366042208894,
      "learning_rate": 3.245141784174913e-06,
      "loss": 0.1724,
      "step": 3741
    },
    {
      "epoch": 4.28,
      "grad_norm": 1.7894924599785673,
      "learning_rate": 3.2351007585081738e-06,
      "loss": 0.1967,
      "step": 3742
    },
    {
      "epoch": 4.28,
      "grad_norm": 1.8599577759056063,
      "learning_rate": 3.2250744058534997e-06,
      "loss": 0.2131,
      "step": 3743
    },
    {
      "epoch": 4.28,
      "grad_norm": 1.5618530956324759,
      "learning_rate": 3.2150627317075244e-06,
      "loss": 0.1572,
      "step": 3744
    },
    {
      "epoch": 4.28,
      "grad_norm": 1.7841789625430315,
      "learning_rate": 3.2050657415588403e-06,
      "loss": 0.172,
      "step": 3745
    },
    {
      "epoch": 4.28,
      "grad_norm": 1.5012994743288999,
      "learning_rate": 3.1950834408879804e-06,
      "loss": 0.1737,
      "step": 3746
    },
    {
      "epoch": 4.28,
      "grad_norm": 1.694681031647407,
      "learning_rate": 3.1851158351674103e-06,
      "loss": 0.1672,
      "step": 3747
    },
    {
      "epoch": 4.28,
      "grad_norm": 1.3855122829023268,
      "learning_rate": 3.1751629298615793e-06,
      "loss": 0.134,
      "step": 3748
    },
    {
      "epoch": 4.28,
      "grad_norm": 1.6896437469009802,
      "learning_rate": 3.165224730426839e-06,
      "loss": 0.1692,
      "step": 3749
    },
    {
      "epoch": 4.29,
      "grad_norm": 1.8973844833276756,
      "learning_rate": 3.1553012423115046e-06,
      "loss": 0.1962,
      "step": 3750
    },
    {
      "epoch": 4.29,
      "grad_norm": 1.585766981232494,
      "learning_rate": 3.1453924709558047e-06,
      "loss": 0.1485,
      "step": 3751
    },
    {
      "epoch": 4.29,
      "grad_norm": 1.654420988712925,
      "learning_rate": 3.135498421791917e-06,
      "loss": 0.1673,
      "step": 3752
    },
    {
      "epoch": 4.29,
      "grad_norm": 1.9283099269687693,
      "learning_rate": 3.125619100243936e-06,
      "loss": 0.1764,
      "step": 3753
    },
    {
      "epoch": 4.29,
      "grad_norm": 1.9014048251149367,
      "learning_rate": 3.1157545117278962e-06,
      "loss": 0.1959,
      "step": 3754
    },
    {
      "epoch": 4.29,
      "grad_norm": 1.5332782520116035,
      "learning_rate": 3.1059046616517415e-06,
      "loss": 0.186,
      "step": 3755
    },
    {
      "epoch": 4.29,
      "grad_norm": 1.8083507686463451,
      "learning_rate": 3.0960695554153427e-06,
      "loss": 0.1822,
      "step": 3756
    },
    {
      "epoch": 4.29,
      "grad_norm": 1.970880312821568,
      "learning_rate": 3.0862491984104867e-06,
      "loss": 0.2155,
      "step": 3757
    },
    {
      "epoch": 4.29,
      "grad_norm": 1.4244531901945765,
      "learning_rate": 3.076443596020877e-06,
      "loss": 0.1404,
      "step": 3758
    },
    {
      "epoch": 4.3,
      "grad_norm": 1.5701042368260012,
      "learning_rate": 3.06665275362213e-06,
      "loss": 0.1651,
      "step": 3759
    },
    {
      "epoch": 4.3,
      "grad_norm": 1.4762187023794675,
      "learning_rate": 3.0568766765817558e-06,
      "loss": 0.148,
      "step": 3760
    },
    {
      "epoch": 4.3,
      "grad_norm": 1.9643663245124343,
      "learning_rate": 3.0471153702591837e-06,
      "loss": 0.1897,
      "step": 3761
    },
    {
      "epoch": 4.3,
      "grad_norm": 1.7218624078435578,
      "learning_rate": 3.0373688400057462e-06,
      "loss": 0.186,
      "step": 3762
    },
    {
      "epoch": 4.3,
      "grad_norm": 1.5602958450742166,
      "learning_rate": 3.0276370911646666e-06,
      "loss": 0.1621,
      "step": 3763
    },
    {
      "epoch": 4.3,
      "grad_norm": 1.6241242214562377,
      "learning_rate": 3.01792012907107e-06,
      "loss": 0.1658,
      "step": 3764
    },
    {
      "epoch": 4.3,
      "grad_norm": 1.6999327091949208,
      "learning_rate": 3.008217959051979e-06,
      "loss": 0.1746,
      "step": 3765
    },
    {
      "epoch": 4.3,
      "grad_norm": 1.6927535383925385,
      "learning_rate": 2.9985305864262947e-06,
      "loss": 0.1801,
      "step": 3766
    },
    {
      "epoch": 4.31,
      "grad_norm": 2.0673832667639944,
      "learning_rate": 2.988858016504826e-06,
      "loss": 0.2171,
      "step": 3767
    },
    {
      "epoch": 4.31,
      "grad_norm": 1.8439790994377017,
      "learning_rate": 2.979200254590242e-06,
      "loss": 0.2182,
      "step": 3768
    },
    {
      "epoch": 4.31,
      "grad_norm": 1.5892006457219734,
      "learning_rate": 2.969557305977113e-06,
      "loss": 0.199,
      "step": 3769
    },
    {
      "epoch": 4.31,
      "grad_norm": 1.5963440670907967,
      "learning_rate": 2.9599291759518743e-06,
      "loss": 0.1607,
      "step": 3770
    },
    {
      "epoch": 4.31,
      "grad_norm": 1.695454776163313,
      "learning_rate": 2.9503158697928536e-06,
      "loss": 0.1767,
      "step": 3771
    },
    {
      "epoch": 4.31,
      "grad_norm": 1.5942969642380385,
      "learning_rate": 2.9407173927702346e-06,
      "loss": 0.1493,
      "step": 3772
    },
    {
      "epoch": 4.31,
      "grad_norm": 2.1105125762486567,
      "learning_rate": 2.9311337501460857e-06,
      "loss": 0.2524,
      "step": 3773
    },
    {
      "epoch": 4.31,
      "grad_norm": 1.631373714950384,
      "learning_rate": 2.9215649471743364e-06,
      "loss": 0.1718,
      "step": 3774
    },
    {
      "epoch": 4.31,
      "grad_norm": 1.5791358135619606,
      "learning_rate": 2.9120109891007785e-06,
      "loss": 0.1418,
      "step": 3775
    },
    {
      "epoch": 4.32,
      "grad_norm": 1.8505539592689288,
      "learning_rate": 2.9024718811630747e-06,
      "loss": 0.1998,
      "step": 3776
    },
    {
      "epoch": 4.32,
      "grad_norm": 1.636048224201295,
      "learning_rate": 2.892947628590732e-06,
      "loss": 0.1722,
      "step": 3777
    },
    {
      "epoch": 4.32,
      "grad_norm": 1.7489945774552291,
      "learning_rate": 2.8834382366051203e-06,
      "loss": 0.1904,
      "step": 3778
    },
    {
      "epoch": 4.32,
      "grad_norm": 1.844492146543618,
      "learning_rate": 2.8739437104194665e-06,
      "loss": 0.2362,
      "step": 3779
    },
    {
      "epoch": 4.32,
      "grad_norm": 1.691763915021842,
      "learning_rate": 2.864464055238857e-06,
      "loss": 0.1747,
      "step": 3780
    },
    {
      "epoch": 4.32,
      "grad_norm": 1.7412970184130097,
      "learning_rate": 2.854999276260197e-06,
      "loss": 0.2078,
      "step": 3781
    },
    {
      "epoch": 4.32,
      "grad_norm": 1.628870476533188,
      "learning_rate": 2.8455493786722573e-06,
      "loss": 0.177,
      "step": 3782
    },
    {
      "epoch": 4.32,
      "grad_norm": 1.6731719278351374,
      "learning_rate": 2.8361143676556534e-06,
      "loss": 0.1986,
      "step": 3783
    },
    {
      "epoch": 4.32,
      "grad_norm": 1.8594224811602127,
      "learning_rate": 2.826694248382831e-06,
      "loss": 0.1887,
      "step": 3784
    },
    {
      "epoch": 4.33,
      "grad_norm": 1.823708888750284,
      "learning_rate": 2.8172890260180684e-06,
      "loss": 0.1805,
      "step": 3785
    },
    {
      "epoch": 4.33,
      "grad_norm": 2.165661914423505,
      "learning_rate": 2.807898705717479e-06,
      "loss": 0.2267,
      "step": 3786
    },
    {
      "epoch": 4.33,
      "grad_norm": 1.6365488828317485,
      "learning_rate": 2.798523292629014e-06,
      "loss": 0.1716,
      "step": 3787
    },
    {
      "epoch": 4.33,
      "grad_norm": 1.6150671650909918,
      "learning_rate": 2.789162791892459e-06,
      "loss": 0.1855,
      "step": 3788
    },
    {
      "epoch": 4.33,
      "grad_norm": 1.710039572313244,
      "learning_rate": 2.7798172086393957e-06,
      "loss": 0.1987,
      "step": 3789
    },
    {
      "epoch": 4.33,
      "grad_norm": 1.5259488454599264,
      "learning_rate": 2.7704865479932562e-06,
      "loss": 0.1659,
      "step": 3790
    },
    {
      "epoch": 4.33,
      "grad_norm": 1.7460289615461273,
      "learning_rate": 2.761170815069275e-06,
      "loss": 0.191,
      "step": 3791
    },
    {
      "epoch": 4.33,
      "grad_norm": 1.7094893715054138,
      "learning_rate": 2.7518700149745137e-06,
      "loss": 0.2098,
      "step": 3792
    },
    {
      "epoch": 4.33,
      "grad_norm": 1.8882857424421766,
      "learning_rate": 2.7425841528078466e-06,
      "loss": 0.2164,
      "step": 3793
    },
    {
      "epoch": 4.34,
      "grad_norm": 1.6434176807237562,
      "learning_rate": 2.7333132336599354e-06,
      "loss": 0.1717,
      "step": 3794
    },
    {
      "epoch": 4.34,
      "grad_norm": 1.6576101621675476,
      "learning_rate": 2.724057262613291e-06,
      "loss": 0.1902,
      "step": 3795
    },
    {
      "epoch": 4.34,
      "grad_norm": 1.5300346178532163,
      "learning_rate": 2.714816244742201e-06,
      "loss": 0.1702,
      "step": 3796
    },
    {
      "epoch": 4.34,
      "grad_norm": 1.5470388543347162,
      "learning_rate": 2.705590185112763e-06,
      "loss": 0.1656,
      "step": 3797
    },
    {
      "epoch": 4.34,
      "grad_norm": 1.769070315643092,
      "learning_rate": 2.6963790887828677e-06,
      "loss": 0.1924,
      "step": 3798
    },
    {
      "epoch": 4.34,
      "grad_norm": 1.9078654209082329,
      "learning_rate": 2.687182960802209e-06,
      "loss": 0.2171,
      "step": 3799
    },
    {
      "epoch": 4.34,
      "grad_norm": 1.9146213762853834,
      "learning_rate": 2.6780018062122813e-06,
      "loss": 0.1977,
      "step": 3800
    },
    {
      "epoch": 4.34,
      "eval_blimp_filtered_avg": 0.707910447761194,
      "eval_blimp_filtered_std": 0.005081030182875898,
      "step": 3800
    },
    {
      "epoch": 4.34,
      "eval_blimp_supplement_avg": 0.8189655172413793,
      "eval_blimp_supplement_std": 0.017403880156794286,
      "step": 3800
    },
    {
      "epoch": 4.34,
      "eval_vqa_filtered_avg": 0.55,
      "eval_vqa_filtered_std": 0.049999999999999996,
      "step": 3800
    },
    {
      "epoch": 4.34,
      "eval_winoground_filtered_avg": 0.62,
      "eval_winoground_filtered_std": 0.048783173121456316,
      "step": 3800
    },
    {
      "epoch": 4.34,
      "grad_norm": 1.9848165159434703,
      "learning_rate": 2.6688356300463577e-06,
      "loss": 0.1937,
      "step": 3801
    },
    {
      "epoch": 4.35,
      "grad_norm": 1.5820438079472403,
      "learning_rate": 2.659684437329499e-06,
      "loss": 0.1667,
      "step": 3802
    },
    {
      "epoch": 4.35,
      "grad_norm": 1.7504436299685544,
      "learning_rate": 2.650548233078569e-06,
      "loss": 0.1955,
      "step": 3803
    },
    {
      "epoch": 4.35,
      "grad_norm": 1.478627003643683,
      "learning_rate": 2.641427022302201e-06,
      "loss": 0.1577,
      "step": 3804
    },
    {
      "epoch": 4.35,
      "grad_norm": 1.5665598513380141,
      "learning_rate": 2.6323208100008145e-06,
      "loss": 0.1679,
      "step": 3805
    },
    {
      "epoch": 4.35,
      "grad_norm": 1.472053196301574,
      "learning_rate": 2.6232296011665934e-06,
      "loss": 0.1661,
      "step": 3806
    },
    {
      "epoch": 4.35,
      "grad_norm": 1.5587290567311014,
      "learning_rate": 2.6141534007835112e-06,
      "loss": 0.1515,
      "step": 3807
    },
    {
      "epoch": 4.35,
      "grad_norm": 1.6797472355211822,
      "learning_rate": 2.605092213827307e-06,
      "loss": 0.1802,
      "step": 3808
    },
    {
      "epoch": 4.35,
      "grad_norm": 1.786623668539048,
      "learning_rate": 2.5960460452654943e-06,
      "loss": 0.2025,
      "step": 3809
    },
    {
      "epoch": 4.35,
      "grad_norm": 1.9099458192921468,
      "learning_rate": 2.5870149000573494e-06,
      "loss": 0.1908,
      "step": 3810
    },
    {
      "epoch": 4.36,
      "grad_norm": 1.9260347534033528,
      "learning_rate": 2.5779987831539085e-06,
      "loss": 0.2182,
      "step": 3811
    },
    {
      "epoch": 4.36,
      "grad_norm": 1.8318191214900343,
      "learning_rate": 2.5689976994979814e-06,
      "loss": 0.2169,
      "step": 3812
    },
    {
      "epoch": 4.36,
      "grad_norm": 1.7222049301465228,
      "learning_rate": 2.5600116540241203e-06,
      "loss": 0.1913,
      "step": 3813
    },
    {
      "epoch": 4.36,
      "grad_norm": 1.6970104471100944,
      "learning_rate": 2.551040651658648e-06,
      "loss": 0.175,
      "step": 3814
    },
    {
      "epoch": 4.36,
      "grad_norm": 1.8479878922440163,
      "learning_rate": 2.5420846973196275e-06,
      "loss": 0.1886,
      "step": 3815
    },
    {
      "epoch": 4.36,
      "grad_norm": 1.4960431652801947,
      "learning_rate": 2.533143795916881e-06,
      "loss": 0.1569,
      "step": 3816
    },
    {
      "epoch": 4.36,
      "grad_norm": 1.7113975186796269,
      "learning_rate": 2.524217952351975e-06,
      "loss": 0.2112,
      "step": 3817
    },
    {
      "epoch": 4.36,
      "grad_norm": 1.7878746295271786,
      "learning_rate": 2.515307171518221e-06,
      "loss": 0.2002,
      "step": 3818
    },
    {
      "epoch": 4.36,
      "grad_norm": 2.073489635252811,
      "learning_rate": 2.506411458300678e-06,
      "loss": 0.2004,
      "step": 3819
