{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 10.0,
"eval_steps": 100,
"global_step": 5260,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"grad_norm": 3.9175004959106445,
"learning_rate": 3.79746835443038e-07,
"loss": 0.1337,
"step": 1
},
{
"epoch": 0.0,
"grad_norm": 5.145163059234619,
"learning_rate": 7.59493670886076e-07,
"loss": 0.1635,
"step": 2
},
{
"epoch": 0.01,
"grad_norm": 5.651113510131836,
"learning_rate": 1.139240506329114e-06,
"loss": 0.1933,
"step": 3
},
{
"epoch": 0.01,
"grad_norm": 5.5048112869262695,
"learning_rate": 1.518987341772152e-06,
"loss": 0.1835,
"step": 4
},
{
"epoch": 0.01,
"grad_norm": 4.910916805267334,
"learning_rate": 1.8987341772151901e-06,
"loss": 0.152,
"step": 5
},
{
"epoch": 0.01,
"grad_norm": 4.256904602050781,
"learning_rate": 2.278481012658228e-06,
"loss": 0.1759,
"step": 6
},
{
"epoch": 0.01,
"grad_norm": 5.162650108337402,
"learning_rate": 2.6582278481012658e-06,
"loss": 0.1767,
"step": 7
},
{
"epoch": 0.02,
"grad_norm": 5.422178268432617,
"learning_rate": 3.037974683544304e-06,
"loss": 0.1196,
"step": 8
},
{
"epoch": 0.02,
"grad_norm": 4.288679122924805,
"learning_rate": 3.4177215189873417e-06,
"loss": 0.1604,
"step": 9
},
{
"epoch": 0.02,
"grad_norm": 5.298702716827393,
"learning_rate": 3.7974683544303802e-06,
"loss": 0.1826,
"step": 10
},
{
"epoch": 0.02,
"grad_norm": 4.901894569396973,
"learning_rate": 4.1772151898734175e-06,
"loss": 0.1333,
"step": 11
},
{
"epoch": 0.02,
"grad_norm": 5.677184581756592,
"learning_rate": 4.556962025316456e-06,
"loss": 0.1542,
"step": 12
},
{
"epoch": 0.02,
"grad_norm": 6.519991874694824,
"learning_rate": 4.936708860759494e-06,
"loss": 0.1496,
"step": 13
},
{
"epoch": 0.03,
"grad_norm": 5.497453689575195,
"learning_rate": 5.3164556962025316e-06,
"loss": 0.1504,
"step": 14
},
{
"epoch": 0.03,
"grad_norm": 6.020430088043213,
"learning_rate": 5.69620253164557e-06,
"loss": 0.231,
"step": 15
},
{
"epoch": 0.03,
"grad_norm": 3.8732798099517822,
"learning_rate": 6.075949367088608e-06,
"loss": 0.0884,
"step": 16
},
{
"epoch": 0.03,
"grad_norm": 5.556426525115967,
"learning_rate": 6.455696202531646e-06,
"loss": 0.1641,
"step": 17
},
{
"epoch": 0.03,
"grad_norm": 7.399602890014648,
"learning_rate": 6.835443037974683e-06,
"loss": 0.2092,
"step": 18
},
{
"epoch": 0.04,
"grad_norm": 7.435789108276367,
"learning_rate": 7.215189873417722e-06,
"loss": 0.1888,
"step": 19
},
{
"epoch": 0.04,
"grad_norm": 6.196465015411377,
"learning_rate": 7.5949367088607605e-06,
"loss": 0.1492,
"step": 20
},
{
"epoch": 0.04,
"grad_norm": 6.390028476715088,
"learning_rate": 7.974683544303797e-06,
"loss": 0.1574,
"step": 21
},
{
"epoch": 0.04,
"grad_norm": 7.8620076179504395,
"learning_rate": 8.354430379746835e-06,
"loss": 0.2148,
"step": 22
},
{
"epoch": 0.04,
"grad_norm": 7.628664016723633,
"learning_rate": 8.734177215189873e-06,
"loss": 0.1886,
"step": 23
},
{
"epoch": 0.05,
"grad_norm": 8.241314888000488,
"learning_rate": 9.113924050632912e-06,
"loss": 0.1481,
"step": 24
},
{
"epoch": 0.05,
"grad_norm": 9.629205703735352,
"learning_rate": 9.49367088607595e-06,
"loss": 0.2949,
"step": 25
},
{
"epoch": 0.05,
"grad_norm": 9.069519996643066,
"learning_rate": 9.873417721518988e-06,
"loss": 0.2439,
"step": 26
},
{
"epoch": 0.05,
"grad_norm": 9.380027770996094,
"learning_rate": 1.0253164556962025e-05,
"loss": 0.269,
"step": 27
},
{
"epoch": 0.05,
"grad_norm": 9.069781303405762,
"learning_rate": 1.0632911392405063e-05,
"loss": 0.2527,
"step": 28
},
{
"epoch": 0.06,
"grad_norm": 8.468887329101562,
"learning_rate": 1.1012658227848103e-05,
"loss": 0.3108,
"step": 29
},
{
"epoch": 0.06,
"grad_norm": 7.798931121826172,
"learning_rate": 1.139240506329114e-05,
"loss": 0.2905,
"step": 30
},
{
"epoch": 0.06,
"grad_norm": 9.545028686523438,
"learning_rate": 1.1772151898734176e-05,
"loss": 0.3323,
"step": 31
},
{
"epoch": 0.06,
"grad_norm": 9.972519874572754,
"learning_rate": 1.2151898734177216e-05,
"loss": 0.3986,
"step": 32
},
{
"epoch": 0.06,
"grad_norm": 9.880655288696289,
"learning_rate": 1.2531645569620253e-05,
"loss": 0.361,
"step": 33
},
{
"epoch": 0.06,
"grad_norm": 10.031686782836914,
"learning_rate": 1.2911392405063291e-05,
"loss": 0.4157,
"step": 34
},
{
"epoch": 0.07,
"grad_norm": 10.452689170837402,
"learning_rate": 1.3291139240506329e-05,
"loss": 0.4637,
"step": 35
},
{
"epoch": 0.07,
"grad_norm": 10.207100868225098,
"learning_rate": 1.3670886075949367e-05,
"loss": 0.4177,
"step": 36
},
{
"epoch": 0.07,
"grad_norm": 9.22084903717041,
"learning_rate": 1.4050632911392406e-05,
"loss": 0.4822,
"step": 37
},
{
"epoch": 0.07,
"grad_norm": 10.47097110748291,
"learning_rate": 1.4430379746835444e-05,
"loss": 0.4341,
"step": 38
},
{
"epoch": 0.07,
"grad_norm": 8.225835800170898,
"learning_rate": 1.4810126582278482e-05,
"loss": 0.393,
"step": 39
},
{
"epoch": 0.08,
"grad_norm": 9.983627319335938,
"learning_rate": 1.5189873417721521e-05,
"loss": 0.5161,
"step": 40
},
{
"epoch": 0.08,
"grad_norm": 9.455209732055664,
"learning_rate": 1.5569620253164557e-05,
"loss": 0.5134,
"step": 41
},
{
"epoch": 0.08,
"grad_norm": 8.63866138458252,
"learning_rate": 1.5949367088607595e-05,
"loss": 0.5609,
"step": 42
},
{
"epoch": 0.08,
"grad_norm": 8.737508773803711,
"learning_rate": 1.6329113924050632e-05,
"loss": 0.4615,
"step": 43
},
{
"epoch": 0.08,
"grad_norm": 8.985445022583008,
"learning_rate": 1.670886075949367e-05,
"loss": 0.5574,
"step": 44
},
{
"epoch": 0.09,
"grad_norm": 8.926046371459961,
"learning_rate": 1.708860759493671e-05,
"loss": 0.504,
"step": 45
},
{
"epoch": 0.09,
"grad_norm": 7.6511616706848145,
"learning_rate": 1.7468354430379746e-05,
"loss": 0.3861,
"step": 46
},
{
"epoch": 0.09,
"grad_norm": 8.432501792907715,
"learning_rate": 1.7848101265822783e-05,
"loss": 0.5432,
"step": 47
},
{
"epoch": 0.09,
"grad_norm": 10.396550178527832,
"learning_rate": 1.8227848101265824e-05,
"loss": 0.6601,
"step": 48
},
{
"epoch": 0.09,
"grad_norm": 8.23548412322998,
"learning_rate": 1.860759493670886e-05,
"loss": 0.4699,
"step": 49
},
{
"epoch": 0.1,
"grad_norm": 8.672127723693848,
"learning_rate": 1.89873417721519e-05,
"loss": 0.5916,
"step": 50
},
{
"epoch": 0.1,
"grad_norm": 9.505621910095215,
"learning_rate": 1.9367088607594938e-05,
"loss": 0.6802,
"step": 51
},
{
"epoch": 0.1,
"grad_norm": 8.878487586975098,
"learning_rate": 1.9746835443037975e-05,