    },
    {
      "epoch": 4.37,
      "grad_norm": 1.5788117527847905,
      "learning_rate": 2.497530817576136e-06,
      "loss": 0.1694,
      "step": 3820
    },
    {
      "epoch": 4.37,
      "grad_norm": 1.3458783819753728,
      "learning_rate": 2.4886652542131295e-06,
      "loss": 0.1431,
      "step": 3821
    },
    {
      "epoch": 4.37,
      "grad_norm": 1.613802588619844,
      "learning_rate": 2.4798147730719246e-06,
      "loss": 0.1729,
      "step": 3822
    },
    {
      "epoch": 4.37,
      "grad_norm": 1.8551094752019834,
      "learning_rate": 2.4709793790045134e-06,
      "loss": 0.2099,
      "step": 3823
    },
    {
      "epoch": 4.37,
      "grad_norm": 1.7321832492171356,
      "learning_rate": 2.462159076854624e-06,
      "loss": 0.1884,
      "step": 3824
    },
    {
      "epoch": 4.37,
      "grad_norm": 1.4331054309953524,
      "learning_rate": 2.4533538714577097e-06,
      "loss": 0.1393,
      "step": 3825
    },
    {
      "epoch": 4.37,
      "grad_norm": 1.744271926456426,
      "learning_rate": 2.4445637676409416e-06,
      "loss": 0.1961,
      "step": 3826
    },
    {
      "epoch": 4.37,
      "grad_norm": 1.5360653554720929,
      "learning_rate": 2.4357887702232353e-06,
      "loss": 0.158,
      "step": 3827
    },
    {
      "epoch": 4.37,
      "grad_norm": 1.5833881559596201,
      "learning_rate": 2.4270288840151856e-06,
      "loss": 0.1751,
      "step": 3828
    },
    {
      "epoch": 4.38,
      "grad_norm": 1.485718058620223,
      "learning_rate": 2.4182841138191313e-06,
      "loss": 0.1637,
      "step": 3829
    },
    {
      "epoch": 4.38,
      "grad_norm": 1.507098629546646,
      "learning_rate": 2.4095544644291177e-06,
      "loss": 0.1662,
      "step": 3830
    },
    {
      "epoch": 4.38,
      "grad_norm": 1.7834279544633096,
      "learning_rate": 2.4008399406309066e-06,
      "loss": 0.1873,
      "step": 3831
    },
    {
      "epoch": 4.38,
      "grad_norm": 1.6893118485771408,
      "learning_rate": 2.3921405472019464e-06,
      "loss": 0.1915,
      "step": 3832
    },
    {
      "epoch": 4.38,
      "grad_norm": 1.6334106467372937,
      "learning_rate": 2.3834562889114155e-06,
      "loss": 0.1825,
      "step": 3833
    },
    {
      "epoch": 4.38,
      "grad_norm": 1.639843100643248,
      "learning_rate": 2.3747871705201718e-06,
      "loss": 0.1816,
      "step": 3834
    },
    {
      "epoch": 4.38,
      "grad_norm": 1.676066245751482,
      "learning_rate": 2.366133196780811e-06,
      "loss": 0.1687,
      "step": 3835
    },
    {
      "epoch": 4.38,
      "grad_norm": 1.3439277138717647,
      "learning_rate": 2.3574943724375774e-06,
      "loss": 0.1459,
      "step": 3836
    },
    {
      "epoch": 4.39,
      "grad_norm": 1.8437339334483593,
      "learning_rate": 2.3488707022264465e-06,
      "loss": 0.1768,
      "step": 3837
    },
    {
      "epoch": 4.39,
      "grad_norm": 1.568111749741232,
      "learning_rate": 2.34026219087507e-06,
      "loss": 0.1792,
      "step": 3838
    },
    {
      "epoch": 4.39,
      "grad_norm": 1.8196028322181468,
      "learning_rate": 2.3316688431027953e-06,
      "loss": 0.2333,
      "step": 3839
    },
    {
      "epoch": 4.39,
      "grad_norm": 1.7249837520342939,
      "learning_rate": 2.3230906636206485e-06,
      "loss": 0.198,
      "step": 3840
    },
    {
      "epoch": 4.39,
      "grad_norm": 1.695179577780692,
      "learning_rate": 2.3145276571313433e-06,
      "loss": 0.1977,
      "step": 3841
    },
    {
      "epoch": 4.39,
      "grad_norm": 1.6877741376212985,
      "learning_rate": 2.305979828329289e-06,
      "loss": 0.1591,
      "step": 3842
    },
    {
      "epoch": 4.39,
      "grad_norm": 1.855127236550373,
      "learning_rate": 2.2974471819005592e-06,
      "loss": 0.2325,
      "step": 3843
    },
    {
      "epoch": 4.39,
      "grad_norm": 1.5211167884825978,
      "learning_rate": 2.2889297225229135e-06,
      "loss": 0.1333,
      "step": 3844
    },
    {
      "epoch": 4.39,
      "grad_norm": 1.6718287404150853,
      "learning_rate": 2.280427454865769e-06,
      "loss": 0.1651,
      "step": 3845
    },
    {
      "epoch": 4.4,
      "grad_norm": 1.676540741591485,
      "learning_rate": 2.2719403835902306e-06,
      "loss": 0.1694,
      "step": 3846
    },
    {
      "epoch": 4.4,
      "grad_norm": 1.9784048131576546,
      "learning_rate": 2.2634685133490674e-06,
      "loss": 0.2124,
      "step": 3847
    },
    {
      "epoch": 4.4,
      "grad_norm": 1.6992407217889358,
      "learning_rate": 2.2550118487867198e-06,
      "loss": 0.1913,
      "step": 3848
    },
    {
      "epoch": 4.4,
      "grad_norm": 1.6567739900756446,
      "learning_rate": 2.2465703945392802e-06,
      "loss": 0.1733,
      "step": 3849
    },
    {
      "epoch": 4.4,
      "grad_norm": 1.1462203177736834,
      "learning_rate": 2.238144155234513e-06,
      "loss": 0.1146,
      "step": 3850
    },
    {
      "epoch": 4.4,
      "grad_norm": 1.6603059175838457,
      "learning_rate": 2.2297331354918395e-06,
      "loss": 0.1713,
      "step": 3851
    },
    {
      "epoch": 4.4,
      "grad_norm": 1.5487916414097593,
      "learning_rate": 2.221337339922338e-06,
      "loss": 0.1788,
      "step": 3852
    },
    {
      "epoch": 4.4,
      "grad_norm": 1.55650628194036,
      "learning_rate": 2.212956773128734e-06,
      "loss": 0.1592,
      "step": 3853
    },
    {
      "epoch": 4.4,
      "grad_norm": 1.7775258650370211,
      "learning_rate": 2.2045914397054078e-06,
      "loss": 0.1826,
      "step": 3854
    },
    {
      "epoch": 4.41,
      "grad_norm": 1.6852258922764494,
      "learning_rate": 2.1962413442383922e-06,
      "loss": 0.1564,
      "step": 3855
    },
    {
      "epoch": 4.41,
      "grad_norm": 1.63264007860541,
      "learning_rate": 2.187906491305365e-06,
      "loss": 0.1682,
      "step": 3856
    },
    {
      "epoch": 4.41,
      "grad_norm": 1.819696181904699,
      "learning_rate": 2.179586885475647e-06,
      "loss": 0.1932,
      "step": 3857
    },
    {
      "epoch": 4.41,
      "grad_norm": 1.7203681513206885,
      "learning_rate": 2.1712825313101946e-06,
      "loss": 0.1587,
      "step": 3858
    },
    {
      "epoch": 4.41,
      "grad_norm": 1.4839581891006914,
      "learning_rate": 2.1629934333616154e-06,
      "loss": 0.179,
      "step": 3859
    },
    {
      "epoch": 4.41,
      "grad_norm": 1.7733261236262234,
      "learning_rate": 2.1547195961741473e-06,
      "loss": 0.1868,
      "step": 3860
    },
    {
      "epoch": 4.41,
      "grad_norm": 1.380667221320021,
      "learning_rate": 2.146461024283649e-06,
      "loss": 0.1392,
      "step": 3861
    },
    {
      "epoch": 4.41,
      "grad_norm": 1.632955378848119,
      "learning_rate": 2.1382177222176293e-06,
      "loss": 0.1596,
      "step": 3862
    },
    {
      "epoch": 4.41,
      "grad_norm": 1.920400662548086,
      "learning_rate": 2.1299896944952157e-06,
      "loss": 0.1861,
      "step": 3863
    },
    {
      "epoch": 4.42,
      "grad_norm": 1.6396921779521352,
      "learning_rate": 2.1217769456271706e-06,
      "loss": 0.1521,
      "step": 3864
    },
    {
      "epoch": 4.42,
      "grad_norm": 1.4956126013282043,
      "learning_rate": 2.1135794801158725e-06,
      "loss": 0.1591,
      "step": 3865
    },
    {
      "epoch": 4.42,
      "grad_norm": 1.6373782966325223,
      "learning_rate": 2.1053973024553195e-06,
      "loss": 0.1808,
      "step": 3866
    },
    {
      "epoch": 4.42,
      "grad_norm": 1.6091375308580322,
      "learning_rate": 2.0972304171311373e-06,
      "loss": 0.1561,
      "step": 3867
    },
    {
      "epoch": 4.42,
      "grad_norm": 1.5712204609872955,
      "learning_rate": 2.089078828620561e-06,
      "loss": 0.1599,
      "step": 3868
    },
    {
      "epoch": 4.42,
      "grad_norm": 1.9270856963484808,
      "learning_rate": 2.0809425413924455e-06,
      "loss": 0.2082,
      "step": 3869
    },
    {
      "epoch": 4.42,
      "grad_norm": 1.8737174316832765,
      "learning_rate": 2.0728215599072462e-06,
      "loss": 0.2092,
      "step": 3870
    },
    {
      "epoch": 4.42,
      "grad_norm": 1.7565266822722814,
      "learning_rate": 2.0647158886170393e-06,
      "loss": 0.2045,
      "step": 3871
    },
    {
      "epoch": 4.43,
      "grad_norm": 1.5369882359206455,
      "learning_rate": 2.0566255319655056e-06,
      "loss": 0.1494,
      "step": 3872
    },
    {
      "epoch": 4.43,
      "grad_norm": 1.500963772145608,
      "learning_rate": 2.0485504943879254e-06,
      "loss": 0.1566,
      "step": 3873
    },
    {
      "epoch": 4.43,
      "grad_norm": 1.8739987516657295,
      "learning_rate": 2.0404907803111828e-06,
      "loss": 0.1915,
      "step": 3874
    },
    {
      "epoch": 4.43,
      "grad_norm": 1.813399469713103,
      "learning_rate": 2.032446394153764e-06,
      "loss": 0.1814,
      "step": 3875
    },
    {
      "epoch": 4.43,
      "grad_norm": 1.4963274087060259,
      "learning_rate": 2.0244173403257484e-06,
      "loss": 0.1597,
      "step": 3876
    },
    {
      "epoch": 4.43,
      "grad_norm": 1.8234693572527871,
      "learning_rate": 2.016403623228821e-06,
      "loss": 0.1852,
      "step": 3877
    },
    {
      "epoch": 4.43,
      "grad_norm": 1.5353683380508785,
      "learning_rate": 2.008405247256231e-06,
      "loss": 0.1718,
      "step": 3878
    },
    {
      "epoch": 4.43,
      "grad_norm": 1.7845997064361374,
      "learning_rate": 2.0004222167928475e-06,
      "loss": 0.2029,
      "step": 3879
    },
    {
      "epoch": 4.43,
      "grad_norm": 1.6067101161700668,
      "learning_rate": 1.9924545362151136e-06,
      "loss": 0.164,
      "step": 3880
    },
    {
      "epoch": 4.44,
      "grad_norm": 1.7809694813089347,
      "learning_rate": 1.9845022098910546e-06,
      "loss": 0.1729,
      "step": 3881
    },
    {
      "epoch": 4.44,
      "grad_norm": 1.9332754470171554,
      "learning_rate": 1.976565242180297e-06,
      "loss": 0.181,
      "step": 3882
    },
    {
      "epoch": 4.44,
      "grad_norm": 1.7240767638280818,
      "learning_rate": 1.96864363743402e-06,
      "loss": 0.1826,
      "step": 3883
    },
    {
      "epoch": 4.44,
      "grad_norm": 1.552218489805046,
      "learning_rate": 1.960737399994993e-06,
      "loss": 0.1344,
      "step": 3884
    },
    {
      "epoch": 4.44,
      "grad_norm": 1.5549442184627702,
      "learning_rate": 1.9528465341975656e-06,
      "loss": 0.1778,
      "step": 3885
    },
    {
      "epoch": 4.44,
      "grad_norm": 1.832381699346936,
      "learning_rate": 1.9449710443676606e-06,
      "loss": 0.1981,
      "step": 3886
    },
    {
      "epoch": 4.44,
      "grad_norm": 1.696020099568385,
      "learning_rate": 1.937110934822758e-06,
      "loss": 0.1891,
      "step": 3887
    },
    {
      "epoch": 4.44,
      "grad_norm": 1.7023921382234626,
      "learning_rate": 1.9292662098719137e-06,
      "loss": 0.1884,
      "step": 3888
    },
    {
      "epoch": 4.44,
      "grad_norm": 1.6819089120534978,
      "learning_rate": 1.9214368738157615e-06,
      "loss": 0.1697,
      "step": 3889
    },
    {
      "epoch": 4.45,
      "grad_norm": 1.6447362090540558,
      "learning_rate": 1.913622930946488e-06,
      "loss": 0.1786,
      "step": 3890
    },
    {
      "epoch": 4.45,
      "grad_norm": 1.915708334615016,
      "learning_rate": 1.9058243855478296e-06,
      "loss": 0.2197,
      "step": 3891
    },
    {
      "epoch": 4.45,
      "grad_norm": 1.9423914370125803,
      "learning_rate": 1.8980412418950998e-06,
      "loss": 0.1924,
      "step": 3892
    },
    {
      "epoch": 4.45,
      "grad_norm": 1.5567303674805966,
      "learning_rate": 1.8902735042551589e-06,
      "loss": 0.1538,
      "step": 3893
    },
    {
      "epoch": 4.45,
      "grad_norm": 1.7107347454112432,
      "learning_rate": 1.8825211768864337e-06,
      "loss": 0.1736,
      "step": 3894
    },
    {
      "epoch": 4.45,
      "grad_norm": 1.6833456072557613,
      "learning_rate": 1.8747842640388745e-06,
      "loss": 0.155,
      "step": 3895
    },
    {
      "epoch": 4.45,
      "grad_norm": 1.7300771341619947,
      "learning_rate": 1.8670627699540156e-06,
      "loss": 0.1524,
      "step": 3896
    },
    {
      "epoch": 4.45,
      "grad_norm": 1.6166971071789003,
      "learning_rate": 1.859356698864917e-06,
      "loss": 0.1939,
      "step": 3897
    },
    {
      "epoch": 4.45,
      "grad_norm": 1.8210037042325553,
      "learning_rate": 1.8516660549961894e-06,
      "loss": 0.1867,
      "step": 3898
    },
    {
      "epoch": 4.46,
      "grad_norm": 1.7760844407786225,
      "learning_rate": 1.843990842563994e-06,
      "loss": 0.163,
      "step": 3899
    },
    {
      "epoch": 4.46,
      "grad_norm": 1.5449849534722269,
      "learning_rate": 1.8363310657760114e-06,
      "loss": 0.1573,
      "step": 3900
    },
    {
      "epoch": 4.46,
      "eval_blimp_filtered_avg": 0.706865671641791,
      "eval_blimp_filtered_std": 0.005078837568083467,
      "step": 3900
    },
    {
      "epoch": 4.46,
      "eval_blimp_supplement_avg": 0.8275862068965517,
      "eval_blimp_supplement_std": 0.01704679409460659,
      "step": 3900
    },
    {
      "epoch": 4.46,
      "eval_vqa_filtered_avg": 0.55,
      "eval_vqa_filtered_std": 0.049999999999999996,
      "step": 3900
    },
    {
      "epoch": 4.46,
      "eval_winoground_filtered_avg": 0.63,
      "eval_winoground_filtered_std": 0.04852365870939099,
      "step": 3900
    },
    {
      "epoch": 4.46,
      "grad_norm": 1.354056509929048,
      "learning_rate": 1.8286867288314758e-06,
      "loss": 0.1364,
      "step": 3901
    },
    {
      "epoch": 4.46,
      "grad_norm": 1.7042012327118783,
      "learning_rate": 1.8210578359211583e-06,
      "loss": 0.1592,
      "step": 3902
    },
    {
      "epoch": 4.46,
      "grad_norm": 1.5791955014081314,
      "learning_rate": 1.8134443912273534e-06,
      "loss": 0.1863,
      "step": 3903
    },
    {
      "epoch": 4.46,
      "grad_norm": 1.6782977727014923,
      "learning_rate": 1.805846398923896e-06,
      "loss": 0.1763,
      "step": 3904
    },
    {
      "epoch": 4.46,
      "grad_norm": 1.6722864533678565,
      "learning_rate": 1.7982638631761472e-06,
      "loss": 0.1692,
      "step": 3905
    },
    {
      "epoch": 4.46,
      "grad_norm": 1.845122647122584,
      "learning_rate": 1.7906967881409919e-06,
      "loss": 0.2049,
      "step": 3906
    },
    {
      "epoch": 4.47,
      "grad_norm": 1.8881758544473501,
      "learning_rate": 1.7831451779668417e-06,
      "loss": 0.2071,
      "step": 3907
    },
    {
      "epoch": 4.47,
      "grad_norm": 1.630648200891546,
      "learning_rate": 1.7756090367936217e-06,
      "loss": 0.1632,
      "step": 3908
    },
    {
      "epoch": 4.47,
      "grad_norm": 1.6756164618361482,
      "learning_rate": 1.7680883687527872e-06,
      "loss": 0.16,
      "step": 3909
    },
    {
      "epoch": 4.47,
      "grad_norm": 1.4887709300303336,
      "learning_rate": 1.76058317796731e-06,
      "loss": 0.1438,
      "step": 3910
    },
    {
      "epoch": 4.47,
      "grad_norm": 1.7470178885999517,
      "learning_rate": 1.7530934685516753e-06,
      "loss": 0.1756,
      "step": 3911
    },
    {
      "epoch": 4.47,
      "grad_norm": 1.5477406659408472,
      "learning_rate": 1.7456192446118756e-06,
      "loss": 0.1677,
      "step": 3912
    },
    {
      "epoch": 4.47,
      "grad_norm": 1.362448658114318,
      "learning_rate": 1.73816051024542e-06,
      "loss": 0.1423,
      "step": 3913
    },
    {
      "epoch": 4.47,
      "grad_norm": 1.715869661443001,
      "learning_rate": 1.7307172695413242e-06,
      "loss": 0.1507,
      "step": 3914
    },
    {
      "epoch": 4.47,
      "grad_norm": 1.9009937606161402,
      "learning_rate": 1.723289526580114e-06,
      "loss": 0.1731,
      "step": 3915
    },
    {
      "epoch": 4.48,
      "grad_norm": 1.4844701433928642,
      "learning_rate": 1.7158772854338156e-06,
      "loss": 0.1373,
      "step": 3916
    },
    {
      "epoch": 4.48,
      "grad_norm": 1.9742908810068918,
      "learning_rate": 1.7084805501659484e-06,
      "loss": 0.2334,
      "step": 3917
    },
    {
      "epoch": 4.48,
      "grad_norm": 1.6975176634693963,
      "learning_rate": 1.7010993248315454e-06,
      "loss": 0.1755,
      "step": 3918
    },
    {
      "epoch": 4.48,
      "grad_norm": 1.6064898065758244,
      "learning_rate": 1.6937336134771265e-06,
      "loss": 0.168,
      "step": 3919
    },
    {
      "epoch": 4.48,
      "grad_norm": 1.7499500664028824,
      "learning_rate": 1.6863834201407147e-06,
      "loss": 0.1695,
      "step": 3920
    },
    {
      "epoch": 4.48,
      "grad_norm": 1.6354290209596087,
      "learning_rate": 1.6790487488518203e-06,
      "loss": 0.1746,
      "step": 3921
    },
    {
      "epoch": 4.48,
      "grad_norm": 1.7234223207251602,
      "learning_rate": 1.6717296036314467e-06,
      "loss": 0.1843,
      "step": 3922
    },
    {
      "epoch": 4.48,
      "grad_norm": 1.5942346075026534,
      "learning_rate": 1.6644259884920844e-06,
      "loss": 0.1665,
      "step": 3923
    },
    {
      "epoch": 4.48,
      "grad_norm": 1.6124844082948093,
      "learning_rate": 1.6571379074377135e-06,
      "loss": 0.1552,
      "step": 3924
    },
    {
      "epoch": 4.49,
      "grad_norm": 1.708164995446054,
      "learning_rate": 1.649865364463785e-06,
      "loss": 0.185,
      "step": 3925
    },
    {
      "epoch": 4.49,
      "grad_norm": 1.5192387520867707,
      "learning_rate": 1.6426083635572464e-06,
      "loss": 0.1476,
      "step": 3926
    },
    {
      "epoch": 4.49,
      "grad_norm": 1.5947464966035467,
      "learning_rate": 1.6353669086965216e-06,
      "loss": 0.163,
      "step": 3927
    },
    {
      "epoch": 4.49,
      "grad_norm": 1.641821800475039,
      "learning_rate": 1.6281410038515054e-06,
      "loss": 0.1671,
      "step": 3928
    },
    {
      "epoch": 4.49,
      "grad_norm": 1.4803049030190218,
      "learning_rate": 1.6209306529835787e-06,
      "loss": 0.1652,
      "step": 3929
    },
    {
      "epoch": 4.49,
      "grad_norm": 1.6084210939131975,
      "learning_rate": 1.6137358600455833e-06,
      "loss": 0.1622,
      "step": 3930
    },
    {
      "epoch": 4.49,
      "grad_norm": 1.6463004001102506,
      "learning_rate": 1.6065566289818402e-06,
      "loss": 0.1902,
      "step": 3931
    },
    {
      "epoch": 4.49,
      "grad_norm": 1.8577103154032104,
      "learning_rate": 1.5993929637281346e-06,
      "loss": 0.2025,
      "step": 3932
    },
    {
      "epoch": 4.49,
      "grad_norm": 1.6519701566613303,
      "learning_rate": 1.5922448682117285e-06,
      "loss": 0.1929,
      "step": 3933
    },
    {
      "epoch": 4.5,
      "grad_norm": 1.797513565015486,
      "learning_rate": 1.5851123463513272e-06,
      "loss": 0.1987,
      "step": 3934
    },
    {
      "epoch": 4.5,
      "grad_norm": 1.710712215508023,
      "learning_rate": 1.577995402057113e-06,
      "loss": 0.1546,
      "step": 3935
    },
    {
      "epoch": 4.5,
      "grad_norm": 2.0382146162510884,
      "learning_rate": 1.5708940392307313e-06,
      "loss": 0.1799,
      "step": 3936
    },
    {
      "epoch": 4.5,
      "grad_norm": 1.7593077035591815,
      "learning_rate": 1.563808261765285e-06,
      "loss": 0.2028,
      "step": 3937
    },
    {
      "epoch": 4.5,
      "grad_norm": 1.7753846211541655,
      "learning_rate": 1.5567380735453196e-06,
      "loss": 0.1561,
      "step": 3938
    },
    {
      "epoch": 4.5,
      "grad_norm": 1.5916905697114871,
      "learning_rate": 1.5496834784468416e-06,
      "loss": 0.154,
      "step": 3939
    },
    {
      "epoch": 4.5,
      "grad_norm": 1.7135287721376375,
      "learning_rate": 1.5426444803373174e-06,
      "loss": 0.1857,
      "step": 3940
    },
    {
      "epoch": 4.5,
      "grad_norm": 1.6121569910406923,
      "learning_rate": 1.5356210830756535e-06,
      "loss": 0.1711,
      "step": 3941
    },
    {
      "epoch": 4.51,
      "grad_norm": 1.5452631366359715,
      "learning_rate": 1.5286132905121964e-06,
      "loss": 0.1587,
      "step": 3942
    },
    {
      "epoch": 4.51,
      "grad_norm": 1.702156152949823,
      "learning_rate": 1.521621106488763e-06,
      "loss": 0.1653,
      "step": 3943
    },
    {
      "epoch": 4.51,
      "grad_norm": 1.9008339496712734,
      "learning_rate": 1.5146445348385872e-06,
      "loss": 0.1854,
      "step": 3944
    },
    {
      "epoch": 4.51,
      "grad_norm": 1.653406230721559,
      "learning_rate": 1.507683579386363e-06,
      "loss": 0.1634,
      "step": 3945
    },
    {
      "epoch": 4.51,
      "grad_norm": 1.604078326134846,
      "learning_rate": 1.5007382439482077e-06,
      "loss": 0.1681,
      "step": 3946
    },
    {
      "epoch": 4.51,
      "grad_norm": 1.467398516069849,
      "learning_rate": 1.4938085323316818e-06,
      "loss": 0.1576,
      "step": 3947
    },
    {
      "epoch": 4.51,
      "grad_norm": 1.6229853284322688,
      "learning_rate": 1.4868944483357839e-06,
      "loss": 0.1624,
      "step": 3948
    },
    {
      "epoch": 4.51,
      "grad_norm": 1.665363611896342,
      "learning_rate": 1.4799959957509413e-06,
      "loss": 0.1972,
      "step": 3949
    },
    {
      "epoch": 4.51,
      "grad_norm": 1.8513715495328429,
      "learning_rate": 1.4731131783590156e-06,
      "loss": 0.2031,
      "step": 3950
    },
    {
      "epoch": 4.52,
      "grad_norm": 2.0881771540265586,
      "learning_rate": 1.4662459999332922e-06,
      "loss": 0.1944,
      "step": 3951
    },
    {
      "epoch": 4.52,
      "grad_norm": 1.9717727185935678,
      "learning_rate": 1.459394464238486e-06,
      "loss": 0.2083,
      "step": 3952
    },
    {
      "epoch": 4.52,
      "grad_norm": 1.6724585364431492,
      "learning_rate": 1.4525585750307357e-06,
      "loss": 0.1707,
      "step": 3953
    },
    {
      "epoch": 4.52,
      "grad_norm": 2.2490174148676125,
      "learning_rate": 1.445738336057607e-06,
      "loss": 0.1989,
      "step": 3954
    },
    {
      "epoch": 4.52,
      "grad_norm": 1.69523707729287,
      "learning_rate": 1.4389337510580692e-06,
      "loss": 0.1633,
      "step": 3955
    },
    {
      "epoch": 4.52,
      "grad_norm": 1.600228089445157,
      "learning_rate": 1.432144823762528e-06,
      "loss": 0.1827,
      "step": 3956
    },
    {
      "epoch": 4.52,
      "grad_norm": 1.5796147544551506,
      "learning_rate": 1.4253715578927994e-06,
      "loss": 0.1795,
      "step": 3957
    },
    {
      "epoch": 4.52,
      "grad_norm": 1.4458941637221985,
      "learning_rate": 1.4186139571621104e-06,
      "loss": 0.1498,
      "step": 3958
    },
    {
      "epoch": 4.52,
      "grad_norm": 1.6909333752042315,
      "learning_rate": 1.4118720252751039e-06,
      "loss": 0.1853,
      "step": 3959
    },
    {
      "epoch": 4.53,
      "grad_norm": 1.4993684008182941,
      "learning_rate": 1.4051457659278333e-06,
      "loss": 0.1554,
      "step": 3960
    },
    {
      "epoch": 4.53,
      "grad_norm": 1.6950753607995859,
      "learning_rate": 1.3984351828077558e-06,
      "loss": 0.1812,
      "step": 3961
    },
    {
      "epoch": 4.53,
      "grad_norm": 1.6989477201236998,
      "learning_rate": 1.3917402795937384e-06,
      "loss": 0.1791,
      "step": 3962
    },
    {
      "epoch": 4.53,
      "grad_norm": 1.656091739628196,
      "learning_rate": 1.3850610599560488e-06,
      "loss": 0.1808,
      "step": 3963
    },
    {
      "epoch": 4.53,
      "grad_norm": 1.5826423586476226,
      "learning_rate": 1.378397527556361e-06,
      "loss": 0.1318,
      "step": 3964
    },
    {
      "epoch": 4.53,
      "grad_norm": 1.7652491397611447,
      "learning_rate": 1.371749686047743e-06,
      "loss": 0.1681,
      "step": 3965
    },
    {
      "epoch": 4.53,
      "grad_norm": 1.7300264916795518,
      "learning_rate": 1.3651175390746696e-06,
      "loss": 0.1906,
      "step": 3966
    },
    {
      "epoch": 4.53,
      "grad_norm": 2.017842183941523,
      "learning_rate": 1.3585010902729989e-06,
      "loss": 0.2025,
      "step": 3967
    },
    {
      "epoch": 4.53,
      "grad_norm": 1.59631641695133,
      "learning_rate": 1.3519003432699962e-06,
      "loss": 0.1712,
      "step": 3968
    },
    {
      "epoch": 4.54,
      "grad_norm": 1.8612184159752005,
      "learning_rate": 1.3453153016843135e-06,
      "loss": 0.1716,
      "step": 3969
    },
    {
      "epoch": 4.54,
      "grad_norm": 1.668258709179667,
      "learning_rate": 1.3387459691259862e-06,
      "loss": 0.1804,
      "step": 3970
    },
    {
      "epoch": 4.54,
      "grad_norm": 1.7993859862957684,
      "learning_rate": 1.3321923491964539e-06,
      "loss": 0.1856,
      "step": 3971
    },
    {
      "epoch": 4.54,
      "grad_norm": 1.7671046318496457,
      "learning_rate": 1.3256544454885221e-06,
      "loss": 0.1868,
      "step": 3972
    },
    {
      "epoch": 4.54,
      "grad_norm": 1.6202496097025165,
      "learning_rate": 1.319132261586391e-06,
      "loss": 0.1513,
      "step": 3973
    },
    {
      "epoch": 4.54,
      "grad_norm": 1.6950828134399187,
      "learning_rate": 1.3126258010656466e-06,
      "loss": 0.1361,
      "step": 3974
    },
    {
      "epoch": 4.54,
      "grad_norm": 1.8712084681922467,
      "learning_rate": 1.3061350674932526e-06,
      "loss": 0.2388,
      "step": 3975
    },
    {
      "epoch": 4.54,
      "grad_norm": 1.8530287460554544,
      "learning_rate": 1.2996600644275424e-06,
      "loss": 0.1946,
      "step": 3976
    },
    {
      "epoch": 4.55,
      "grad_norm": 1.7597506557772802,
      "learning_rate": 1.2932007954182367e-06,
      "loss": 0.1689,
      "step": 3977
    },
    {
      "epoch": 4.55,
      "grad_norm": 1.292841324251992,
      "learning_rate": 1.2867572640064263e-06,
      "loss": 0.1299,
      "step": 3978
    },
    {
      "epoch": 4.55,
      "grad_norm": 1.6397553170189763,
      "learning_rate": 1.2803294737245752e-06,
      "loss": 0.1824,
      "step": 3979
    },
    {
      "epoch": 4.55,
      "grad_norm": 1.704715755313884,
      "learning_rate": 1.2739174280965117e-06,
      "loss": 0.1603,
      "step": 3980