"loss": 0.6344,
"step": 52
},
{
"epoch": 0.1,
"grad_norm": 8.002931594848633,
"learning_rate": 2.0126582278481013e-05,
"loss": 0.6238,
"step": 53
},
{
"epoch": 0.1,
"grad_norm": 8.54853343963623,
"learning_rate": 2.050632911392405e-05,
"loss": 0.76,
"step": 54
},
{
"epoch": 0.1,
"grad_norm": 7.649876594543457,
"learning_rate": 2.0886075949367092e-05,
"loss": 0.5577,
"step": 55
},
{
"epoch": 0.11,
"grad_norm": 7.788110256195068,
"learning_rate": 2.1265822784810126e-05,
"loss": 0.6395,
"step": 56
},
{
"epoch": 0.11,
"grad_norm": 8.369818687438965,
"learning_rate": 2.1645569620253164e-05,
"loss": 0.6485,
"step": 57
},
{
"epoch": 0.11,
"grad_norm": 8.506150245666504,
"learning_rate": 2.2025316455696205e-05,
"loss": 0.6646,
"step": 58
},
{
"epoch": 0.11,
"grad_norm": 7.9867424964904785,
"learning_rate": 2.240506329113924e-05,
"loss": 0.5646,
"step": 59
},
{
"epoch": 0.11,
"grad_norm": 7.857216835021973,
"learning_rate": 2.278481012658228e-05,
"loss": 0.7844,
"step": 60
},
{
"epoch": 0.12,
"grad_norm": 7.138265609741211,
"learning_rate": 2.3164556962025318e-05,
"loss": 0.5942,
"step": 61
},
{
"epoch": 0.12,
"grad_norm": 7.170198917388916,
"learning_rate": 2.3544303797468353e-05,
"loss": 0.7432,
"step": 62
},
{
"epoch": 0.12,
"grad_norm": 7.172253608703613,
"learning_rate": 2.3924050632911394e-05,
"loss": 0.647,
"step": 63
},
{
"epoch": 0.12,
"grad_norm": 7.111423015594482,
"learning_rate": 2.430379746835443e-05,
"loss": 0.716,
"step": 64
},
{
"epoch": 0.12,
"grad_norm": 7.36029577255249,
"learning_rate": 2.468354430379747e-05,
"loss": 0.5896,
"step": 65
},
{
"epoch": 0.13,
"grad_norm": 8.06987476348877,
"learning_rate": 2.5063291139240507e-05,
"loss": 0.6631,
"step": 66
},
{
"epoch": 0.13,
"grad_norm": 7.407128810882568,
"learning_rate": 2.5443037974683545e-05,
"loss": 0.6989,
"step": 67
},
{
"epoch": 0.13,
"grad_norm": 6.675311088562012,
"learning_rate": 2.5822784810126582e-05,
"loss": 0.5181,
"step": 68
},
{
"epoch": 0.13,
"grad_norm": 7.581515789031982,
"learning_rate": 2.620253164556962e-05,
"loss": 0.6752,
"step": 69
},
{
"epoch": 0.13,
"grad_norm": 6.89380407333374,
"learning_rate": 2.6582278481012658e-05,
"loss": 0.5794,
"step": 70
},
{
"epoch": 0.13,
"grad_norm": 6.9637322425842285,
"learning_rate": 2.69620253164557e-05,
"loss": 0.6325,
"step": 71
},
{
"epoch": 0.14,
"grad_norm": 7.424550533294678,
"learning_rate": 2.7341772151898733e-05,
"loss": 0.6926,
"step": 72
},
{
"epoch": 0.14,
"grad_norm": 5.9254374504089355,
"learning_rate": 2.7721518987341774e-05,
"loss": 0.5896,
"step": 73
},
{
"epoch": 0.14,
"grad_norm": 6.71732759475708,
"learning_rate": 2.8101265822784812e-05,
"loss": 0.7428,
"step": 74
},
{
"epoch": 0.14,
"grad_norm": 7.140275955200195,
"learning_rate": 2.8481012658227846e-05,
"loss": 0.7038,
"step": 75
},
{
"epoch": 0.14,
"grad_norm": 6.515181064605713,
"learning_rate": 2.8860759493670888e-05,
"loss": 0.6555,
"step": 76
},
{
"epoch": 0.15,
"grad_norm": 6.07747220993042,
"learning_rate": 2.9240506329113925e-05,
"loss": 0.6585,
"step": 77
},
{
"epoch": 0.15,
"grad_norm": 5.888715744018555,
"learning_rate": 2.9620253164556963e-05,
"loss": 0.6201,
"step": 78
},
{
"epoch": 0.15,
"grad_norm": 6.26497220993042,
"learning_rate": 3e-05,
"loss": 0.7353,
"step": 79
},
{
"epoch": 0.15,
"grad_norm": 5.932974338531494,
"learning_rate": 3.0379746835443042e-05,
"loss": 0.6054,
"step": 80
},
{
"epoch": 0.15,
"grad_norm": 7.151474475860596,
"learning_rate": 3.075949367088607e-05,
"loss": 0.7763,
"step": 81
},
{
"epoch": 0.16,
"grad_norm": 6.203202724456787,
"learning_rate": 3.1139240506329114e-05,
"loss": 0.6086,
"step": 82
},
{
"epoch": 0.16,
"grad_norm": 6.468818187713623,
"learning_rate": 3.1518987341772155e-05,
"loss": 0.7511,
"step": 83
},
{
"epoch": 0.16,
"grad_norm": 6.266534328460693,
"learning_rate": 3.189873417721519e-05,
"loss": 0.649,
"step": 84
},
{
"epoch": 0.16,
"grad_norm": 6.891850471496582,
"learning_rate": 3.227848101265823e-05,
"loss": 0.7872,
"step": 85
},
{
"epoch": 0.16,
"grad_norm": 5.686375141143799,
"learning_rate": 3.2658227848101265e-05,
"loss": 0.6628,
"step": 86
},
{
"epoch": 0.17,
"grad_norm": 7.013491153717041,
"learning_rate": 3.3037974683544306e-05,
"loss": 0.7877,
"step": 87
},
{
"epoch": 0.17,
"grad_norm": 6.100419998168945,
"learning_rate": 3.341772151898734e-05,
"loss": 0.6175,
"step": 88
},
{
"epoch": 0.17,
"grad_norm": 6.089512825012207,
"learning_rate": 3.379746835443038e-05,
"loss": 0.6103,
"step": 89
},
{
"epoch": 0.17,
"grad_norm": 6.411644458770752,
"learning_rate": 3.417721518987342e-05,
"loss": 0.7606,
"step": 90
},
{
"epoch": 0.17,
"grad_norm": 5.83514404296875,
"learning_rate": 3.455696202531646e-05,
"loss": 0.5909,
"step": 91
},
{
"epoch": 0.17,
"grad_norm": 5.9389543533325195,
"learning_rate": 3.493670886075949e-05,
"loss": 0.6151,
"step": 92
},
{
"epoch": 0.18,
"grad_norm": 5.754659175872803,
"learning_rate": 3.531645569620253e-05,
"loss": 0.6139,
"step": 93
},
{
"epoch": 0.18,
"grad_norm": 6.1849470138549805,
"learning_rate": 3.569620253164557e-05,
"loss": 0.6415,
"step": 94
},
{
"epoch": 0.18,
"grad_norm": 6.594564437866211,
"learning_rate": 3.607594936708861e-05,
"loss": 0.6701,
"step": 95
},
{
"epoch": 0.18,
"grad_norm": 5.387152671813965,
"learning_rate": 3.645569620253165e-05,
"loss": 0.6131,
"step": 96
},
{
"epoch": 0.18,
"grad_norm": 5.76530647277832,
"learning_rate": 3.683544303797469e-05,
"loss": 0.7383,
"step": 97
},
{
"epoch": 0.19,
"grad_norm": 6.206262588500977,
"learning_rate": 3.721518987341772e-05,
"loss": 0.6689,
"step": 98
},
{
"epoch": 0.19,
"grad_norm": 5.932553291320801,
"learning_rate": 3.759493670886076e-05,
"loss": 0.6902,
"step": 99
},
{
"epoch": 0.19,
"grad_norm": 5.317256927490234,
"learning_rate": 3.79746835443038e-05,
"loss": 0.5967,
"step": 100
},
{
"epoch": 0.19,
"eval_blimp_filtered_avg": 0.735820895522388,
"eval_blimp_filtered_std": 0.004840559180894505,
"step": 100
},
{
"epoch": 0.19,
"eval_blimp_supplement_avg": 0.8017241379310345,
"eval_blimp_supplement_std": 0.01733198480824205,
"step": 100
},
{
"epoch": 0.19,
"eval_vqa_filtered_avg": 0.32,
"eval_vqa_filtered_std": 0.046882617226215034,
"step": 100
},
{
"epoch": 0.19,
"eval_winoground_filtered_avg": 0.49,
"eval_winoground_filtered_std": 0.05024183937956912,
"step": 100
},
{
"epoch": 0.19,
"grad_norm": 4.965583801269531,
"learning_rate": 3.8354430379746834e-05,
"loss": 0.5232,
"step": 101
},
{
"epoch": 0.19,
"grad_norm": 6.1376752853393555,
"learning_rate": 3.8734177215189875e-05,
"loss": 0.6635,
"step": 102
},
{
"epoch": 0.2,
"grad_norm": 6.0280327796936035,
"learning_rate": 3.9113924050632916e-05,
"loss": 0.6847,
"step": 103
},
{
"epoch": 0.2,
"grad_norm": 5.339182376861572,
"learning_rate": 3.949367088607595e-05,
"loss": 0.6508,
"step": 104
},
{
"epoch": 0.2,
"grad_norm": 5.721656799316406,
"learning_rate": 3.9873417721518985e-05,
"loss": 0.6717,
"step": 105
},
{
"epoch": 0.2,
"grad_norm": 5.537835597991943,
"learning_rate": 4.0253164556962026e-05,
"loss": 0.6903,
"step": 106
},
{
"epoch": 0.2,
"grad_norm": 4.880278587341309,
"learning_rate": 4.063291139240506e-05,
"loss": 0.5948,
"step": 107
},
{
"epoch": 0.21,
"grad_norm": 5.917506217956543,
"learning_rate": 4.10126582278481e-05,
"loss": 0.7949,
"step": 108
},
{
"epoch": 0.21,
"grad_norm": 5.594561576843262,
"learning_rate": 4.139240506329114e-05,
"loss": 0.6717,
"step": 109
},
{
"epoch": 0.21,
"grad_norm": 5.249439239501953,
"learning_rate": 4.1772151898734184e-05,
"loss": 0.6142,
"step": 110
},
{
"epoch": 0.21,
"grad_norm": 5.3499226570129395,
"learning_rate": 4.215189873417721e-05,
"loss": 0.6437,
"step": 111
},
{
"epoch": 0.21,
"grad_norm": 5.097590446472168,
"learning_rate": 4.253164556962025e-05,
"loss": 0.5324,
"step": 112
},
{
"epoch": 0.21,
"grad_norm": 5.144162178039551,
"learning_rate": 4.2911392405063294e-05,
"loss": 0.5807,
"step": 113
},
{
"epoch": 0.22,
"grad_norm": 5.480279445648193,
"learning_rate": 4.329113924050633e-05,
"loss": 0.5897,
"step": 114
},
{
"epoch": 0.22,
"grad_norm": 5.903713226318359,
"learning_rate": 4.367088607594937e-05,
"loss": 0.5655,
"step": 115
},
{
"epoch": 0.22,
"grad_norm": 6.119063377380371,
"learning_rate": 4.405063291139241e-05,
"loss": 0.7654,
"step": 116
},
{
"epoch": 0.22,
"grad_norm": 4.897018909454346,
"learning_rate": 4.443037974683544e-05,
"loss": 0.6129,
"step": 117
},
{
"epoch": 0.22,
"grad_norm": 5.184539794921875,
"learning_rate": 4.481012658227848e-05,
"loss": 0.5702,
"step": 118
},
{
"epoch": 0.23,
"grad_norm": 4.406834602355957,
"learning_rate": 4.518987341772152e-05,
"loss": 0.5389,
"step": 119
},
{
"epoch": 0.23,
"grad_norm": 4.665469646453857,
"learning_rate": 4.556962025316456e-05,
"loss": 0.5516,
"step": 120
},
{
"epoch": 0.23,
"grad_norm": 5.4455766677856445,
"learning_rate": 4.5949367088607595e-05,
"loss": 0.7366,
"step": 121
},
{
"epoch": 0.23,
"grad_norm": 5.428802013397217,
"learning_rate": 4.6329113924050637e-05,
"loss": 0.5887,
"step": 122
},
{
"epoch": 0.23,
"grad_norm": 5.139505863189697,
"learning_rate": 4.670886075949368e-05,
"loss": 0.6599,
"step": 123
},
{
"epoch": 0.24,
"grad_norm": 4.818654537200928,
"learning_rate": 4.7088607594936705e-05,
"loss": 0.5744,
"step": 124
},
{
"epoch": 0.24,
"grad_norm": 4.986923694610596,
"learning_rate": 4.7468354430379746e-05,
"loss": 0.6954,
"step": 125
},
{
"epoch": 0.24,
"grad_norm": 5.777828216552734,
"learning_rate": 4.784810126582279e-05,
"loss": 0.7288,
"step": 126
},
{
"epoch": 0.24,
"grad_norm": 5.308794021606445,
"learning_rate": 4.822784810126582e-05,
"loss": 0.6869,
"step": 127
},
{
"epoch": 0.24,
"grad_norm": 5.15963888168335,
"learning_rate": 4.860759493670886e-05,
"loss": 0.6327,
"step": 128
},
{
"epoch": 0.25,
"grad_norm": 5.143080234527588,
"learning_rate": 4.8987341772151904e-05,
"loss": 0.6528,
"step": 129
},
{
"epoch": 0.25,
"grad_norm": 5.111945152282715,
"learning_rate": 4.936708860759494e-05,
"loss": 0.7721,
"step": 130
},
{
"epoch": 0.25,
"grad_norm": 4.487850666046143,
"learning_rate": 4.974683544303797e-05,
"loss": 0.6298,
"step": 131
},
{
"epoch": 0.25,
"grad_norm": 4.99354887008667,
"learning_rate": 5.0126582278481014e-05,
"loss": 0.6803,
"step": 132
},
{
"epoch": 0.25,
"grad_norm": 4.28816556930542,
"learning_rate": 5.0506329113924055e-05,
"loss": 0.4907,
"step": 133
},
{
"epoch": 0.25,
"grad_norm": 4.963744640350342,
"learning_rate": 5.088607594936709e-05,
"loss": 0.5726,
"step": 134
},
{
"epoch": 0.26,
"grad_norm": 4.753849506378174,
"learning_rate": 5.126582278481013e-05,
"loss": 0.6057,
"step": 135
},
{
"epoch": 0.26,
"grad_norm": 5.178636074066162,
"learning_rate": 5.1645569620253165e-05,
"loss": 0.4703,
"step": 136
},
{
"epoch": 0.26,
"grad_norm": 5.6553263664245605,
"learning_rate": 5.20253164556962e-05,
"loss": 0.6415,
"step": 137
},
{
"epoch": 0.26,
"grad_norm": 4.706320285797119,
"learning_rate": 5.240506329113924e-05,
"loss": 0.5295,
"step": 138
},
{
"epoch": 0.26,
"grad_norm": 4.684479713439941,
"learning_rate": 5.278481012658228e-05,
"loss": 0.5974,
"step": 139
},
{
"epoch": 0.27,
"grad_norm": 4.653369426727295,
"learning_rate": 5.3164556962025316e-05,
"loss": 0.543,
"step": 140
},
{
"epoch": 0.27,
"grad_norm": 5.25187349319458,
"learning_rate": 5.354430379746836e-05,
"loss": 0.6223,
"step": 141
},
{
"epoch": 0.27,
"grad_norm": 3.9385440349578857,
"learning_rate": 5.39240506329114e-05,
"loss": 0.4875,
"step": 142
},
{
"epoch": 0.27,
"grad_norm": 4.36478853225708,
"learning_rate": 5.430379746835443e-05,
"loss": 0.6691,
"step": 143
},
{
"epoch": 0.27,
"grad_norm": 4.22128963470459,
"learning_rate": 5.4683544303797467e-05,
"loss": 0.553,
"step": 144
},
{
"epoch": 0.28,
"grad_norm": 4.6783905029296875,
"learning_rate": 5.506329113924051e-05,
"loss": 0.5891,
"step": 145
},
{
"epoch": 0.28,
"grad_norm": 4.668182373046875,
"learning_rate": 5.544303797468355e-05,
"loss": 0.5683,
"step": 146
},
{
"epoch": 0.28,
"grad_norm": 4.611398696899414,
"learning_rate": 5.582278481012658e-05,
"loss": 0.4798,
"step": 147
},
{
"epoch": 0.28,
"grad_norm": 4.3566460609436035,
"learning_rate": 5.6202531645569624e-05,
"loss": 0.477,
"step": 148
},
{
"epoch": 0.28,
"grad_norm": 4.709928035736084,
"learning_rate": 5.658227848101266e-05,
"loss": 0.5788,
"step": 149
},
{
"epoch": 0.29,
"grad_norm": 6.524693965911865,
"learning_rate": 5.696202531645569e-05,
"loss": 0.6396,
"step": 150
},
{
"epoch": 0.29,
"grad_norm": 5.2940144538879395,
"learning_rate": 5.7341772151898734e-05,
"loss": 0.626,
"step": 151
},
{
"epoch": 0.29,
"grad_norm": 4.811652660369873,
"learning_rate": 5.7721518987341775e-05,
"loss": 0.5884,
"step": 152
},
{
"epoch": 0.29,
"grad_norm": 5.316815376281738,
"learning_rate": 5.8101265822784816e-05,
"loss": 0.5479,