    },
    {
      "epoch": 4.55,
      "grad_norm": 1.5106918927886992,
      "learning_rate": 1.267521130637437e-06,
      "loss": 0.1468,
      "step": 3981
    },
    {
      "epoch": 4.55,
      "grad_norm": 2.0297596380257255,
      "learning_rate": 1.2611405848539226e-06,
      "loss": 0.2058,
      "step": 3982
    },
    {
      "epoch": 4.55,
      "grad_norm": 1.8839850548207473,
      "learning_rate": 1.2547757942439075e-06,
      "loss": 0.1933,
      "step": 3983
    },
    {
      "epoch": 4.55,
      "grad_norm": 1.7238241406307995,
      "learning_rate": 1.2484267622966772e-06,
      "loss": 0.191,
      "step": 3984
    },
    {
      "epoch": 4.55,
      "grad_norm": 1.4778910614414653,
      "learning_rate": 1.242093492492894e-06,
      "loss": 0.1458,
      "step": 3985
    },
    {
      "epoch": 4.56,
      "grad_norm": 1.5759770765396623,
      "learning_rate": 1.2357759883045738e-06,
      "loss": 0.1447,
      "step": 3986
    },
    {
      "epoch": 4.56,
      "grad_norm": 1.5660194531066942,
      "learning_rate": 1.2294742531950864e-06,
      "loss": 0.194,
      "step": 3987
    },
    {
      "epoch": 4.56,
      "grad_norm": 1.4814182487241383,
      "learning_rate": 1.2231882906191617e-06,
      "loss": 0.1523,
      "step": 3988
    },
    {
      "epoch": 4.56,
      "grad_norm": 1.6806336864719533,
      "learning_rate": 1.216918104022876e-06,
      "loss": 0.1786,
      "step": 3989
    },
    {
      "epoch": 4.56,
      "grad_norm": 1.8649983725244876,
      "learning_rate": 1.2106636968436636e-06,
      "loss": 0.1925,
      "step": 3990
    },
    {
      "epoch": 4.56,
      "grad_norm": 1.6141465592205102,
      "learning_rate": 1.2044250725103113e-06,
      "loss": 0.1495,
      "step": 3991
    },
    {
      "epoch": 4.56,
      "grad_norm": 1.5407436076803396,
      "learning_rate": 1.1982022344429433e-06,
      "loss": 0.1608,
      "step": 3992
    },
    {
      "epoch": 4.56,
      "grad_norm": 1.8233482414597937,
      "learning_rate": 1.1919951860530343e-06,
      "loss": 0.2077,
      "step": 3993
    },
    {
      "epoch": 4.56,
      "grad_norm": 1.7111809018527957,
      "learning_rate": 1.185803930743402e-06,
      "loss": 0.1782,
      "step": 3994
    },
    {
      "epoch": 4.57,
      "grad_norm": 1.6720851556948202,
      "learning_rate": 1.1796284719082084e-06,
      "loss": 0.1313,
      "step": 3995
    },
    {
      "epoch": 4.57,
      "grad_norm": 1.7855395370917473,
      "learning_rate": 1.1734688129329585e-06,
      "loss": 0.193,
      "step": 3996
    },
    {
      "epoch": 4.57,
      "grad_norm": 1.589652194107256,
      "learning_rate": 1.1673249571944811e-06,
      "loss": 0.1564,
      "step": 3997
    },
    {
      "epoch": 4.57,
      "grad_norm": 1.84327567836781,
      "learning_rate": 1.1611969080609586e-06,
      "loss": 0.1615,
      "step": 3998
    },
    {
      "epoch": 4.57,
      "grad_norm": 1.664946520566539,
      "learning_rate": 1.1550846688919003e-06,
      "loss": 0.1751,
      "step": 3999
    },
    {
      "epoch": 4.57,
      "grad_norm": 1.7274055611454835,
      "learning_rate": 1.1489882430381493e-06,
      "loss": 0.1975,
      "step": 4000
    },
    {
      "epoch": 4.57,
      "eval_blimp_filtered_avg": 0.7080597014925373,
      "eval_blimp_filtered_std": 0.005076355892422563,
      "step": 4000
    },
    {
      "epoch": 4.57,
      "eval_blimp_supplement_avg": 0.8297413793103449,
      "eval_blimp_supplement_std": 0.016984707085279518,
      "step": 4000
    },
    {
      "epoch": 4.57,
      "eval_vqa_filtered_avg": 0.55,
      "eval_vqa_filtered_std": 0.049999999999999996,
      "step": 4000
    },
    {
      "epoch": 4.57,
      "eval_winoground_filtered_avg": 0.62,
      "eval_winoground_filtered_std": 0.04878317312145633,
      "step": 4000
    },
    {
      "epoch": 4.57,
      "grad_norm": 1.931600834850389,
      "learning_rate": 1.142907633841872e-06,
      "loss": 0.2066,
      "step": 4001
    },
    {
      "epoch": 4.57,
      "grad_norm": 1.4980776909462328,
      "learning_rate": 1.1368428446365786e-06,
      "loss": 0.1764,
      "step": 4002
    },
    {
      "epoch": 4.57,
      "grad_norm": 1.4926290676582301,
      "learning_rate": 1.1307938787470929e-06,
      "loss": 0.1552,
      "step": 4003
    },
    {
      "epoch": 4.58,
      "grad_norm": 2.085130797598516,
      "learning_rate": 1.1247607394895753e-06,
      "loss": 0.2119,
      "step": 4004
    },
    {
      "epoch": 4.58,
      "grad_norm": 1.5038271394742293,
      "learning_rate": 1.1187434301715004e-06,
      "loss": 0.1398,
      "step": 4005
    },
    {
      "epoch": 4.58,
      "grad_norm": 1.824491560533093,
      "learning_rate": 1.1127419540916694e-06,
      "loss": 0.1766,
      "step": 4006
    },
    {
      "epoch": 4.58,
      "grad_norm": 1.7437204709945326,
      "learning_rate": 1.1067563145402004e-06,
      "loss": 0.1939,
      "step": 4007
    },
    {
      "epoch": 4.58,
      "grad_norm": 1.8709355890249975,
      "learning_rate": 1.100786514798532e-06,
      "loss": 0.199,
      "step": 4008
    },
    {
      "epoch": 4.58,
      "grad_norm": 1.651757726760737,
      "learning_rate": 1.0948325581394259e-06,
      "loss": 0.19,
      "step": 4009
    },
    {
      "epoch": 4.58,
      "grad_norm": 1.581364175384204,
      "learning_rate": 1.0888944478269413e-06,
      "loss": 0.1533,
      "step": 4010
    },
    {
      "epoch": 4.58,
      "grad_norm": 1.711314521519743,
      "learning_rate": 1.0829721871164643e-06,
      "loss": 0.1726,
      "step": 4011
    },
    {
      "epoch": 4.59,
      "grad_norm": 1.7094209934781743,
      "learning_rate": 1.0770657792546845e-06,
      "loss": 0.1754,
      "step": 4012
    },
    {
      "epoch": 4.59,
      "grad_norm": 1.8734968783535153,
      "learning_rate": 1.0711752274796083e-06,
      "loss": 0.2126,
      "step": 4013
    },
    {
      "epoch": 4.59,
      "grad_norm": 1.5360652429494819,
      "learning_rate": 1.0653005350205424e-06,
      "loss": 0.1341,
      "step": 4014
    },
    {
      "epoch": 4.59,
      "grad_norm": 1.9095886035342458,
      "learning_rate": 1.059441705098101e-06,
      "loss": 0.1938,
      "step": 4015
    },
    {
      "epoch": 4.59,
      "grad_norm": 1.6535525834928424,
      "learning_rate": 1.0535987409242043e-06,
      "loss": 0.1678,
      "step": 4016
    },
    {
      "epoch": 4.59,
      "grad_norm": 1.940343720340287,
      "learning_rate": 1.0477716457020803e-06,
      "loss": 0.1914,
      "step": 4017
    },
    {
      "epoch": 4.59,
      "grad_norm": 1.4937618131869408,
      "learning_rate": 1.0419604226262336e-06,
      "loss": 0.1669,
      "step": 4018
    },
    {
      "epoch": 4.59,
      "grad_norm": 1.554584238006475,
      "learning_rate": 1.0361650748824991e-06,
      "loss": 0.1401,
      "step": 4019
    },
    {
      "epoch": 4.59,
      "grad_norm": 1.6672536019751836,
      "learning_rate": 1.0303856056479822e-06,
      "loss": 0.1858,
      "step": 4020
    },
    {
      "epoch": 4.6,
      "grad_norm": 1.4342157504333215,
      "learning_rate": 1.0246220180911015e-06,
      "loss": 0.1486,
      "step": 4021
    },
    {
      "epoch": 4.6,
      "grad_norm": 1.6989938949223848,
      "learning_rate": 1.0188743153715629e-06,
      "loss": 0.1616,
      "step": 4022
    },
    {
      "epoch": 4.6,
      "grad_norm": 1.9557564436260257,
      "learning_rate": 1.0131425006403627e-06,
      "loss": 0.217,
      "step": 4023
    },
    {
      "epoch": 4.6,
      "grad_norm": 1.7661745243987428,
      "learning_rate": 1.007426577039784e-06,
      "loss": 0.2221,
      "step": 4024
    },
    {
      "epoch": 4.6,
      "grad_norm": 1.6733367775950827,
      "learning_rate": 1.0017265477033998e-06,
      "loss": 0.1696,
      "step": 4025
    },
    {
      "epoch": 4.6,
      "grad_norm": 1.6260445727285806,
      "learning_rate": 9.960424157560844e-07,
      "loss": 0.1718,
      "step": 4026
    },
    {
      "epoch": 4.6,
      "grad_norm": 1.4999585186141624,
      "learning_rate": 9.90374184313968e-07,
      "loss": 0.1622,
      "step": 4027
    },
    {
      "epoch": 4.6,
      "grad_norm": 1.7076409153446404,
      "learning_rate": 9.847218564844884e-07,
      "loss": 0.1722,
      "step": 4028
    },
    {
      "epoch": 4.6,
      "grad_norm": 1.6191127551059803,
      "learning_rate": 9.790854353663503e-07,
      "loss": 0.1605,
      "step": 4029
    },
    {
      "epoch": 4.61,
      "grad_norm": 1.5924856135354057,
      "learning_rate": 9.73464924049552e-07,
      "loss": 0.1521,
      "step": 4030
    },
    {
      "epoch": 4.61,
      "grad_norm": 1.7412036070600552,
      "learning_rate": 9.678603256153585e-07,
      "loss": 0.1815,
      "step": 4031
    },
    {
      "epoch": 4.61,
      "grad_norm": 1.8526129533783429,
      "learning_rate": 9.622716431363088e-07,
      "loss": 0.2062,
      "step": 4032
    },
    {
      "epoch": 4.61,
      "grad_norm": 1.6727884616358468,
      "learning_rate": 9.56698879676229e-07,
      "loss": 0.1633,
      "step": 4033
    },
    {
      "epoch": 4.61,
      "grad_norm": 1.4118725085143675,
      "learning_rate": 9.511420382902114e-07,
      "loss": 0.1585,
      "step": 4034
    },
    {
      "epoch": 4.61,
      "grad_norm": 1.7416769502659841,
      "learning_rate": 9.45601122024613e-07,
      "loss": 0.2069,
      "step": 4035
    },
    {
      "epoch": 4.61,
      "grad_norm": 2.0105982185599176,
      "learning_rate": 9.400761339170639e-07,
      "loss": 0.2113,
      "step": 4036
    },
    {
      "epoch": 4.61,
      "grad_norm": 1.6560241912596207,
      "learning_rate": 9.345670769964742e-07,
      "loss": 0.1592,
      "step": 4037
    },
    {
      "epoch": 4.61,
      "grad_norm": 1.839594028396661,
      "learning_rate": 9.290739542830051e-07,
      "loss": 0.1826,
      "step": 4038
    },
    {
      "epoch": 4.62,
      "grad_norm": 1.9938359079602435,
      "learning_rate": 9.235967687880941e-07,
      "loss": 0.1978,
      "step": 4039
    },
    {
      "epoch": 4.62,
      "grad_norm": 1.4779139809700963,
      "learning_rate": 9.181355235144262e-07,
      "loss": 0.1756,
      "step": 4040
    },
    {
      "epoch": 4.62,
      "grad_norm": 1.6884515964906324,
      "learning_rate": 9.126902214559629e-07,
      "loss": 0.1802,
      "step": 4041
    },
    {
      "epoch": 4.62,
      "grad_norm": 1.700354857405424,
      "learning_rate": 9.072608655979164e-07,
      "loss": 0.1773,
      "step": 4042
    },
    {
      "epoch": 4.62,
      "grad_norm": 1.8980848638812369,
      "learning_rate": 9.018474589167591e-07,
      "loss": 0.1964,
      "step": 4043
    },
    {
      "epoch": 4.62,
      "grad_norm": 1.6155989002046396,
      "learning_rate": 8.964500043802271e-07,
      "loss": 0.1466,
      "step": 4044
    },
    {
      "epoch": 4.62,
      "grad_norm": 1.6009448574254344,
      "learning_rate": 8.910685049472967e-07,
      "loss": 0.1664,
      "step": 4045
    },
    {
      "epoch": 4.62,
      "grad_norm": 1.8208911667675858,
      "learning_rate": 8.857029635682112e-07,
      "loss": 0.1688,
      "step": 4046
    },
    {
      "epoch": 4.63,
      "grad_norm": 1.4286352824423236,
      "learning_rate": 8.803533831844579e-07,
      "loss": 0.1519,
      "step": 4047
    },
    {
      "epoch": 4.63,
      "grad_norm": 1.5701216974676369,
      "learning_rate": 8.750197667287707e-07,
      "loss": 0.1374,
      "step": 4048
    },
    {
      "epoch": 4.63,
      "grad_norm": 1.770588278316926,
      "learning_rate": 8.69702117125144e-07,
      "loss": 0.1917,
      "step": 4049
    },
    {
      "epoch": 4.63,
      "grad_norm": 1.9670190589270773,
      "learning_rate": 8.644004372888059e-07,
      "loss": 0.2088,
      "step": 4050
    },
    {
      "epoch": 4.63,
      "grad_norm": 1.7078835831398982,
      "learning_rate": 8.591147301262381e-07,
      "loss": 0.1665,
      "step": 4051
    },
    {
      "epoch": 4.63,
      "grad_norm": 2.103801546681517,
      "learning_rate": 8.538449985351627e-07,
      "loss": 0.1758,
      "step": 4052
    },
    {
      "epoch": 4.63,
      "grad_norm": 2.025408686531091,
      "learning_rate": 8.485912454045419e-07,
      "loss": 0.2009,
      "step": 4053
    },
    {
      "epoch": 4.63,
      "grad_norm": 1.5021731352010161,
      "learning_rate": 8.433534736145821e-07,
      "loss": 0.1517,
      "step": 4054
    },
    {
      "epoch": 4.63,
      "grad_norm": 1.6662192948349954,
      "learning_rate": 8.381316860367294e-07,
      "loss": 0.1596,
      "step": 4055
    },
    {
      "epoch": 4.64,
      "grad_norm": 1.7248772761779576,
      "learning_rate": 8.329258855336574e-07,
      "loss": 0.161,
      "step": 4056
    },
    {
      "epoch": 4.64,
      "grad_norm": 1.6875160744310185,
      "learning_rate": 8.277360749592866e-07,
      "loss": 0.1674,
      "step": 4057
    },
    {
      "epoch": 4.64,
      "grad_norm": 1.610527642396192,
      "learning_rate": 8.225622571587643e-07,
      "loss": 0.1776,
      "step": 4058
    },
    {
      "epoch": 4.64,
      "grad_norm": 1.5697621582254087,
      "learning_rate": 8.174044349684784e-07,