"step": 153
},
{
"epoch": 0.29,
"grad_norm": 4.054238319396973,
"learning_rate": 5.848101265822785e-05,
"loss": 0.463,
"step": 154
},
{
"epoch": 0.29,
"grad_norm": 4.240988254547119,
"learning_rate": 5.8860759493670885e-05,
"loss": 0.5653,
"step": 155
},
{
"epoch": 0.3,
"grad_norm": 5.164887428283691,
"learning_rate": 5.9240506329113926e-05,
"loss": 0.5604,
"step": 156
},
{
"epoch": 0.3,
"grad_norm": 4.015778064727783,
"learning_rate": 5.962025316455696e-05,
"loss": 0.4601,
"step": 157
},
{
"epoch": 0.3,
"grad_norm": 5.371922016143799,
"learning_rate": 6e-05,
"loss": 0.6121,
"step": 158
},
{
"epoch": 0.3,
"grad_norm": 5.462612628936768,
"learning_rate": 5.999999431264834e-05,
"loss": 0.7026,
"step": 159
},
{
"epoch": 0.3,
"grad_norm": 6.297367095947266,
"learning_rate": 5.999997725059553e-05,
"loss": 0.5357,
"step": 160
},
{
"epoch": 0.31,
"grad_norm": 3.897402286529541,
"learning_rate": 5.999994881384802e-05,
"loss": 0.503,
"step": 161
},
{
"epoch": 0.31,
"grad_norm": 4.3078508377075195,
"learning_rate": 5.9999909002416605e-05,
"loss": 0.5621,
"step": 162
},
{
"epoch": 0.31,
"grad_norm": 3.8923463821411133,
"learning_rate": 5.999985781631638e-05,
"loss": 0.4929,
"step": 163
},
{
"epoch": 0.31,
"grad_norm": 3.8754007816314697,
"learning_rate": 5.999979525556675e-05,
"loss": 0.5162,
"step": 164
},
{
"epoch": 0.31,
"grad_norm": 4.165163516998291,
"learning_rate": 5.999972132019143e-05,
"loss": 0.6166,
"step": 165
},
{
"epoch": 0.32,
"grad_norm": 5.141818046569824,
"learning_rate": 5.999963601021846e-05,
"loss": 0.6601,
"step": 166
},
{
"epoch": 0.32,
"grad_norm": 4.144015789031982,
"learning_rate": 5.999953932568019e-05,
"loss": 0.5514,
"step": 167
},
{
"epoch": 0.32,
"grad_norm": 4.668071746826172,
"learning_rate": 5.999943126661326e-05,
"loss": 0.5702,
"step": 168
},
{
"epoch": 0.32,
"grad_norm": 3.8531084060668945,
"learning_rate": 5.999931183305866e-05,
"loss": 0.4969,
"step": 169
},
{
"epoch": 0.32,
"grad_norm": 4.355401992797852,
"learning_rate": 5.9999181025061666e-05,
"loss": 0.5173,
"step": 170
},
{
"epoch": 0.33,
"grad_norm": 4.722582817077637,
"learning_rate": 5.999903884267189e-05,
"loss": 0.5285,
"step": 171
},
{
"epoch": 0.33,
"grad_norm": 4.063629627227783,
"learning_rate": 5.999888528594321e-05,
"loss": 0.4357,
"step": 172
},
{
"epoch": 0.33,
"grad_norm": 3.639040470123291,
"learning_rate": 5.9998720354933876e-05,
"loss": 0.4925,
"step": 173
},
{
"epoch": 0.33,
"grad_norm": 4.768002033233643,
"learning_rate": 5.9998544049706406e-05,
"loss": 0.6511,
"step": 174
},
{
"epoch": 0.33,
"grad_norm": 4.156386852264404,
"learning_rate": 5.999835637032766e-05,
"loss": 0.5163,
"step": 175
},
{
"epoch": 0.33,
"grad_norm": 4.105283737182617,
"learning_rate": 5.999815731686879e-05,
"loss": 0.6785,
"step": 176
},
{
"epoch": 0.34,
"grad_norm": 3.8351871967315674,
"learning_rate": 5.999794688940527e-05,
"loss": 0.4706,
"step": 177
},
{
"epoch": 0.34,
"grad_norm": 3.964378595352173,
"learning_rate": 5.999772508801689e-05,
"loss": 0.5669,
"step": 178
},
{
"epoch": 0.34,
"grad_norm": 4.290960311889648,
"learning_rate": 5.999749191278774e-05,
"loss": 0.6151,
"step": 179
},
{
"epoch": 0.34,
"grad_norm": 3.683363437652588,
"learning_rate": 5.999724736380623e-05,
"loss": 0.5293,
"step": 180
},
{
"epoch": 0.34,
"grad_norm": 4.145986080169678,
"learning_rate": 5.9996991441165087e-05,
"loss": 0.7017,
"step": 181
},
{
"epoch": 0.35,
"grad_norm": 3.628763198852539,
"learning_rate": 5.999672414496134e-05,
"loss": 0.622,
"step": 182
},
{
"epoch": 0.35,
"grad_norm": 3.490804672241211,
"learning_rate": 5.9996445475296353e-05,
"loss": 0.5202,
"step": 183
},
{
"epoch": 0.35,
"grad_norm": 3.346330404281616,
"learning_rate": 5.999615543227577e-05,
"loss": 0.4609,
"step": 184
},
{
"epoch": 0.35,
"grad_norm": 3.364408016204834,
"learning_rate": 5.999585401600956e-05,
"loss": 0.427,
"step": 185
},
{
"epoch": 0.35,
"grad_norm": 4.008015155792236,
"learning_rate": 5.999554122661201e-05,
"loss": 0.4696,
"step": 186
},
{
"epoch": 0.36,
"grad_norm": 3.96929669380188,
"learning_rate": 5.999521706420172e-05,
"loss": 0.5513,
"step": 187
},
{
"epoch": 0.36,
"grad_norm": 4.280902862548828,
"learning_rate": 5.9994881528901594e-05,
"loss": 0.4387,
"step": 188
},
{
"epoch": 0.36,
"grad_norm": 4.011152744293213,
"learning_rate": 5.9994534620838866e-05,
"loss": 0.5087,
"step": 189
},
{
"epoch": 0.36,
"grad_norm": 4.213669776916504,
"learning_rate": 5.999417634014506e-05,
"loss": 0.5272,
"step": 190
},
{
"epoch": 0.36,
"grad_norm": 4.03300142288208,
"learning_rate": 5.9993806686956003e-05,
"loss": 0.4772,
"step": 191
},
{
"epoch": 0.37,
"grad_norm": 3.8939173221588135,
"learning_rate": 5.999342566141188e-05,
"loss": 0.4971,
"step": 192
},
{
"epoch": 0.37,
"grad_norm": 3.5354299545288086,
"learning_rate": 5.999303326365714e-05,
"loss": 0.5406,
"step": 193
},
{
"epoch": 0.37,
"grad_norm": 3.1534574031829834,
"learning_rate": 5.9992629493840576e-05,
"loss": 0.3613,
"step": 194
},
{
"epoch": 0.37,
"grad_norm": 3.8146002292633057,
"learning_rate": 5.999221435211527e-05,
"loss": 0.379,
"step": 195
},
{
"epoch": 0.37,
"grad_norm": 3.565364122390747,
"learning_rate": 5.999178783863863e-05,
"loss": 0.4033,
"step": 196
},
{
"epoch": 0.37,
"grad_norm": 4.264044761657715,
"learning_rate": 5.9991349953572375e-05,
"loss": 0.4949,
"step": 197
},
{
"epoch": 0.38,
"grad_norm": 3.5022003650665283,
"learning_rate": 5.9990900697082524e-05,
"loss": 0.5189,
"step": 198
},
{
"epoch": 0.38,
"grad_norm": 3.3596351146698,
"learning_rate": 5.999044006933942e-05,
"loss": 0.4359,
"step": 199
},
{
"epoch": 0.38,
"grad_norm": 3.9610064029693604,
"learning_rate": 5.9989968070517715e-05,
"loss": 0.4154,
"step": 200
},
{
"epoch": 0.38,
"eval_blimp_filtered_avg": 0.7302985074626865,
"eval_blimp_filtered_std": 0.0048809494702359655,
"step": 200
},
{
"epoch": 0.38,
"eval_blimp_supplement_avg": 0.7844827586206896,
"eval_blimp_supplement_std": 0.017560328710745126,
"step": 200
},
{
"epoch": 0.38,
"eval_vqa_filtered_avg": 0.4,
"eval_vqa_filtered_std": 0.049236596391733084,
"step": 200
},
{
"epoch": 0.38,
"eval_winoground_filtered_avg": 0.51,
"eval_winoground_filtered_std": 0.05024183937956912,
"step": 200
},
{
"epoch": 0.38,
"grad_norm": 3.9232940673828125,
"learning_rate": 5.998948470079638e-05,
"loss": 0.4778,
"step": 201
},
{
"epoch": 0.38,