      "loss": 0.1763,
      "step": 4059
    },
    {
      "epoch": 4.64,
      "grad_norm": 1.8597160337699385,
      "learning_rate": 8.122626112160336e-07,
      "loss": 0.1979,
      "step": 4060
    },
    {
      "epoch": 4.64,
      "grad_norm": 1.7696796090396307,
      "learning_rate": 8.071367887202819e-07,
      "loss": 0.1682,
      "step": 4061
    },
    {
      "epoch": 4.64,
      "grad_norm": 1.6426874704322243,
      "learning_rate": 8.020269702912886e-07,
      "loss": 0.1761,
      "step": 4062
    },
    {
      "epoch": 4.64,
      "grad_norm": 1.7248393449704518,
      "learning_rate": 7.969331587303497e-07,
      "loss": 0.1733,
      "step": 4063
    },
    {
      "epoch": 4.64,
      "grad_norm": 1.652995537508509,
      "learning_rate": 7.918553568299913e-07,
      "loss": 0.1746,
      "step": 4064
    },
    {
      "epoch": 4.65,
      "grad_norm": 1.6219171844351,
      "learning_rate": 7.867935673739535e-07,
      "loss": 0.1675,
      "step": 4065
    },
    {
      "epoch": 4.65,
      "grad_norm": 1.824889195223463,
      "learning_rate": 7.817477931372063e-07,
      "loss": 0.2058,
      "step": 4066
    },
    {
      "epoch": 4.65,
      "grad_norm": 1.7485836270628718,
      "learning_rate": 7.767180368859306e-07,
      "loss": 0.169,
      "step": 4067
    },
    {
      "epoch": 4.65,
      "grad_norm": 1.6998021897719668,
      "learning_rate": 7.717043013775405e-07,
      "loss": 0.1877,
      "step": 4068
    },
    {
      "epoch": 4.65,
      "grad_norm": 1.5278783834242422,
      "learning_rate": 7.667065893606506e-07,
      "loss": 0.1519,
      "step": 4069
    },
    {
      "epoch": 4.65,
      "grad_norm": 1.7898353842836094,
      "learning_rate": 7.617249035751028e-07,
      "loss": 0.1785,
      "step": 4070
    },
    {
      "epoch": 4.65,
      "grad_norm": 1.5403014005957336,
      "learning_rate": 7.56759246751949e-07,
      "loss": 0.166,
      "step": 4071
    },
    {
      "epoch": 4.65,
      "grad_norm": 1.7698742346483354,
      "learning_rate": 7.518096216134552e-07,
      "loss": 0.1917,
      "step": 4072
    },
    {
      "epoch": 4.65,
      "grad_norm": 1.3488175373697622,
      "learning_rate": 7.468760308730938e-07,
      "loss": 0.139,
      "step": 4073
    },
    {
      "epoch": 4.66,
      "grad_norm": 1.731679052739123,
      "learning_rate": 7.419584772355515e-07,
      "loss": 0.1711,
      "step": 4074
    },
    {
      "epoch": 4.66,
      "grad_norm": 1.7565381215232179,
      "learning_rate": 7.370569633967217e-07,
      "loss": 0.1828,
      "step": 4075
    },
    {
      "epoch": 4.66,
      "grad_norm": 1.5935760079129373,
      "learning_rate": 7.321714920437051e-07,
      "loss": 0.1481,
      "step": 4076
    },
    {
      "epoch": 4.66,
      "grad_norm": 1.6284366735166198,
      "learning_rate": 7.273020658548124e-07,
      "loss": 0.1591,
      "step": 4077
    },
    {
      "epoch": 4.66,
      "grad_norm": 2.0681498443551245,
      "learning_rate": 7.224486874995451e-07,
      "loss": 0.2332,
      "step": 4078
    },
    {
      "epoch": 4.66,
      "grad_norm": 1.5039282009389316,
      "learning_rate": 7.176113596386214e-07,
      "loss": 0.1517,
      "step": 4079
    },
    {
      "epoch": 4.66,
      "grad_norm": 1.5149668162283836,
      "learning_rate": 7.127900849239466e-07,
      "loss": 0.149,
      "step": 4080
    },
    {
      "epoch": 4.66,
      "grad_norm": 1.7742567247015328,
      "learning_rate": 7.079848659986433e-07,
      "loss": 0.1702,
      "step": 4081
    },
    {
      "epoch": 4.67,
      "grad_norm": 1.6802385722510147,
      "learning_rate": 7.031957054970073e-07,
      "loss": 0.1644,
      "step": 4082
    },
    {
      "epoch": 4.67,
      "grad_norm": 1.7715587327012614,
      "learning_rate": 6.98422606044552e-07,
      "loss": 0.1786,
      "step": 4083
    },
    {
      "epoch": 4.67,
      "grad_norm": 1.7353756573503913,
      "learning_rate": 6.936655702579808e-07,
      "loss": 0.1919,
      "step": 4084
    },
    {
      "epoch": 4.67,
      "grad_norm": 1.9576618890886268,
      "learning_rate": 6.889246007451911e-07,
      "loss": 0.2025,
      "step": 4085
    },
    {
      "epoch": 4.67,
      "grad_norm": 1.8524366377104768,
      "learning_rate": 6.841997001052603e-07,
      "loss": 0.1722,
      "step": 4086
    },
    {
      "epoch": 4.67,
      "grad_norm": 1.6796608398695736,
      "learning_rate": 6.794908709284731e-07,
      "loss": 0.1764,
      "step": 4087
    },
    {
      "epoch": 4.67,
      "grad_norm": 1.6352140329999119,
      "learning_rate": 6.747981157962913e-07,
      "loss": 0.1371,
      "step": 4088
    },
    {
      "epoch": 4.67,
      "grad_norm": 1.7407258262450536,
      "learning_rate": 6.701214372813802e-07,
      "loss": 0.1629,
      "step": 4089
    },
    {
      "epoch": 4.67,
      "grad_norm": 1.8941973435660886,
      "learning_rate": 6.654608379475657e-07,
      "loss": 0.1909,
      "step": 4090
    },
    {
      "epoch": 4.68,
      "grad_norm": 1.9017183563243754,
      "learning_rate": 6.608163203498907e-07,
      "loss": 0.1904,
      "step": 4091
    },
    {
      "epoch": 4.68,
      "grad_norm": 1.5820684169951538,
      "learning_rate": 6.56187887034555e-07,
      "loss": 0.1543,
      "step": 4092
    },
    {
      "epoch": 4.68,
      "grad_norm": 1.6419470701979069,
      "learning_rate": 6.515755405389523e-07,
      "loss": 0.1741,
      "step": 4093
    },
    {
      "epoch": 4.68,
      "grad_norm": 1.593365748824792,
      "learning_rate": 6.469792833916632e-07,
      "loss": 0.1753,
      "step": 4094
    },
    {
      "epoch": 4.68,
      "grad_norm": 1.6562337959096531,
      "learning_rate": 6.423991181124355e-07,
      "loss": 0.154,
      "step": 4095
    },
    {
      "epoch": 4.68,
      "grad_norm": 1.446989163023385,
      "learning_rate": 6.378350472121975e-07,
      "loss": 0.1524,
      "step": 4096
    },
    {
      "epoch": 4.68,
      "grad_norm": 1.6992940161795667,
      "learning_rate": 6.332870731930607e-07,
      "loss": 0.1629,
      "step": 4097
    },
    {
      "epoch": 4.68,
      "grad_norm": 1.773871913461549,
      "learning_rate": 6.287551985483075e-07,
      "loss": 0.1671,
      "step": 4098
    },
    {
      "epoch": 4.68,
      "grad_norm": 1.7727937658140043,
      "learning_rate": 6.242394257623974e-07,
      "loss": 0.1771,
      "step": 4099
    },
    {
      "epoch": 4.69,
      "grad_norm": 1.5382941511809407,
      "learning_rate": 6.197397573109565e-07,
      "loss": 0.1517,
      "step": 4100
    },
    {
      "epoch": 4.69,
      "eval_blimp_filtered_avg": 0.7083582089552238,
      "eval_blimp_filtered_std": 0.005072631102482144,
      "step": 4100
    },
    {
      "epoch": 4.69,
      "eval_blimp_supplement_avg": 0.8254310344827587,
      "eval_blimp_supplement_std": 0.017125114504825824,
      "step": 4100
    },
    {
      "epoch": 4.69,
      "eval_vqa_filtered_avg": 0.55,
      "eval_vqa_filtered_std": 0.049999999999999996,
      "step": 4100
    },
    {
      "epoch": 4.69,
      "eval_winoground_filtered_avg": 0.63,
      "eval_winoground_filtered_std": 0.04852365870939099,
      "step": 4100
    },
    {
      "epoch": 4.69,
      "grad_norm": 1.6748924753738572,
      "learning_rate": 6.152561956607916e-07,
      "loss": 0.1757,
      "step": 4101
    },
    {
      "epoch": 4.69,
      "grad_norm": 1.728021063962419,
      "learning_rate": 6.1078874326987e-07,
      "loss": 0.1726,
      "step": 4102
    },
    {
      "epoch": 4.69,
      "grad_norm": 1.5840773666037529,
      "learning_rate": 6.063374025873325e-07,
      "loss": 0.1594,
      "step": 4103
    },
    {
      "epoch": 4.69,
      "grad_norm": 1.718563043307126,
      "learning_rate": 6.01902176053487e-07,
      "loss": 0.2007,
      "step": 4104
    },
    {
      "epoch": 4.69,
      "grad_norm": 1.6481428568043974,
      "learning_rate": 5.974830660998021e-07,
      "loss": 0.1766,
      "step": 4105
    },
    {
      "epoch": 4.69,
      "grad_norm": 1.4731354935182221,
      "learning_rate": 5.930800751489229e-07,
      "loss": 0.1459,
      "step": 4106
    },
    {
      "epoch": 4.69,
      "grad_norm": 1.7315385267160621,
      "learning_rate": 5.886932056146454e-07,
      "loss": 0.1991,
      "step": 4107
    },
    {
      "epoch": 4.69,
      "grad_norm": 1.8466093201173261,
      "learning_rate": 5.843224599019325e-07,
      "loss": 0.2393,
      "step": 4108
    },
    {
      "epoch": 4.7,
      "grad_norm": 1.6563147991339584,
      "learning_rate": 5.799678404069109e-07,
      "loss": 0.1734,
      "step": 4109
    },
    {
      "epoch": 4.7,
      "grad_norm": 1.748332787725972,
      "learning_rate": 5.756293495168608e-07,
      "loss": 0.1644,
      "step": 4110
    },
    {
      "epoch": 4.7,
      "grad_norm": 1.8412766552983522,
      "learning_rate": 5.713069896102263e-07,
      "loss": 0.1945,
      "step": 4111
    },
    {
      "epoch": 4.7,
      "grad_norm": 1.575135679863595,
      "learning_rate": 5.670007630566021e-07,
      "loss": 0.1834,
      "step": 4112
    },
    {
      "epoch": 4.7,
      "grad_norm": 1.793307852188892,
      "learning_rate": 5.627106722167396e-07,
      "loss": 0.1953,
      "step": 4113
    },
    {
      "epoch": 4.7,
      "grad_norm": 1.7875729188370852,
      "learning_rate": 5.584367194425477e-07,
      "loss": 0.1739,
      "step": 4114
    },
    {
      "epoch": 4.7,
      "grad_norm": 1.5836307456963465,
      "learning_rate": 5.541789070770852e-07,
      "loss": 0.1713,
      "step": 4115
    },
    {
      "epoch": 4.7,
      "grad_norm": 1.5833272905640778,
      "learning_rate": 5.499372374545652e-07,
      "loss": 0.1824,
      "step": 4116
    },
    {
      "epoch": 4.71,
      "grad_norm": 1.6557380104450752,
      "learning_rate": 5.457117129003475e-07,
      "loss": 0.1777,
      "step": 4117
    },
    {
      "epoch": 4.71,
      "grad_norm": 1.5272302687352872,
      "learning_rate": 5.415023357309456e-07,
      "loss": 0.1477,
      "step": 4118
    },
    {
      "epoch": 4.71,
      "grad_norm": 1.6656862804488692,
      "learning_rate": 5.373091082540172e-07,
      "loss": 0.1616,
      "step": 4119
    },
    {
      "epoch": 4.71,
      "grad_norm": 1.446343881718617,
      "learning_rate": 5.331320327683664e-07,
      "loss": 0.1358,
      "step": 4120
    },
    {
      "epoch": 4.71,
      "grad_norm": 1.9709258107835732,
      "learning_rate": 5.289711115639384e-07,
      "loss": 0.2244,
      "step": 4121
    },
    {
      "epoch": 4.71,
      "grad_norm": 1.595945063153684,
      "learning_rate": 5.248263469218351e-07,
      "loss": 0.1679,
      "step": 4122
    },
    {
      "epoch": 4.71,
      "grad_norm": 1.615475065807381,
      "learning_rate": 5.206977411142855e-07,
      "loss": 0.1663,
      "step": 4123
    },
    {
      "epoch": 4.71,
      "grad_norm": 1.8267928242776135,
      "learning_rate": 5.165852964046724e-07,
      "loss": 0.1979,
      "step": 4124
    },
    {
      "epoch": 4.71,
      "grad_norm": 1.729700643890717,
      "learning_rate": 5.124890150475159e-07,
      "loss": 0.1839,
      "step": 4125
    },
    {
      "epoch": 4.72,
      "grad_norm": 1.4456494499397914,
      "learning_rate": 5.084088992884661e-07,
      "loss": 0.1357,
      "step": 4126
    },
    {
      "epoch": 4.72,
      "grad_norm": 1.6797178418080758,
      "learning_rate": 5.043449513643239e-07,
      "loss": 0.1937,
      "step": 4127
    },
    {
      "epoch": 4.72,
      "grad_norm": 1.9957905753339638,
      "learning_rate": 5.002971735030171e-07,
      "loss": 0.1971,
      "step": 4128
    },
    {
      "epoch": 4.72,
      "grad_norm": 1.514198991222978,
      "learning_rate": 4.962655679236139e-07,
      "loss": 0.1569,
      "step": 4129
    },
    {
      "epoch": 4.72,
      "grad_norm": 1.93549972177563,
      "learning_rate": 4.92250136836313e-07,
      "loss": 0.1939,
      "step": 4130
    },
    {
      "epoch": 4.72,
      "grad_norm": 1.7661041224417735,
      "learning_rate": 4.882508824424503e-07,
      "loss": 0.1763,
      "step": 4131
    },
    {
      "epoch": 4.72,
      "grad_norm": 1.7068490095149578,
      "learning_rate": 4.842678069344952e-07,
      "loss": 0.1455,
      "step": 4132
    },
    {
      "epoch": 4.72,
      "grad_norm": 1.6759870649657793,
      "learning_rate": 4.803009124960345e-07,
      "loss": 0.1734,
      "step": 4133
    },
    {
      "epoch": 4.72,
      "grad_norm": 1.6779563984432133,
      "learning_rate": 4.763502013017951e-07,
      "loss": 0.1694,
      "step": 4134
    },
    {
      "epoch": 4.73,
      "grad_norm": 1.605588836637454,
      "learning_rate": 4.7241567551763123e-07,
      "loss": 0.1642,
      "step": 4135
    },
    {
      "epoch": 4.73,
      "grad_norm": 1.659913600316028,
      "learning_rate": 4.6849733730052747e-07,
      "loss": 0.1804,
      "step": 4136
    },
    {
      "epoch": 4.73,
      "grad_norm": 1.8518076411158613,
      "learning_rate": 4.645951887985789e-07,
      "loss": 0.1882,
      "step": 4137
    },
    {
      "epoch": 4.73,
      "grad_norm": 1.5860782434536027,
      "learning_rate": 4.607092321510209e-07,