"grad_norm": 2.9619343280792236,
"learning_rate": 5.998898996035866e-05,
"loss": 0.3576,
"step": 202
},
{
"epoch": 0.39,
"grad_norm": 3.633815288543701,
"learning_rate": 5.998848384939216e-05,
"loss": 0.4828,
"step": 203
},
{
"epoch": 0.39,
"grad_norm": 3.7032175064086914,
"learning_rate": 5.9987966368088775e-05,
"loss": 0.5225,
"step": 204
},
{
"epoch": 0.39,
"grad_norm": 4.0977702140808105,
"learning_rate": 5.9987437516644706e-05,
"loss": 0.5723,
"step": 205
},
{
"epoch": 0.39,
"grad_norm": 2.977778434753418,
"learning_rate": 5.998689729526047e-05,
"loss": 0.4184,
"step": 206
},
{
"epoch": 0.39,
"grad_norm": 3.0592050552368164,
"learning_rate": 5.99863457041409e-05,
"loss": 0.3867,
"step": 207
},
{
"epoch": 0.4,
"grad_norm": 3.603689432144165,
"learning_rate": 5.998578274349514e-05,
"loss": 0.407,
"step": 208
},
{
"epoch": 0.4,
"grad_norm": 3.3270907402038574,
"learning_rate": 5.998520841353662e-05,
"loss": 0.4736,
"step": 209
},
{
"epoch": 0.4,
"grad_norm": 3.208359718322754,
"learning_rate": 5.998462271448312e-05,
"loss": 0.4541,
"step": 210
},
{
"epoch": 0.4,
"grad_norm": 4.0348968505859375,
"learning_rate": 5.998402564655671e-05,
"loss": 0.4859,
"step": 211
},
{
"epoch": 0.4,
"grad_norm": 3.537198305130005,
"learning_rate": 5.998341720998377e-05,
"loss": 0.4231,
"step": 212
},
{
"epoch": 0.4,
"grad_norm": 3.096646785736084,
"learning_rate": 5.998279740499498e-05,
"loss": 0.3793,
"step": 213
},
{
"epoch": 0.41,
"grad_norm": 2.963459014892578,
"learning_rate": 5.998216623182537e-05,
"loss": 0.4014,
"step": 214
},
{
"epoch": 0.41,
"grad_norm": 3.7130398750305176,
"learning_rate": 5.9981523690714234e-05,
"loss": 0.4625,
"step": 215
},
{
"epoch": 0.41,
"grad_norm": 3.2156856060028076,
"learning_rate": 5.99808697819052e-05,
"loss": 0.3755,
"step": 216
},
{
"epoch": 0.41,
"grad_norm": 3.309335231781006,
"learning_rate": 5.998020450564621e-05,
"loss": 0.3959,
"step": 217
},
{
"epoch": 0.41,
"grad_norm": 3.4760210514068604,
"learning_rate": 5.997952786218949e-05,
"loss": 0.3846,
"step": 218
},
{
"epoch": 0.42,
"grad_norm": 3.2478973865509033,
"learning_rate": 5.997883985179161e-05,
"loss": 0.3814,
"step": 219
},
{
"epoch": 0.42,
"grad_norm": 3.0604562759399414,
"learning_rate": 5.9978140474713426e-05,
"loss": 0.3534,
"step": 220
},
{
"epoch": 0.42,
"grad_norm": 3.1047887802124023,
"learning_rate": 5.997742973122011e-05,
"loss": 0.3316,
"step": 221
},
{
"epoch": 0.42,
"grad_norm": 3.838047981262207,
"learning_rate": 5.997670762158116e-05,
"loss": 0.4588,
"step": 222
},
{
"epoch": 0.42,
"grad_norm": 3.3866701126098633,
"learning_rate": 5.997597414607035e-05,
"loss": 0.4649,
"step": 223
},
{
"epoch": 0.43,
"grad_norm": 3.1786577701568604,
"learning_rate": 5.9975229304965784e-05,
"loss": 0.427,
"step": 224
},
{
"epoch": 0.43,
"grad_norm": 2.7126779556274414,
"learning_rate": 5.9974473098549884e-05,
"loss": 0.3141,
"step": 225
},
{
"epoch": 0.43,
"grad_norm": 3.3140058517456055,
"learning_rate": 5.9973705527109374e-05,
"loss": 0.4621,
"step": 226
},
{
"epoch": 0.43,
"grad_norm": 3.076749324798584,
"learning_rate": 5.997292659093528e-05,
"loss": 0.3511,
"step": 227
},
{
"epoch": 0.43,
"grad_norm": 3.834329128265381,
"learning_rate": 5.997213629032293e-05,
"loss": 0.4479,
"step": 228
},
{
"epoch": 0.44,
"grad_norm": 2.914842367172241,
"learning_rate": 5.997133462557197e-05,
"loss": 0.3534,
"step": 229
},
{
"epoch": 0.44,
"grad_norm": 3.1591084003448486,
"learning_rate": 5.997052159698638e-05,
"loss": 0.4687,
"step": 230
},
{
"epoch": 0.44,
"grad_norm": 2.6038105487823486,
"learning_rate": 5.99696972048744e-05,
"loss": 0.3447,
"step": 231
},
{
"epoch": 0.44,
"grad_norm": 3.7994561195373535,
"learning_rate": 5.996886144954862e-05,
"loss": 0.4482,
"step": 232
},
{
"epoch": 0.44,
"grad_norm": 3.483107328414917,
"learning_rate": 5.996801433132592e-05,
"loss": 0.3826,
"step": 233
},
{
"epoch": 0.44,
"grad_norm": 2.49519419670105,
"learning_rate": 5.996715585052748e-05,
"loss": 0.311,
"step": 234
},
{
"epoch": 0.45,
"grad_norm": 3.222121238708496,
"learning_rate": 5.996628600747881e-05,
"loss": 0.3879,
"step": 235
},
{
"epoch": 0.45,
"grad_norm": 2.7663979530334473,
"learning_rate": 5.996540480250971e-05,
"loss": 0.3721,
"step": 236
},
{
"epoch": 0.45,
"grad_norm": 3.325986862182617,
"learning_rate": 5.996451223595431e-05,
"loss": 0.4078,
"step": 237
},
{
"epoch": 0.45,
"grad_norm": 2.6873891353607178,
"learning_rate": 5.996360830815101e-05,
"loss": 0.3523,
"step": 238
},
{
"epoch": 0.45,
"grad_norm": 3.3118350505828857,
"learning_rate": 5.996269301944255e-05,
"loss": 0.3625,
"step": 239
},
{
"epoch": 0.46,
"grad_norm": 3.242655038833618,
"learning_rate": 5.996176637017597e-05,
"loss": 0.3411,
"step": 240
},
{
"epoch": 0.46,
"grad_norm": 3.603785276412964,
"learning_rate": 5.9960828360702625e-05,
"loss": 0.4841,
"step": 241
},
{
"epoch": 0.46,
"grad_norm": 2.7061522006988525,
"learning_rate": 5.995987899137815e-05,
"loss": 0.3832,
"step": 242
},
{
"epoch": 0.46,
"grad_norm": 3.7037487030029297,
"learning_rate": 5.9958918262562516e-05,
"loss": 0.5065,
"step": 243
},
{
"epoch": 0.46,
"grad_norm": 3.384439706802368,
"learning_rate": 5.995794617461998e-05,
"loss": 0.3535,
"step": 244
},
{
"epoch": 0.47,
"grad_norm": 2.943293333053589,
"learning_rate": 5.995696272791913e-05,
"loss": 0.318,
"step": 245
},
{
"epoch": 0.47,
"grad_norm": 3.12994122505188,
"learning_rate": 5.995596792283283e-05,
"loss": 0.3944,
"step": 246
},
{
"epoch": 0.47,
"grad_norm": 3.0718719959259033,
"learning_rate": 5.995496175973827e-05,
"loss": 0.3687,
"step": 247
},
{
"epoch": 0.47,
"grad_norm": 3.240448236465454,
"learning_rate": 5.9953944239016966e-05,
"loss": 0.4989,
"step": 248
},
{
"epoch": 0.47,
"grad_norm": 3.2008259296417236,
"learning_rate": 5.995291536105468e-05,
"loss": 0.4174,
"step": 249
},
{
"epoch": 0.48,
"grad_norm": 3.0582709312438965,
"learning_rate": 5.9951875126241556e-05,
"loss": 0.3905,
"step": 250
},
{
"epoch": 0.48,
"grad_norm": 3.0488216876983643,
"learning_rate": 5.995082353497198e-05,
"loss": 0.3459,
"step": 251
},
{
"epoch": 0.48,
"grad_norm": 3.3751935958862305,
"learning_rate": 5.994976058764468e-05,
"loss": 0.375,
"step": 252
},
{
"epoch": 0.48,
"grad_norm": 2.9127864837646484,
"learning_rate": 5.994868628466268e-05,
"loss": 0.3926,
"step": 253
},
{
"epoch": 0.48,
"grad_norm": 3.7527294158935547,