      "loss": 0.1704,
      "step": 4138
    },
    {
      "epoch": 4.73,
      "grad_norm": 1.8525503708654103,
      "learning_rate": 4.5683946948820945e-07,
      "loss": 0.1981,
      "step": 4139
    },
    {
      "epoch": 4.73,
      "grad_norm": 1.6959568117830472,
      "learning_rate": 4.5298590293161744e-07,
      "loss": 0.1732,
      "step": 4140
    },
    {
      "epoch": 4.73,
      "grad_norm": 1.597151037913245,
      "learning_rate": 4.491485345938418e-07,
      "loss": 0.1883,
      "step": 4141
    },
    {
      "epoch": 4.73,
      "grad_norm": 1.6930937598789055,
      "learning_rate": 4.4532736657859286e-07,
      "loss": 0.1688,
      "step": 4142
    },
    {
      "epoch": 4.73,
      "grad_norm": 1.9266892101790405,
      "learning_rate": 4.4152240098071505e-07,
      "loss": 0.1806,
      "step": 4143
    },
    {
      "epoch": 4.74,
      "grad_norm": 1.7038022344124253,
      "learning_rate": 4.3773363988615645e-07,
      "loss": 0.1759,
      "step": 4144
    },
    {
      "epoch": 4.74,
      "grad_norm": 1.6259725811157641,
      "learning_rate": 4.339610853719822e-07,
      "loss": 0.1675,
      "step": 4145
    },
    {
      "epoch": 4.74,
      "grad_norm": 1.836977492649986,
      "learning_rate": 4.302047395063813e-07,
      "loss": 0.1871,
      "step": 4146
    },
    {
      "epoch": 4.74,
      "grad_norm": 1.6996564294381378,
      "learning_rate": 4.264646043486531e-07,
      "loss": 0.2053,
      "step": 4147
    },
    {
      "epoch": 4.74,
      "grad_norm": 1.7537053419204507,
      "learning_rate": 4.2274068194920415e-07,
      "loss": 0.1666,
      "step": 4148
    },
    {
      "epoch": 4.74,
      "grad_norm": 1.6812166189672537,
      "learning_rate": 4.1903297434956134e-07,
      "loss": 0.1783,
      "step": 4149
    },
    {
      "epoch": 4.74,
      "grad_norm": 1.8363604793554071,
      "learning_rate": 4.1534148358235214e-07,
      "loss": 0.1994,
      "step": 4150
    },
    {
      "epoch": 4.74,
      "grad_norm": 1.8749175596503191,
      "learning_rate": 4.1166621167132766e-07,
      "loss": 0.1774,
      "step": 4151
    },
    {
      "epoch": 4.75,
      "grad_norm": 1.5497108796875068,
      "learning_rate": 4.0800716063133625e-07,
      "loss": 0.1829,
      "step": 4152
    },
    {
      "epoch": 4.75,
      "grad_norm": 1.72865179499611,
      "learning_rate": 4.043643324683366e-07,
      "loss": 0.17,
      "step": 4153
    },
    {
      "epoch": 4.75,
      "grad_norm": 1.748872575404866,
      "learning_rate": 4.0073772917939453e-07,
      "loss": 0.1786,
      "step": 4154
    },
    {
      "epoch": 4.75,
      "grad_norm": 1.7609792260675674,
      "learning_rate": 3.971273527526831e-07,
      "loss": 0.1963,
      "step": 4155
    },
    {
      "epoch": 4.75,
      "grad_norm": 1.8637703749165802,
      "learning_rate": 3.935332051674756e-07,
      "loss": 0.1966,
      "step": 4156
    },
    {
      "epoch": 4.75,
      "grad_norm": 1.8620192842930694,
      "learning_rate": 3.8995528839415597e-07,
      "loss": 0.2057,
      "step": 4157
    },
    {
      "epoch": 4.75,
      "grad_norm": 1.8040131256340257,
      "learning_rate": 3.8639360439419183e-07,
      "loss": 0.1959,
      "step": 4158
    },
    {
      "epoch": 4.75,
      "grad_norm": 1.6949874429397778,
      "learning_rate": 3.828481551201746e-07,
      "loss": 0.1765,
      "step": 4159
    },
    {
      "epoch": 4.75,
      "grad_norm": 1.842288333611181,
      "learning_rate": 3.793189425157795e-07,
      "loss": 0.1929,
      "step": 4160
    },
    {
      "epoch": 4.76,
      "grad_norm": 1.8787318937650501,
      "learning_rate": 3.7580596851578887e-07,
      "loss": 0.1875,
      "step": 4161
    },
    {
      "epoch": 4.76,
      "grad_norm": 1.7192865279134641,
      "learning_rate": 3.723092350460788e-07,
      "loss": 0.1602,
      "step": 4162
    },
    {
      "epoch": 4.76,
      "grad_norm": 1.8052033589383758,
      "learning_rate": 3.6882874402362597e-07,
      "loss": 0.1889,
      "step": 4163
    },
    {
      "epoch": 4.76,
      "grad_norm": 1.7684353640416697,
      "learning_rate": 3.6536449735649403e-07,
      "loss": 0.1699,
      "step": 4164
    },
    {
      "epoch": 4.76,
      "grad_norm": 1.574808848343125,
      "learning_rate": 3.6191649694385045e-07,
      "loss": 0.1413,
      "step": 4165
    },
    {
      "epoch": 4.76,
      "grad_norm": 1.6858538460008905,
      "learning_rate": 3.584847446759565e-07,
      "loss": 0.1682,
      "step": 4166
    },
    {
      "epoch": 4.76,
      "grad_norm": 1.6867743648492322,
      "learning_rate": 3.550692424341573e-07,
      "loss": 0.1851,
      "step": 4167
    },
    {
      "epoch": 4.76,
      "grad_norm": 1.6289675797351881,
      "learning_rate": 3.51669992090895e-07,
      "loss": 0.1652,
      "step": 4168
    },
    {
      "epoch": 4.76,
      "grad_norm": 1.8282026721118854,
      "learning_rate": 3.482869955096957e-07,
      "loss": 0.1939,
      "step": 4169
    },
    {
      "epoch": 4.77,
      "grad_norm": 1.6727091825325344,
      "learning_rate": 3.4492025454519574e-07,
      "loss": 0.1798,
      "step": 4170
    },
    {
      "epoch": 4.77,
      "grad_norm": 1.740688385694393,
      "learning_rate": 3.415697710430921e-07,
      "loss": 0.181,
      "step": 4171
    },
    {
      "epoch": 4.77,
      "grad_norm": 1.651684736580399,
      "learning_rate": 3.3823554684018207e-07,
      "loss": 0.1722,
      "step": 4172
    },
    {
      "epoch": 4.77,
      "grad_norm": 1.6688248910723953,
      "learning_rate": 3.34917583764357e-07,
      "loss": 0.1694,
      "step": 4173
    },
    {
      "epoch": 4.77,
      "grad_norm": 1.3622123475018657,
      "learning_rate": 3.316158836345784e-07,
      "loss": 0.1342,
      "step": 4174
    },
    {
      "epoch": 4.77,
      "grad_norm": 1.8062408187874204,
      "learning_rate": 3.283304482609017e-07,
      "loss": 0.2015,
      "step": 4175
    },
    {
      "epoch": 4.77,
      "grad_norm": 1.5740008841025745,
      "learning_rate": 3.2506127944446297e-07,
      "loss": 0.1806,
      "step": 4176
    },
    {
      "epoch": 4.77,
      "grad_norm": 1.8472492025678775,
      "learning_rate": 3.218083789774784e-07,
      "loss": 0.2001,
      "step": 4177
    },
    {
      "epoch": 4.77,
      "grad_norm": 1.7136273083788351,
      "learning_rate": 3.185717486432549e-07,
      "loss": 0.2136,
      "step": 4178
    },
    {
      "epoch": 4.78,
      "grad_norm": 1.7103003405807478,
      "learning_rate": 3.1535139021616645e-07,
      "loss": 0.2179,
      "step": 4179
    },
    {
      "epoch": 4.78,
      "grad_norm": 1.633444639722959,
      "learning_rate": 3.1214730546167416e-07,
      "loss": 0.151,
      "step": 4180
    },
    {
      "epoch": 4.78,
      "grad_norm": 1.5680137332729076,
      "learning_rate": 3.089594961363196e-07,
      "loss": 0.153,
      "step": 4181
    },
    {
      "epoch": 4.78,
      "grad_norm": 1.708116251744714,
      "learning_rate": 3.057879639877148e-07,
      "loss": 0.186,
      "step": 4182
    },
    {
      "epoch": 4.78,
      "grad_norm": 2.1474770087811996,
      "learning_rate": 3.026327107545557e-07,
      "loss": 0.2155,
      "step": 4183
    },
    {
      "epoch": 4.78,
      "grad_norm": 1.6348690341712562,
      "learning_rate": 2.994937381666085e-07,
      "loss": 0.1698,
      "step": 4184
    },
    {
      "epoch": 4.78,
      "grad_norm": 1.4532915624742586,
      "learning_rate": 2.963710479447168e-07,
      "loss": 0.1872,
      "step": 4185
    },
    {
      "epoch": 4.78,
      "grad_norm": 1.9785508917209873,
      "learning_rate": 2.932646418007945e-07,
      "loss": 0.2172,
      "step": 4186
    },
    {
      "epoch": 4.79,
      "grad_norm": 1.6905594335836855,
      "learning_rate": 2.9017452143783594e-07,
      "loss": 0.1653,
      "step": 4187
    },
    {
      "epoch": 4.79,
      "grad_norm": 1.7972292132930021,
      "learning_rate": 2.871006885498995e-07,
      "loss": 0.183,
      "step": 4188
    },
    {
      "epoch": 4.79,
      "grad_norm": 1.9633189952742964,
      "learning_rate": 2.840431448221137e-07,
      "loss": 0.185,
      "step": 4189
    },
    {
      "epoch": 4.79,
      "grad_norm": 1.628299828298247,
      "learning_rate": 2.810018919306845e-07,
      "loss": 0.1524,
      "step": 4190
    },
    {
      "epoch": 4.79,
      "grad_norm": 1.6025172269427412,
      "learning_rate": 2.7797693154288485e-07,
      "loss": 0.1535,
      "step": 4191
    },
    {
      "epoch": 4.79,
      "grad_norm": 1.6867230761426106,
      "learning_rate": 2.7496826531704823e-07,
      "loss": 0.1812,
      "step": 4192
    },
    {
      "epoch": 4.79,
      "grad_norm": 1.6611355357405166,
      "learning_rate": 2.719758949025819e-07,
      "loss": 0.1779,
      "step": 4193
    },
    {
      "epoch": 4.79,
      "grad_norm": 1.7293110038673385,
      "learning_rate": 2.6899982193996366e-07,
      "loss": 0.1834,
      "step": 4194
    },
    {
      "epoch": 4.79,
      "grad_norm": 1.7259897052681612,
      "learning_rate": 2.660400480607283e-07,
      "loss": 0.1878,
      "step": 4195
    },
    {
      "epoch": 4.8,
      "grad_norm": 1.8080220872776218,
      "learning_rate": 2.6309657488747806e-07,
      "loss": 0.1784,
      "step": 4196
    },
    {
      "epoch": 4.8,
      "grad_norm": 1.613690063043456,
      "learning_rate": 2.60169404033882e-07,
      "loss": 0.1547,
      "step": 4197
    },
    {
      "epoch": 4.8,
      "grad_norm": 1.8355204884712948,
      "learning_rate": 2.572585371046632e-07,
      "loss": 0.1817,
      "step": 4198
    },
    {
      "epoch": 4.8,
      "grad_norm": 1.5551401287454572,
      "learning_rate": 2.543639756956184e-07,
      "loss": 0.1758,
      "step": 4199
    },
    {
      "epoch": 4.8,
      "grad_norm": 1.6037239729893256,
      "learning_rate": 2.514857213935984e-07,
      "loss": 0.1526,
      "step": 4200
    },
    {
      "epoch": 4.8,
      "eval_blimp_filtered_avg": 0.7082089552238806,
      "eval_blimp_filtered_std": 0.005082712760541137,
      "step": 4200
    },
    {
      "epoch": 4.8,
      "eval_blimp_supplement_avg": 0.8297413793103449,
      "eval_blimp_supplement_std": 0.016993084807057846,
      "step": 4200
    },
    {
      "epoch": 4.8,
      "eval_vqa_filtered_avg": 0.55,
      "eval_vqa_filtered_std": 0.049999999999999996,
      "step": 4200
    },
    {
      "epoch": 4.8,
      "eval_winoground_filtered_avg": 0.63,
      "eval_winoground_filtered_std": 0.04852365870939099,
      "step": 4200
    },
    {
      "epoch": 4.8,
      "grad_norm": 2.1358883298655384,
      "learning_rate": 2.4862377577651417e-07,
      "loss": 0.2368,
      "step": 4201
    },
    {
      "epoch": 4.8,
      "grad_norm": 1.5845238407957647,
      "learning_rate": 2.4577814041333746e-07,
      "loss": 0.162,
      "step": 4202
    },
    {
      "epoch": 4.8,
      "grad_norm": 1.7771704254516887,
      "learning_rate": 2.429488168641003e-07,
      "loss": 0.1725,
      "step": 4203
    },
    {
      "epoch": 4.8,
      "grad_norm": 1.8972401066823792,
      "learning_rate": 2.4013580667988864e-07,
      "loss": 0.2056,
      "step": 4204
    },
    {
      "epoch": 4.81,
      "grad_norm": 1.577962801161097,
      "learning_rate": 2.373391114028456e-07,
      "loss": 0.1564,
      "step": 4205
    },
    {
      "epoch": 4.81,
      "grad_norm": 1.7196543196710692,
      "learning_rate": 2.3455873256617134e-07,
      "loss": 0.1996,
      "step": 4206
    },
    {
      "epoch": 4.81,
      "grad_norm": 1.590065331293314,
      "learning_rate": 2.3179467169411993e-07,
      "loss": 0.162,
      "step": 4207
    },
    {
      "epoch": 4.81,
      "grad_norm": 1.5892028840225745,
      "learning_rate": 2.2904693030200262e-07,
      "loss": 0.1509,
      "step": 4208
    },
    {
      "epoch": 4.81,
      "grad_norm": 1.6510592423174393,
      "learning_rate": 2.2631550989618443e-07,
      "loss": 0.1817,
      "step": 4209
    },
    {
      "epoch": 4.81,
      "grad_norm": 1.5743015256352912,
      "learning_rate": 2.236004119740742e-07,
      "loss": 0.1528,
      "step": 4210
    },
    {
      "epoch": 4.81,
      "grad_norm": 1.4648597750908197,
      "learning_rate": 2.2090163802414132e-07,
      "loss": 0.163,
      "step": 4211
    },
    {
      "epoch": 4.81,
      "grad_norm": 1.668258084247246,
      "learning_rate": 2.1821918952590891e-07,
      "loss": 0.1661,
      "step": 4212
    },
    {
      "epoch": 4.81,
      "grad_norm": 1.6453881067718439,
      "learning_rate": 2.1555306794993734e-07,
      "loss": 0.1724,
      "step": 4213
    },
    {
      "epoch": 4.82,
      "grad_norm": 1.53727279292723,
      "learning_rate": 2.1290327475784744e-07,
      "loss": 0.1585,
      "step": 4214
    },
    {
      "epoch": 4.82,
      "grad_norm": 1.6663560172088907,
      "learning_rate": 2.1026981140230051e-07,
      "loss": 0.1642,
      "step": 4215
    },
    {
      "epoch": 4.82,
      "grad_norm": 1.7056679783664932,
      "learning_rate": 2.0765267932701503e-07,
      "loss": 0.1625,
      "step": 4216
    },
    {
      "epoch": 4.82,