"learning_rate": 5.994760062643331e-05,
"loss": 0.4839,
"step": 254
},
{
"epoch": 0.48,
"grad_norm": 2.8750929832458496,
"learning_rate": 5.9946503613368196e-05,
"loss": 0.3853,
"step": 255
},
{
"epoch": 0.49,
"grad_norm": 3.118471622467041,
"learning_rate": 5.994539524588329e-05,
"loss": 0.3904,
"step": 256
},
{
"epoch": 0.49,
"grad_norm": 3.176503896713257,
"learning_rate": 5.994427552439882e-05,
"loss": 0.433,
"step": 257
},
{
"epoch": 0.49,
"grad_norm": 3.249119997024536,
"learning_rate": 5.9943144449339356e-05,
"loss": 0.3782,
"step": 258
},
{
"epoch": 0.49,
"grad_norm": 3.520930051803589,
"learning_rate": 5.9942002021133745e-05,
"loss": 0.3889,
"step": 259
},
{
"epoch": 0.49,
"grad_norm": 2.975569248199463,
"learning_rate": 5.994084824021515e-05,
"loss": 0.3669,
"step": 260
},
{
"epoch": 0.5,
"grad_norm": 2.6710500717163086,
"learning_rate": 5.9939683107021025e-05,
"loss": 0.3284,
"step": 261
},
{
"epoch": 0.5,
"grad_norm": 2.95792555809021,
"learning_rate": 5.993850662199314e-05,
"loss": 0.3677,
"step": 262
},
{
"epoch": 0.5,
"grad_norm": 3.169322967529297,
"learning_rate": 5.993731878557758e-05,
"loss": 0.3687,
"step": 263
},
{
"epoch": 0.5,
"grad_norm": 2.588524341583252,
"learning_rate": 5.9936119598224706e-05,
"loss": 0.3302,
"step": 264
},
{
"epoch": 0.5,
"grad_norm": 3.0526788234710693,
"learning_rate": 5.99349090603892e-05,
"loss": 0.3534,
"step": 265
},
{
"epoch": 0.51,
"grad_norm": 2.759631633758545,
"learning_rate": 5.993368717253006e-05,
"loss": 0.3192,
"step": 266
},
{
"epoch": 0.51,
"grad_norm": 2.9315197467803955,
"learning_rate": 5.993245393511055e-05,
"loss": 0.3962,
"step": 267
},
{
"epoch": 0.51,
"grad_norm": 3.271660327911377,
"learning_rate": 5.993120934859828e-05,
"loss": 0.4351,
"step": 268
},
{
"epoch": 0.51,
"grad_norm": 2.787395715713501,
"learning_rate": 5.9929953413465135e-05,
"loss": 0.2393,
"step": 269
},
{
"epoch": 0.51,
"grad_norm": 2.905775547027588,
"learning_rate": 5.992868613018731e-05,
"loss": 0.3505,
"step": 270
},
{
"epoch": 0.52,
"grad_norm": 2.675422430038452,
"learning_rate": 5.992740749924531e-05,
"loss": 0.3108,
"step": 271
},
{
"epoch": 0.52,
"grad_norm": 3.0469627380371094,
"learning_rate": 5.992611752112394e-05,
"loss": 0.3476,
"step": 272
},
{
"epoch": 0.52,
"grad_norm": 2.6333208084106445,
"learning_rate": 5.992481619631229e-05,
"loss": 0.3207,
"step": 273
},
{
"epoch": 0.52,
"grad_norm": 2.866960048675537,
"learning_rate": 5.992350352530377e-05,
"loss": 0.3437,
"step": 274
},
{
"epoch": 0.52,
"grad_norm": 2.8716375827789307,
"learning_rate": 5.992217950859609e-05,
"loss": 0.311,
"step": 275
},
{
"epoch": 0.52,
"grad_norm": 3.6173741817474365,
"learning_rate": 5.992084414669127e-05,
"loss": 0.3991,
"step": 276
},
{
"epoch": 0.53,
"grad_norm": 3.2167680263519287,
"learning_rate": 5.991949744009561e-05,
"loss": 0.3438,
"step": 277
},
{
"epoch": 0.53,
"grad_norm": 2.6045103073120117,
"learning_rate": 5.9918139389319726e-05,
"loss": 0.2851,
"step": 278
},
{
"epoch": 0.53,
"grad_norm": 2.835850954055786,
"learning_rate": 5.9916769994878534e-05,
"loss": 0.2958,
"step": 279
},
{
"epoch": 0.53,
"grad_norm": 2.8578131198883057,
"learning_rate": 5.991538925729125e-05,
"loss": 0.3422,
"step": 280
},
{
"epoch": 0.53,
"grad_norm": 3.084900140762329,
"learning_rate": 5.991399717708138e-05,
"loss": 0.3734,
"step": 281
},
{
"epoch": 0.54,
"grad_norm": 2.6663057804107666,
"learning_rate": 5.991259375477674e-05,
"loss": 0.3193,
"step": 282
},
{
"epoch": 0.54,
"grad_norm": 2.924835443496704,
"learning_rate": 5.991117899090947e-05,
"loss": 0.2643,
"step": 283
},
{
"epoch": 0.54,
"grad_norm": 3.4481570720672607,
"learning_rate": 5.990975288601597e-05,
"loss": 0.4303,
"step": 284
},
{
"epoch": 0.54,
"grad_norm": 2.655031681060791,
"learning_rate": 5.990831544063696e-05,
"loss": 0.3411,
"step": 285
},
{
"epoch": 0.54,
"grad_norm": 2.879248857498169,
"learning_rate": 5.990686665531745e-05,
"loss": 0.3146,
"step": 286
},
{
"epoch": 0.55,
"grad_norm": 2.674999237060547,
"learning_rate": 5.990540653060677e-05,
"loss": 0.2896,
"step": 287
},
{
"epoch": 0.55,
"grad_norm": 3.2149717807769775,
"learning_rate": 5.9903935067058524e-05,
"loss": 0.3174,
"step": 288
},
{
"epoch": 0.55,
"grad_norm": 2.646902561187744,
"learning_rate": 5.990245226523064e-05,
"loss": 0.3133,
"step": 289
},
{
"epoch": 0.55,
"grad_norm": 3.5633413791656494,
"learning_rate": 5.990095812568532e-05,
"loss": 0.4423,
"step": 290
},
{
"epoch": 0.55,
"grad_norm": 3.1335041522979736,
"learning_rate": 5.989945264898908e-05,
"loss": 0.2432,
"step": 291
},
{
"epoch": 0.56,
"grad_norm": 2.7509536743164062,
"learning_rate": 5.989793583571274e-05,
"loss": 0.3036,
"step": 292
},
{
"epoch": 0.56,
"grad_norm": 2.829868793487549,
"learning_rate": 5.98964076864314e-05,
"loss": 0.3358,
"step": 293
},
{
"epoch": 0.56,
"grad_norm": 3.1890668869018555,
"learning_rate": 5.9894868201724475e-05,
"loss": 0.4812,
"step": 294
},
{
"epoch": 0.56,
"grad_norm": 2.7375717163085938,
"learning_rate": 5.9893317382175666e-05,
"loss": 0.2799,
"step": 295
},
{
"epoch": 0.56,
"grad_norm": 2.607553005218506,
"learning_rate": 5.989175522837298e-05,
"loss": 0.2984,
"step": 296
},
{
"epoch": 0.56,
"grad_norm": 2.415358781814575,
"learning_rate": 5.9890181740908724e-05,
"loss": 0.3197,
"step": 297
},
{
"epoch": 0.57,
"grad_norm": 2.593463182449341,
"learning_rate": 5.9888596920379485e-05,
"loss": 0.311,
"step": 298
},
{
"epoch": 0.57,
"grad_norm": 2.4831888675689697,
"learning_rate": 5.988700076738616e-05,
"loss": 0.2185,
"step": 299
},
{
"epoch": 0.57,
"grad_norm": 2.832648277282715,
"learning_rate": 5.988539328253395e-05,
"loss": 0.3542,
"step": 300
},
{
"epoch": 0.57,
"eval_blimp_filtered_avg": 0.7332835820895522,
"eval_blimp_filtered_std": 0.004823865259635913,
"step": 300
},
{
"epoch": 0.57,
"eval_blimp_supplement_avg": 0.7974137931034483,
"eval_blimp_supplement_std": 0.0175288828427962,
"step": 300
},
{
"epoch": 0.57,
"eval_vqa_filtered_avg": 0.33,
"eval_vqa_filtered_std": 0.04725815626252604,
"step": 300
},
{
"epoch": 0.57,
"eval_winoground_filtered_avg": 0.51,
"eval_winoground_filtered_std": 0.05024183937956912,
"step": 300
},
{
"epoch": 0.57,
"grad_norm": 2.9080259799957275,
"learning_rate": 5.988377446643234e-05,
"loss": 0.3458,
"step": 301
},
{
"epoch": 0.57,
"grad_norm": 3.2709133625030518,
"learning_rate": 5.9882144319695104e-05,
"loss": 0.4492,
"step": 302
},
{
"epoch": 0.58,
"grad_norm": 2.6878743171691895,
"learning_rate": 5.9880502842940344e-05,
"loss": 0.3158,
"step": 303
},
{
"epoch": 0.58,
"grad_norm": 2.6411502361297607,
"learning_rate": 5.987885003679042e-05,
"loss": 0.285,
"step": 304
},
{
"epoch": 0.58,
"grad_norm": 3.1085729598999023,
"learning_rate": 5.987718590187201e-05,
"loss": 0.3284,
"step": 305
},
{
"epoch": 0.58,
"grad_norm": 2.469777822494507,
"learning_rate": 5.987551043881608e-05,
"loss": 0.3012,
"step": 306
},
{
"epoch": 0.58,
"grad_norm": 2.7271270751953125,
"learning_rate": 5.98738236482579e-05,
"loss": 0.3199,
"step": 307
},
{
"epoch": 0.59,
"grad_norm": 2.7930829524993896,
"learning_rate": 5.987212553083702e-05,
"loss": 0.3015,
"step": 308
},
{
"epoch": 0.59,
"grad_norm": 2.495415449142456,
"learning_rate": 5.98704160871973e-05,
"loss": 0.305,
"step": 309
},
{
"epoch": 0.59,
"grad_norm": 3.050403356552124,
"learning_rate": 5.986869531798688e-05,
"loss": 0.4023,
"step": 310
},
{
"epoch": 0.59,
"grad_norm": 2.747694730758667,
"learning_rate": 5.9866963223858204e-05,
"loss": 0.2798,
"step": 311
},
{
"epoch": 0.59,
"grad_norm": 2.3856823444366455,
"learning_rate": 5.986521980546801e-05,
"loss": 0.2851,
"step": 312
},
{
"epoch": 0.6,
"grad_norm": 2.4634649753570557,
"learning_rate": 5.9863465063477325e-05,
"loss": 0.2961,
"step": 313
},
{
"epoch": 0.6,
"grad_norm": 2.812249183654785,
"learning_rate": 5.986169899855147e-05,
"loss": 0.2782,
"step": 314
},
{
"epoch": 0.6,
"grad_norm": 2.2254583835601807,
"learning_rate": 5.985992161136007e-05,
"loss": 0.2412,
"step": 315
},
{
"epoch": 0.6,
"grad_norm": 2.301006555557251,
"learning_rate": 5.9858132902577014e-05,
"loss": 0.2335,
"step": 316
},
{
"epoch": 0.6,
"grad_norm": 2.2392797470092773,
"learning_rate": 5.9856332872880516e-05,
"loss": 0.2136,
"step": 317
},
{
"epoch": 0.6,
"grad_norm": 2.7301435470581055,
"learning_rate": 5.9854521522953065e-05,
"loss": 0.235,
"step": 318
},
{
"epoch": 0.61,
"grad_norm": 3.6057446002960205,
"learning_rate": 5.985269885348146e-05,
"loss": 0.3155,
"step": 319
},
{
"epoch": 0.61,
"grad_norm": 3.4933650493621826,
"learning_rate": 5.9850864865156764e-05,
"loss": 0.4632,
"step": 320
},
{
"epoch": 0.61,
"grad_norm": 3.3500263690948486,
"learning_rate": 5.9849019558674344e-05,
"loss": 0.3438,
"step": 321
},
{
"epoch": 0.61,
"grad_norm": 2.497302293777466,
"learning_rate": 5.984716293473387e-05,
"loss": 0.2354,
"step": 322
},
{
"epoch": 0.61,
"grad_norm": 2.8048901557922363,
"learning_rate": 5.984529499403929e-05,
"loss": 0.3728,
"step": 323
},
{
"epoch": 0.62,
"grad_norm": 2.21390700340271,
"learning_rate": 5.9843415737298844e-05,
"loss": 0.1969,
"step": 324
},
{
"epoch": 0.62,
"grad_norm": 2.8044192790985107,
"learning_rate": 5.984152516522507e-05,
"loss": 0.3633,
"step": 325
},
{
"epoch": 0.62,
"grad_norm": 2.369816780090332,
"learning_rate": 5.983962327853479e-05,
"loss": 0.2581,
"step": 326
},
{
"epoch": 0.62,
"grad_norm": 2.613353967666626,
"learning_rate": 5.983771007794911e-05,
"loss": 0.2472,
"step": 327
},
{
"epoch": 0.62,
"grad_norm": 2.20285964012146,
"learning_rate": 5.983578556419344e-05,
"loss": 0.2884,
"step": 328
},
{
"epoch": 0.63,
"grad_norm": 2.223172903060913,
"learning_rate": 5.983384973799747e-05,
"loss": 0.2574,
"step": 329
},
{
"epoch": 0.63,
"grad_norm": 2.0567522048950195,
"learning_rate": 5.983190260009519e-05,
"loss": 0.223,
"step": 330
},
{
"epoch": 0.63,
"grad_norm": 2.5754551887512207,
"learning_rate": 5.982994415122485e-05,
"loss": 0.2393,
"step": 331
},
{
"epoch": 0.63,
"grad_norm": 2.5476841926574707,
"learning_rate": 5.982797439212904e-05,
"loss": 0.3017,
"step": 332
},
{
"epoch": 0.63,
"grad_norm": 2.534669876098633,
"learning_rate": 5.982599332355458e-05,
"loss": 0.2866,
"step": 333
},
{
"epoch": 0.63,
"grad_norm": 2.7685904502868652,
"learning_rate": 5.982400094625261e-05,
"loss": 0.2611,
"step": 334
},
{
"epoch": 0.64,
"grad_norm": 2.3333921432495117,
"learning_rate": 5.9821997260978565e-05,
"loss": 0.22,
"step": 335
},
{
"epoch": 0.64,
"grad_norm": 2.71771240234375,
"learning_rate": 5.981998226849215e-05,
"loss": 0.3065,
"step": 336
},
{
"epoch": 0.64,
"grad_norm": 2.8669044971466064,
"learning_rate": 5.981795596955735e-05,
"loss": 0.3463,
"step": 337
},
{
"epoch": 0.64,
"grad_norm": 2.467029094696045,
"learning_rate": 5.981591836494248e-05,
"loss": 0.2554,
"step": 338
},
{
"epoch": 0.64,
"grad_norm": 2.549724578857422,
"learning_rate": 5.981386945542009e-05,
"loss": 0.2698,
"step": 339
},
{
"epoch": 0.65,
"grad_norm": 2.6562583446502686,
"learning_rate": 5.981180924176704e-05,
"loss": 0.3041,
"step": 340
},
{
"epoch": 0.65,
"grad_norm": 2.2014517784118652,
"learning_rate": 5.980973772476448e-05,
"loss": 0.281,
"step": 341
},
{
"epoch": 0.65,
"grad_norm": 2.36507511138916,
"learning_rate": 5.9807654905197826e-05,
"loss": 0.2162,
"step": 342
},
{
"epoch": 0.65,
"grad_norm": 2.680798292160034,
"learning_rate": 5.9805560783856805e-05,
"loss": 0.3116,
"step": 343
},
{
"epoch": 0.65,
"grad_norm": 2.1825125217437744,
"learning_rate": 5.980345536153543e-05,
"loss": 0.2448,
"step": 344
},
{
"epoch": 0.66,
"grad_norm": 2.7784745693206787,
"learning_rate": 5.9801338639031956e-05,
"loss": 0.2922,
"step": 345
},
{
"epoch": 0.66,
"grad_norm": 2.0150933265686035,
"learning_rate": 5.979921061714898e-05,
"loss": 0.1972,
"step": 346
},
{
"epoch": 0.66,
"grad_norm": 2.6845476627349854,
"learning_rate": 5.9797071296693325e-05,
"loss": 0.3012,
"step": 347
},
{
"epoch": 0.66,
"grad_norm": 2.7155652046203613,
"learning_rate": 5.979492067847616e-05,
"loss": 0.3163,
"step": 348
},
{
"epoch": 0.66,
"grad_norm": 2.534285545349121,
"learning_rate": 5.9792758763312895e-05,
"loss": 0.235,
"step": 349
},
{
"epoch": 0.67,
"grad_norm": 2.1968131065368652,
"learning_rate": 5.979058555202324e-05,
"loss": 0.2348,
"step": 350
},
{
"epoch": 0.67,
"grad_norm": 2.2630422115325928,
"learning_rate": 5.978840104543117e-05,
"loss": 0.2078,
"step": 351
},
{
"epoch": 0.67,
"grad_norm": 2.380323886871338,
"learning_rate": 5.9786205244364954e-05,
"loss": 0.2549,
"step": 352
},
{
"epoch": 0.67,
"grad_norm": 2.1898386478424072,
"learning_rate": 5.9783998149657166e-05,
"loss": 0.189,
"step": 353
},
{
"epoch": 0.67,
"grad_norm": 2.5150115489959717,
"learning_rate": 5.978177976214462e-05,
"loss": 0.2343,