      "grad_norm": 1.431755910638817,
      "learning_rate": 2.0505187996675e-07,
      "loss": 0.1408,
      "step": 4217
    },
    {
      "epoch": 4.82,
      "grad_norm": 1.3954656388787623,
      "learning_rate": 2.0246741474731158e-07,
      "loss": 0.1236,
      "step": 4218
    },
    {
      "epoch": 4.82,
      "grad_norm": 1.5783210298781891,
      "learning_rate": 1.9989928508554968e-07,
      "loss": 0.1747,
      "step": 4219
    },
    {
      "epoch": 4.82,
      "grad_norm": 1.8656983840354258,
      "learning_rate": 1.9734749238936144e-07,
      "loss": 0.1773,
      "step": 4220
    },
    {
      "epoch": 4.82,
      "grad_norm": 1.8017000916885524,
      "learning_rate": 1.9481203805768788e-07,
      "loss": 0.172,
      "step": 4221
    },
    {
      "epoch": 4.83,
      "grad_norm": 1.4816749832892309,
      "learning_rate": 1.922929234805071e-07,
      "loss": 0.1643,
      "step": 4222
    },
    {
      "epoch": 4.83,
      "grad_norm": 1.7800378536548793,
      "learning_rate": 1.8979015003885104e-07,
      "loss": 0.1843,
      "step": 4223
    },
    {
      "epoch": 4.83,
      "grad_norm": 1.8309706032576392,
      "learning_rate": 1.8730371910477883e-07,
      "loss": 0.1811,
      "step": 4224
    },
    {
      "epoch": 4.83,
      "grad_norm": 1.6047033931856074,
      "learning_rate": 1.848336320414068e-07,
      "loss": 0.1933,
      "step": 4225
    },
    {
      "epoch": 4.83,
      "grad_norm": 1.938188550973194,
      "learning_rate": 1.823798902028784e-07,
      "loss": 0.2043,
      "step": 4226
    },
    {
      "epoch": 4.83,
      "grad_norm": 1.9576427235761895,
      "learning_rate": 1.799424949343742e-07,
      "loss": 0.1847,
      "step": 4227
    },
    {
      "epoch": 4.83,
      "grad_norm": 1.7207632679309004,
      "learning_rate": 1.7752144757212875e-07,
      "loss": 0.1774,
      "step": 4228
    },
    {
      "epoch": 4.83,
      "grad_norm": 1.7822265841892735,
      "learning_rate": 1.751167494434003e-07,
      "loss": 0.2069,
      "step": 4229
    },
    {
      "epoch": 4.83,
      "grad_norm": 1.7823663433603851,
      "learning_rate": 1.7272840186648764e-07,
      "loss": 0.212,
      "step": 4230
    },
    {
      "epoch": 4.84,
      "grad_norm": 1.783959409659246,
      "learning_rate": 1.703564061507268e-07,
      "loss": 0.201,
      "step": 4231
    },
    {
      "epoch": 4.84,
      "grad_norm": 1.6513029199180789,
      "learning_rate": 1.680007635964942e-07,
      "loss": 0.1866,
      "step": 4232
    },
    {
      "epoch": 4.84,
      "grad_norm": 1.5699090799457305,
      "learning_rate": 1.656614754951935e-07,
      "loss": 0.1729,
      "step": 4233
    },
    {
      "epoch": 4.84,
      "grad_norm": 1.4686909629127158,
      "learning_rate": 1.6333854312926555e-07,
      "loss": 0.1521,
      "step": 4234
    },
    {
      "epoch": 4.84,
      "grad_norm": 1.6729147073614774,
      "learning_rate": 1.6103196777218498e-07,
      "loss": 0.1879,
      "step": 4235
    },
    {
      "epoch": 4.84,
      "grad_norm": 1.5343658686927966,
      "learning_rate": 1.587417506884603e-07,
      "loss": 0.1363,
      "step": 4236
    },
    {
      "epoch": 4.84,
      "grad_norm": 1.9303633710630477,
      "learning_rate": 1.5646789313362718e-07,
      "loss": 0.2207,
      "step": 4237
    },
    {
      "epoch": 4.84,
      "grad_norm": 1.83244643222208,
      "learning_rate": 1.5421039635426183e-07,
      "loss": 0.2155,
      "step": 4238
    },
    {
      "epoch": 4.84,
      "grad_norm": 1.633718838422072,
      "learning_rate": 1.5196926158796088e-07,
      "loss": 0.1972,
      "step": 4239
    },
    {
      "epoch": 4.85,
      "grad_norm": 1.710171842749256,
      "learning_rate": 1.4974449006335489e-07,
      "loss": 0.1369,
      "step": 4240
    },
    {
      "epoch": 4.85,
      "grad_norm": 1.5231805620972798,
      "learning_rate": 1.4753608300011156e-07,
      "loss": 0.1535,
      "step": 4241
    },
    {
      "epoch": 4.85,
      "grad_norm": 1.7142452376675024,
      "learning_rate": 1.4534404160891578e-07,
      "loss": 0.1752,
      "step": 4242
    },
    {
      "epoch": 4.85,
      "grad_norm": 1.6551233691406522,
      "learning_rate": 1.4316836709148296e-07,
      "loss": 0.1514,
      "step": 4243
    },
    {
      "epoch": 4.85,
      "grad_norm": 1.7856553879228187,
      "learning_rate": 1.4100906064056228e-07,
      "loss": 0.1672,
      "step": 4244
    },
    {
      "epoch": 4.85,
      "grad_norm": 1.6920361769211127,
      "learning_rate": 1.3886612343992355e-07,
      "loss": 0.1812,
      "step": 4245
    },
    {
      "epoch": 4.85,
      "grad_norm": 1.712124932644748,
      "learning_rate": 1.36739556664367e-07,
      "loss": 0.1633,
      "step": 4246
    },
    {
      "epoch": 4.85,
      "grad_norm": 1.6311084289878006,
      "learning_rate": 1.3462936147971006e-07,
      "loss": 0.1842,
      "step": 4247
    },
    {
      "epoch": 4.85,
      "grad_norm": 1.7242411896691712,
      "learning_rate": 1.3253553904280736e-07,
      "loss": 0.1754,
      "step": 4248
    },
    {
      "epoch": 4.86,
      "grad_norm": 1.6980640147201027,
      "learning_rate": 1.3045809050152736e-07,
      "loss": 0.1649,
      "step": 4249
    },
    {
      "epoch": 4.86,
      "grad_norm": 1.9206634232511077,
      "learning_rate": 1.283970169947657e-07,
      "loss": 0.1837,
      "step": 4250
    },
    {
      "epoch": 4.86,
      "grad_norm": 1.7136707527236534,
      "learning_rate": 1.2635231965244188e-07,
      "loss": 0.1624,
      "step": 4251
    },
    {
      "epoch": 4.86,
      "grad_norm": 1.8965984891818417,
      "learning_rate": 1.243239995954959e-07,
      "loss": 0.1953,
      "step": 4252
    },
    {
      "epoch": 4.86,
      "grad_norm": 1.9486986296513862,
      "learning_rate": 1.2231205793588828e-07,
      "loss": 0.204,
      "step": 4253
    },
    {
      "epoch": 4.86,
      "grad_norm": 1.637122100173338,
      "learning_rate": 1.2031649577660676e-07,
      "loss": 0.1619,
      "step": 4254
    },
    {
      "epoch": 4.86,
      "grad_norm": 1.780085987974145,
      "learning_rate": 1.1833731421165617e-07,
      "loss": 0.1853,
      "step": 4255
    },
    {
      "epoch": 4.86,
      "grad_norm": 1.487817373060785,
      "learning_rate": 1.1637451432605528e-07,
      "loss": 0.1687,
      "step": 4256
    },
    {
      "epoch": 4.87,
      "grad_norm": 1.9041081172011134,
      "learning_rate": 1.1442809719585001e-07,
      "loss": 0.1954,
      "step": 4257
    },
    {
      "epoch": 4.87,
      "grad_norm": 1.5197685381715085,
      "learning_rate": 1.1249806388810347e-07,
      "loss": 0.1386,
      "step": 4258
    },
    {
      "epoch": 4.87,
      "grad_norm": 1.5141299750171282,
      "learning_rate": 1.105844154608926e-07,
      "loss": 0.1333,
      "step": 4259
    },
    {
      "epoch": 4.87,
      "grad_norm": 1.9031092950821267,
      "learning_rate": 1.0868715296331822e-07,
      "loss": 0.1767,
      "step": 4260
    },
    {
      "epoch": 4.87,
      "grad_norm": 1.8526030875860238,
      "learning_rate": 1.0680627743549165e-07,
      "loss": 0.1877,
      "step": 4261
    },
    {
      "epoch": 4.87,
      "grad_norm": 1.6483027025258055,
      "learning_rate": 1.0494178990854475e-07,
      "loss": 0.1769,
      "step": 4262
    },
    {
      "epoch": 4.87,
      "grad_norm": 1.6368415158429606,
      "learning_rate": 1.0309369140462322e-07,
      "loss": 0.1559,
      "step": 4263
    },
    {
      "epoch": 4.87,
      "grad_norm": 1.5685614548117601,
      "learning_rate": 1.0126198293688993e-07,
      "loss": 0.1636,
      "step": 4264
    },
    {
      "epoch": 4.87,
      "grad_norm": 1.598043450285483,
      "learning_rate": 9.944666550952164e-08,
      "loss": 0.1624,
      "step": 4265
    },
    {
      "epoch": 4.88,
      "grad_norm": 1.8626147934520072,
      "learning_rate": 9.76477401177056e-08,
      "loss": 0.1845,
      "step": 4266
    },
    {
      "epoch": 4.88,
      "grad_norm": 1.7957371443570325,
      "learning_rate": 9.586520774765296e-08,
      "loss": 0.1993,
      "step": 4267
    },
    {
      "epoch": 4.88,
      "grad_norm": 2.032620735066742,
      "learning_rate": 9.409906937657198e-08,
      "loss": 0.1996,
      "step": 4268
    },
    {
      "epoch": 4.88,
      "grad_norm": 1.5177743874877734,
      "learning_rate": 9.234932597270151e-08,
      "loss": 0.1542,
      "step": 4269
    },
    {
      "epoch": 4.88,
      "grad_norm": 1.679535164619695,
      "learning_rate": 9.061597849527425e-08,
      "loss": 0.1837,
      "step": 4270
    },
    {
      "epoch": 4.88,
      "grad_norm": 1.6663906455272264,
      "learning_rate": 8.889902789454673e-08,
      "loss": 0.1755,
      "step": 4271
    },
    {
      "epoch": 4.88,
      "grad_norm": 1.9848231011289936,
      "learning_rate": 8.719847511178936e-08,
      "loss": 0.2162,
      "step": 4272
    },
    {
      "epoch": 4.88,
      "grad_norm": 1.9711765606841256,
      "learning_rate": 8.551432107926638e-08,
      "loss": 0.233,
      "step": 4273
    },
    {
      "epoch": 4.88,
      "grad_norm": 1.6884598241590432,
      "learning_rate": 8.384656672026925e-08,
      "loss": 0.1769,
      "step": 4274
    },
    {
      "epoch": 4.89,
      "grad_norm": 1.717778132420375,
      "learning_rate": 8.21952129490866e-08,
      "loss": 0.1581,
      "step": 4275
    },
    {
      "epoch": 4.89,
      "grad_norm": 1.9084299420936592,
      "learning_rate": 8.056026067102762e-08,
      "loss": 0.1907,
      "step": 4276
    },
    {
      "epoch": 4.89,
      "grad_norm": 1.590740926720445,
      "learning_rate": 7.894171078239532e-08,
      "loss": 0.1486,
      "step": 4277
    },
    {
      "epoch": 4.89,
      "grad_norm": 1.6963789590489036,
      "learning_rate": 7.733956417051325e-08,
      "loss": 0.1613,
      "step": 4278
    },
    {
      "epoch": 4.89,
      "grad_norm": 1.7051928326275672,
      "learning_rate": 7.575382171370881e-08,
      "loss": 0.164,
      "step": 4279
    },
    {
      "epoch": 4.89,
      "grad_norm": 1.6382569991705374,
      "learning_rate": 7.418448428131331e-08,
      "loss": 0.1954,
      "step": 4280
    },
    {
      "epoch": 4.89,
      "grad_norm": 1.4969088598589102,
      "learning_rate": 7.263155273366851e-08,
      "loss": 0.1407,
      "step": 4281
    },
    {
      "epoch": 4.89,
      "grad_norm": 1.7478999127813215,
      "learning_rate": 7.109502792211675e-08,
      "loss": 0.1667,
      "step": 4282
    },
    {
      "epoch": 4.89,
      "grad_norm": 1.640472564416056,
      "learning_rate": 6.95749106890109e-08,
      "loss": 0.1616,
      "step": 4283
    },
    {
      "epoch": 4.9,
      "grad_norm": 1.732702165514099,
      "learning_rate": 6.807120186770766e-08,
      "loss": 0.1609,
      "step": 4284
    },
    {
      "epoch": 4.9,
      "grad_norm": 1.8142569651865335,
      "learning_rate": 6.658390228256761e-08,
      "loss": 0.18,
      "step": 4285
    },
    {
      "epoch": 4.9,
      "grad_norm": 1.9492977476686126,
      "learning_rate": 6.511301274895854e-08,
      "loss": 0.2428,
      "step": 4286
    },
    {
      "epoch": 4.9,
      "grad_norm": 1.4858277359894116,
      "learning_rate": 6.365853407324207e-08,
      "loss": 0.1257,
      "step": 4287
    },
    {
      "epoch": 4.9,
      "grad_norm": 1.4774371582411832,
      "learning_rate": 6.222046705280038e-08,
      "loss": 0.1801,
      "step": 4288
    },
    {
      "epoch": 4.9,
      "grad_norm": 1.6097289211589743,
      "learning_rate": 6.07988124759995e-08,
      "loss": 0.1837,
      "step": 4289
    },
    {
      "epoch": 4.9,
      "grad_norm": 1.6933316661956161,
      "learning_rate": 5.9393571122222653e-08,
      "loss": 0.159,
      "step": 4290
    },
    {
      "epoch": 4.9,
      "grad_norm": 1.6650027592999528,
      "learning_rate": 5.800474376184362e-08,
      "loss": 0.182,
      "step": 4291
    },
    {
      "epoch": 4.91,
      "grad_norm": 1.4629347292520003,
      "learning_rate": 5.6632331156249994e-08,
      "loss": 0.1537,
      "step": 4292
    },
    {
      "epoch": 4.91,
      "grad_norm": 1.824800768409321,
      "learning_rate": 5.527633405781662e-08,
      "loss": 0.1764,
      "step": 4293
    },
    {
      "epoch": 4.91,
      "grad_norm": 1.7502214661629538,
      "learning_rate": 5.3936753209932185e-08,
      "loss": 0.1546,
      "step": 4294
    },
    {
      "epoch": 4.91,
      "grad_norm": 1.7010285026222332,
      "learning_rate": 5.261358934697591e-08,
      "loss": 0.2042,
      "step": 4295
    },
    {
      "epoch": 4.91,
      "grad_norm": 1.4033028112361023,
      "learning_rate": 5.1306843194330875e-08,
      "loss": 0.1369,
      "step": 4296
    },
    {
      "epoch": 4.91,
      "grad_norm": 1.7768006695382847,
      "learning_rate": 5.001651546838071e-08,
      "loss": 0.1991,
      "step": 4297
    },
    {
      "epoch": 4.91,
      "grad_norm": 1.6052266507751531,
      "learning_rate": 4.874260687650622e-08,
      "loss": 0.1917,
      "step": 4298
    },
    {
      "epoch": 4.91,
      "grad_norm": 1.7257989855977833,
      "learning_rate": 4.7485118117088775e-08,
      "loss": 0.1761,
      "step": 4299
    },
    {
      "epoch": 4.91,
      "grad_norm": 1.859518349733145,
      "learning_rate": 4.624404987950692e-08,
      "loss": 0.1806,
      "step": 4300
    },
    {
      "epoch": 4.91,
      "eval_blimp_filtered_avg": 0.7070149253731344,
      "eval_blimp_filtered_std": 0.0050798786244165926,
      "step": 4300
    },
    {
      "epoch": 4.91,
      "eval_blimp_supplement_avg": 0.8297413793103449,
      "eval_blimp_supplement_std": 0.016984707085279518,
      "step": 4300
    },
    {
      "epoch": 4.91,
      "eval_vqa_filtered_avg": 0.54,
      "eval_vqa_filtered_std": 0.05009082659620333,
      "step": 4300
    },
    {
      "epoch": 4.91,
      "eval_winoground_filtered_avg": 0.61,
      "eval_winoground_filtered_std": 0.04902071300001975,
      "step": 4300
    },
    {
      "epoch": 4.92,
      "grad_norm": 1.658300954145524,
      "learning_rate": 4.501940284413309e-08,
      "loss": 0.143,
      "step": 4301
    },
    {
      "epoch": 4.92,
      "grad_norm": 1.7173045978173385,
      "learning_rate": 4.381117768234688e-08,
      "loss": 0.1603,
      "step": 4302
    },
    {
      "epoch": 4.92,
      "grad_norm": 1.7746174234358307,
      "learning_rate": 4.2619375056518475e-08,
      "loss": 0.1955,
      "step": 4303
    },
    {
      "epoch": 4.92,
      "grad_norm": 1.5518296176157136,
      "learning_rate": 4.144399562001522e-08,
      "loss": 0.1515,
      "step": 4304
    },
    {
      "epoch": 4.92,
      "grad_norm": 1.689059187266479,
      "learning_rate": 4.028504001719835e-08,
      "loss": 0.1862,
      "step": 4305
    },
    {
      "epoch": 4.92,
      "grad_norm": 1.8390359229857962,
      "learning_rate": 3.914250888343629e-08,
      "loss": 0.2135,
      "step": 4306
    },
    {
      "epoch": 4.92,
      "grad_norm": 1.5640205076549674,
      "learning_rate": 3.801640284507801e-08,
      "loss": 0.1497,
      "step": 4307
    },
    {
      "epoch": 4.92,
      "grad_norm": 1.8940306908441105,
      "learning_rate": 3.6906722519479687e-08,
      "loss": 0.2127,
      "step": 4308
    },
    {
      "epoch": 4.92,
      "grad_norm": 1.5701968539750373,
      "learning_rate": 3.581346851498801e-08,
      "loss": 0.1689,
      "step": 4309
    },
    {
      "epoch": 4.93,
      "grad_norm": 1.4552145469167725,
      "learning_rate": 3.47366414309469e-08,
      "loss": 0.1309,
      "step": 4310
    },
    {
      "epoch": 4.93,
      "grad_norm": 1.8690179080875535,
      "learning_rate": 3.3676241857687474e-08,
      "loss": 0.1879,
      "step": 4311
    },
    {
      "epoch": 4.93,
      "grad_norm": 1.721622288990227,
      "learning_rate": 3.263227037654804e-08,
      "loss": 0.2226,
      "step": 4312
    },
    {
      "epoch": 4.93,
      "grad_norm": 1.6469746183661493,
      "learning_rate": 3.160472755984412e-08,
      "loss": 0.1847,
      "step": 4313
    },
    {
      "epoch": 4.93,
      "grad_norm": 1.686537243060933,
      "learning_rate": 3.059361397090177e-08,
      "loss": 0.1721,
      "step": 4314
    },
    {
      "epoch": 4.93,
      "grad_norm": 1.6196362809382412,
      "learning_rate": 2.9598930164027593e-08,
      "loss": 0.1531,
      "step": 4315
    },
    {
      "epoch": 4.93,
      "grad_norm": 1.4726182986182101,
      "learning_rate": 2.8620676684528725e-08,
      "loss": 0.1538,
      "step": 4316
    },
    {
      "epoch": 4.93,
      "grad_norm": 1.5874577631575562,
      "learning_rate": 2.765885406869617e-08,
      "loss": 0.1252,
      "step": 4317
    },
    {
      "epoch": 4.93,
      "grad_norm": 1.5118798488588863,
      "learning_rate": 2.671346284382481e-08,
      "loss": 0.169,
      "step": 4318
    },
    {
      "epoch": 4.94,
      "grad_norm": 1.5528067281400093,
      "learning_rate": 2.5784503528193394e-08,
      "loss": 0.1591,
      "step": 4319
    },
    {
      "epoch": 4.94,
      "grad_norm": 1.7138724030565835,
      "learning_rate": 2.4871976631071215e-08,
      "loss": 0.187,
      "step": 4320
    },
    {
      "epoch": 4.94,
      "grad_norm": 1.4726393594263332,
      "learning_rate": 2.3975882652728097e-08,
      "loss": 0.1651,
      "step": 4321
    },
    {
      "epoch": 4.94,
      "grad_norm": 1.815500868064678,
      "learning_rate": 2.309622208441775e-08,
      "loss": 0.1792,
      "step": 4322
    },
    {
      "epoch": 4.94,
      "grad_norm": 1.6798398437494637,
      "learning_rate": 2.2232995408384415e-08,
      "loss": 0.2009,
      "step": 4323
    },
    {
      "epoch": 4.94,
      "grad_norm": 1.6351537999009862,
      "learning_rate": 2.138620309786621e-08,
      "loss": 0.1601,
      "step": 4324
    },
    {
      "epoch": 4.94,
      "grad_norm": 1.6377985014844092,
      "learning_rate": 2.0555845617088452e-08,
      "loss": 0.1754,
      "step": 4325
    },
    {
      "epoch": 4.94,
      "grad_norm": 1.6749306816277274,
      "learning_rate": 1.974192342127035e-08,
      "loss": 0.1705,
      "step": 4326
    },
    {
      "epoch": 4.95,
      "grad_norm": 1.965087081359588,
      "learning_rate": 1.8944436956621624e-08,
      "loss": 0.2127,
      "step": 4327
    },
    {
      "epoch": 4.95,
      "grad_norm": 2.012277924723148,
      "learning_rate": 1.816338666033257e-08,
      "loss": 0.1955,
      "step": 4328
    },
    {
      "epoch": 4.95,
      "grad_norm": 1.7004230210884352,
      "learning_rate": 1.7398772960594002e-08,
      "loss": 0.1775,
      "step": 4329
    },
    {
      "epoch": 4.95,
      "grad_norm": 2.0351918433652387,
      "learning_rate": 1.665059627658061e-08,
      "loss": 0.2253,
      "step": 4330
    },
    {
      "epoch": 4.95,
      "grad_norm": 1.7377435392437706,
      "learning_rate": 1.5918857018457635e-08,
      "loss": 0.1776,
      "step": 4331
    },
    {
      "epoch": 4.95,
      "grad_norm": 1.7232733929221407,
      "learning_rate": 1.5203555587374185e-08,
      "loss": 0.1825,
      "step": 4332
    },
    {
      "epoch": 4.95,
      "grad_norm": 1.7486421699531023,
      "learning_rate": 1.450469237547658e-08,
      "loss": 0.1718,
      "step": 4333
    },
    {
      "epoch": 4.95,
      "grad_norm": 1.7780018159597923,
      "learning_rate": 1.3822267765891683e-08,
      "loss": 0.1567,
      "step": 4334
    },
    {
      "epoch": 4.95,
      "grad_norm": 1.5244488976107824,
      "learning_rate": 1.3156282132736896e-08,
      "loss": 0.1474,
      "step": 4335
    },
    {
      "epoch": 4.96,
      "grad_norm": 1.6245059403686068,
      "learning_rate": 1.2506735841120165e-08,
      "loss": 0.1546,
      "step": 4336
    },
    {
      "epoch": 4.96,
      "grad_norm": 1.619153761120578,
      "learning_rate": 1.187362924712998e-08,
      "loss": 0.1779,
      "step": 4337
    },
    {
      "epoch": 4.96,
      "grad_norm": 1.622627703571449,
      "learning_rate": 1.1256962697852036e-08,
      "loss": 0.1551,
      "step": 4338
    },
    {
      "epoch": 4.96,
      "grad_norm": 1.959220063169498,
      "learning_rate": 1.0656736531355904e-08,
      "loss": 0.1958,
      "step": 4339
    },
    {
      "epoch": 4.96,
      "grad_norm": 1.78254801046785,
      "learning_rate": 1.0072951076688374e-08,
      "loss": 0.203,
      "step": 4340
    },
    {
      "epoch": 4.96,
      "grad_norm": 1.9347809781490521,
      "learning_rate": 9.5056066539001e-09,
      "loss": 0.1951,
      "step": 4341
    },
    {
      "epoch": 4.96,
      "grad_norm": 1.6896744848016434,
      "learning_rate": 8.954703574015621e-09,
      "loss": 0.1344,
      "step": 4342
    },
    {
      "epoch": 4.96,
      "grad_norm": 1.4392602264124519,
      "learning_rate": 8.420242139050017e-09,
      "loss": 0.1768,
      "step": 4343
    },
    {
      "epoch": 4.96,
      "grad_norm": 1.6270488306600346,
      "learning_rate": 7.902222642005575e-09,
      "loss": 0.1636,
      "step": 4344
    },
    {
      "epoch": 4.97,
      "grad_norm": 1.621243535662127,
      "learning_rate": 7.400645366871794e-09,
      "loss": 0.1526,
      "step": 4345
    },
    {
      "epoch": 4.97,
      "grad_norm": 1.5087247063831792,
      "learning_rate": 6.915510588618723e-09,
      "loss": 0.1635,
      "step": 4346
    },
    {
      "epoch": 4.97,
      "grad_norm": 1.4504646534891135,
      "learning_rate": 6.446818573210278e-09,
      "loss": 0.1386,
      "step": 4347
    },
    {
      "epoch": 4.97,
      "grad_norm": 1.7960921311989198,
      "learning_rate": 5.994569577590925e-09,
      "loss": 0.197,
      "step": 4348
    },
    {
      "epoch": 4.97,
      "grad_norm": 1.6097240724744943,
      "learning_rate": 5.558763849689008e-09,
      "loss": 0.1608,
      "step": 4349
    },
    {
      "epoch": 4.97,
      "grad_norm": 1.705313305347482,
      "learning_rate": 5.139401628423413e-09,
      "loss": 0.1717,
      "step": 4350
    },
    {
      "epoch": 4.97,
      "grad_norm": 1.556569224223631,
      "learning_rate": 4.7364831436969015e-09,
      "loss": 0.1745,
      "step": 4351
    },
    {
      "epoch": 4.97,
      "grad_norm": 1.7865811454033467,
      "learning_rate": 4.350008616399448e-09,
      "loss": 0.2048,
      "step": 4352
    },
    {
      "epoch": 4.97,
      "grad_norm": 2.0277118065577304,
      "learning_rate": 3.979978258398243e-09,
      "loss": 0.2204,
      "step": 4353
    },
    {
      "epoch": 4.98,
      "grad_norm": 1.744395709111818,
      "learning_rate": 3.626392272551016e-09,
      "loss": 0.1652,
      "step": 4354
    },
    {
      "epoch": 4.98,
      "grad_norm": 1.6579037125982428,
      "learning_rate": 3.2892508527027074e-09,
      "loss": 0.1672,
      "step": 4355
    },
    {
      "epoch": 4.98,
      "grad_norm": 1.5905522672766175,
      "learning_rate": 2.968554183682137e-09,
      "loss": 0.1506,
      "step": 4356
    },
    {
      "epoch": 4.98,
      "grad_norm": 1.721908858711472,
      "learning_rate": 2.6643024412986716e-09,
      "loss": 0.1554,
      "step": 4357
    },
    {
      "epoch": 4.98,
      "grad_norm": 1.4701881485853636,
      "learning_rate": 2.3764957923455565e-09,
      "loss": 0.1575,
      "step": 4358
    },
    {
      "epoch": 4.98,
      "grad_norm": 1.737261813456956,
      "learning_rate": 2.105134394609909e-09,
      "loss": 0.1929,
      "step": 4359
    },
    {
      "epoch": 4.98,
      "grad_norm": 1.6651755970251632,
      "learning_rate": 1.850218396852732e-09,
      "loss": 0.1946,
      "step": 4360
    },
    {
      "epoch": 4.98,
      "grad_norm": 1.7250888605835115,
      "learning_rate": 1.6117479388255697e-09,
      "loss": 0.1945,
      "step": 4361
    },
    {
      "epoch": 4.99,
      "grad_norm": 1.7823916196259166,
      "learning_rate": 1.389723151263844e-09,
      "loss": 0.1877,
      "step": 4362
    },
    {
      "epoch": 4.99,
      "grad_norm": 1.8966838383185736,
      "learning_rate": 1.1841441558801957e-09,
      "loss": 0.1906,
      "step": 4363
    },
    {
      "epoch": 4.99,
      "grad_norm": 1.9067323039178679,
      "learning_rate": 9.95011065381135e-10,
      "loss": 0.2256,
      "step": 4364
    },
    {
      "epoch": 4.99,
      "grad_norm": 1.6136779813654527,
      "learning_rate": 8.223239834537211e-10,
      "loss": 0.1683,
      "step": 4365
    },
    {
      "epoch": 4.99,
      "grad_norm": 1.7806553244340622,
      "learning_rate": 6.660830047655609e-10,
      "loss": 0.1823,
      "step": 4366
    },
    {
      "epoch": 4.99,
      "grad_norm": 1.5877497834882028,
      "learning_rate": 5.262882149748017e-10,
      "loss": 0.1427,
      "step": 4367
    },
    {
      "epoch": 4.99,
      "grad_norm": 1.6942752680954523,
      "learning_rate": 4.0293969071347747e-10,
      "loss": 0.1679,
      "step": 4368
    },
    {
      "epoch": 4.99,
      "grad_norm": 1.361962979335187,
      "learning_rate": 2.9603749960749326e-10,
      "loss": 0.1261,
      "step": 4369
    },
    {
      "epoch": 4.99,
      "grad_norm": 1.6634328200625963,
      "learning_rate": 2.0558170025997137e-10,
      "loss": 0.1607,
      "step": 4370
    },
    {
      "epoch": 5.0,
      "grad_norm": 1.6290968191860016,
      "learning_rate": 1.3157234226457427e-10,
      "loss": 0.188,
      "step": 4371
    },
    {
      "epoch": 5.0,
      "grad_norm": 1.5307856067439978,
      "learning_rate": 7.400946619218196e-11,
      "loss": 0.1334,
      "step": 4372
    },
    {
      "epoch": 5.0,
      "grad_norm": 1.6033702135788142,
      "learning_rate": 3.289310359755326e-11,
      "loss": 0.1819,
      "step": 4373
    },
    {
      "epoch": 5.0,
      "grad_norm": 1.5401766254813891,
      "learning_rate": 8.223277025987131e-12,
      "loss": 0.1638,
      "step": 4374
    },
    {
      "epoch": 5.0,
      "grad_norm": 1.5276898308891989,
      "learning_rate": 0.0,
      "loss": 0.1611,
      "step": 4375
    },
    {
      "epoch": 5.0,
      "step": 4375,
      "total_flos": 75070389288960.0,
      "train_loss": 0.8224289282832827,
      "train_runtime": 40693.8773,
      "train_samples_per_second": 13.747,
      "train_steps_per_second": 0.108
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 4375,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "total_flos": 75070389288960.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}