"step": 354
},
{
"epoch": 0.67,
"grad_norm": 2.9260966777801514,
"learning_rate": 5.9779550082668446e-05,
"loss": 0.2712,
"step": 355
},
{
"epoch": 0.68,
"grad_norm": 2.7630796432495117,
"learning_rate": 5.977730911207404e-05,
"loss": 0.2323,
"step": 356
},
{
"epoch": 0.68,
"grad_norm": 2.4779183864593506,
"learning_rate": 5.9775056851211065e-05,
"loss": 0.2212,
"step": 357
},
{
"epoch": 0.68,
"grad_norm": 2.9302053451538086,
"learning_rate": 5.9772793300933505e-05,
"loss": 0.2594,
"step": 358
},
{
"epoch": 0.68,
"grad_norm": 2.833881139755249,
"learning_rate": 5.9770518462099586e-05,
"loss": 0.28,
"step": 359
},
{
"epoch": 0.68,
"grad_norm": 2.6423261165618896,
"learning_rate": 5.976823233557184e-05,
"loss": 0.2743,
"step": 360
},
{
"epoch": 0.69,
"grad_norm": 2.441531181335449,
"learning_rate": 5.976593492221705e-05,
"loss": 0.2194,
"step": 361
},
{
"epoch": 0.69,
"grad_norm": 2.8908536434173584,
"learning_rate": 5.9763626222906304e-05,
"loss": 0.4459,
"step": 362
},
{
"epoch": 0.69,
"grad_norm": 2.0470006465911865,
"learning_rate": 5.976130623851497e-05,
"loss": 0.1857,
"step": 363
},
{
"epoch": 0.69,
"grad_norm": 2.313760995864868,
"learning_rate": 5.975897496992267e-05,
"loss": 0.2712,
"step": 364
},
{
"epoch": 0.69,
"grad_norm": 1.9525102376937866,
"learning_rate": 5.975663241801333e-05,
"loss": 0.1876,
"step": 365
},
{
"epoch": 0.7,
"grad_norm": 2.240333080291748,
"learning_rate": 5.975427858367514e-05,
"loss": 0.3036,
"step": 366
},
{
"epoch": 0.7,
"grad_norm": 2.6497061252593994,
"learning_rate": 5.9751913467800584e-05,
"loss": 0.3427,
"step": 367
},
{
"epoch": 0.7,
"grad_norm": 2.374687910079956,
"learning_rate": 5.97495370712864e-05,
"loss": 0.26,
"step": 368
},
{
"epoch": 0.7,
"grad_norm": 2.4449923038482666,
"learning_rate": 5.974714939503362e-05,
"loss": 0.2521,
"step": 369
},
{
"epoch": 0.7,
"grad_norm": 2.520024061203003,
"learning_rate": 5.974475043994753e-05,
"loss": 0.291,
"step": 370
},
{
"epoch": 0.71,
"grad_norm": 2.644174098968506,
"learning_rate": 5.974234020693775e-05,
"loss": 0.2917,
"step": 371
},
{
"epoch": 0.71,
"grad_norm": 2.1318678855895996,
"learning_rate": 5.973991869691809e-05,
"loss": 0.1999,
"step": 372
},
{
"epoch": 0.71,
"grad_norm": 2.4748287200927734,
"learning_rate": 5.973748591080671e-05,
"loss": 0.23,
"step": 373
},
{
"epoch": 0.71,
"grad_norm": 2.7718019485473633,
"learning_rate": 5.973504184952602e-05,
"loss": 0.3271,
"step": 374
},
{
"epoch": 0.71,
"grad_norm": 2.431978464126587,
"learning_rate": 5.973258651400268e-05,
"loss": 0.2583,
"step": 375
},
{
"epoch": 0.71,
"grad_norm": 2.8330860137939453,
"learning_rate": 5.973011990516767e-05,
"loss": 0.2124,
"step": 376
},
{
"epoch": 0.72,
"grad_norm": 2.2848148345947266,
"learning_rate": 5.97276420239562e-05,
"loss": 0.2089,
"step": 377
},
{
"epoch": 0.72,
"grad_norm": 2.3330984115600586,
"learning_rate": 5.97251528713078e-05,
"loss": 0.2074,
"step": 378
},
{
"epoch": 0.72,
"grad_norm": 2.6895229816436768,
"learning_rate": 5.972265244816623e-05,
"loss": 0.2894,
"step": 379
},
{
"epoch": 0.72,
"grad_norm": 2.2100677490234375,
"learning_rate": 5.972014075547954e-05,
"loss": 0.2044,
"step": 380
},
{
"epoch": 0.72,
"grad_norm": 2.1728501319885254,
"learning_rate": 5.971761779420008e-05,
"loss": 0.2613,
"step": 381
},
{
"epoch": 0.73,
"grad_norm": 2.547057628631592,
"learning_rate": 5.9715083565284426e-05,
"loss": 0.3105,
"step": 382
},
{
"epoch": 0.73,
"grad_norm": 2.376161575317383,
"learning_rate": 5.971253806969345e-05,
"loss": 0.2437,
"step": 383
},
{
"epoch": 0.73,
"grad_norm": 2.8268380165100098,
"learning_rate": 5.97099813083923e-05,
"loss": 0.2624,
"step": 384
},
{
"epoch": 0.73,
"grad_norm": 2.4889893531799316,
"learning_rate": 5.970741328235038e-05,
"loss": 0.2324,
"step": 385
},
{
"epoch": 0.73,
"grad_norm": 1.7808310985565186,
"learning_rate": 5.970483399254139e-05,
"loss": 0.1899,
"step": 386
},
{
"epoch": 0.74,
"grad_norm": 1.9073035717010498,
"learning_rate": 5.970224343994328e-05,
"loss": 0.1857,
"step": 387
},
{
"epoch": 0.74,
"grad_norm": 2.750807762145996,
"learning_rate": 5.9699641625538276e-05,
"loss": 0.3467,
"step": 388
},
{
"epoch": 0.74,
"grad_norm": 2.2048373222351074,
"learning_rate": 5.969702855031286e-05,
"loss": 0.2478,
"step": 389
},
{
"epoch": 0.74,
"grad_norm": 3.4341559410095215,
"learning_rate": 5.9694404215257804e-05,
"loss": 0.3996,
"step": 390
},
{
"epoch": 0.74,
"grad_norm": 2.1668033599853516,
"learning_rate": 5.969176862136815e-05,
"loss": 0.2748,
"step": 391
},
{
"epoch": 0.75,
"grad_norm": 2.460563898086548,
"learning_rate": 5.96891217696432e-05,
"loss": 0.2857,
"step": 392
},
{
"epoch": 0.75,
"grad_norm": 2.2946317195892334,
"learning_rate": 5.968646366108652e-05,
"loss": 0.2249,
"step": 393
},
{
"epoch": 0.75,
"grad_norm": 2.5212395191192627,
"learning_rate": 5.968379429670596e-05,
"loss": 0.2365,
"step": 394
},
{
"epoch": 0.75,
"grad_norm": 2.290038824081421,
"learning_rate": 5.96811136775136e-05,
"loss": 0.2533,
"step": 395
},
{
"epoch": 0.75,
"grad_norm": 2.3398196697235107,
"learning_rate": 5.967842180452585e-05,
"loss": 0.2328,
"step": 396
},
{
"epoch": 0.75,
"grad_norm": 2.202530860900879,
"learning_rate": 5.9675718678763345e-05,
"loss": 0.2564,
"step": 397
},
{
"epoch": 0.76,
"grad_norm": 2.6087045669555664,
"learning_rate": 5.9673004301250975e-05,
"loss": 0.3086,
"step": 398
},
{
"epoch": 0.76,
"grad_norm": 2.5613410472869873,
"learning_rate": 5.967027867301792e-05,
"loss": 0.2826,
"step": 399
},
{
"epoch": 0.76,
"grad_norm": 2.278445243835449,
"learning_rate": 5.9667541795097633e-05,
"loss": 0.2394,
"step": 400
},
{
"epoch": 0.76,
"eval_blimp_filtered_avg": 0.7356716417910448,
"eval_blimp_filtered_std": 0.004821205670038822,
"step": 400
},
{
"epoch": 0.76,
"eval_blimp_supplement_avg": 0.7866379310344828,
"eval_blimp_supplement_std": 0.017667706349291778,
"step": 400
},
{
"epoch": 0.76,
"eval_vqa_filtered_avg": 0.32,
"eval_vqa_filtered_std": 0.046882617226215034,
"step": 400
},
{
"epoch": 0.76,
"eval_winoground_filtered_avg": 0.51,
"eval_winoground_filtered_std": 0.05024183937956912,
"step": 400
},
{
"epoch": 0.76,
"grad_norm": 2.1116080284118652,
"learning_rate": 5.966479366852781e-05,
"loss": 0.1971,
"step": 401
},
{
"epoch": 0.76,
"grad_norm": 3.073458433151245,
"learning_rate": 5.9662034294350426e-05,
"loss": 0.3245,
"step": 402
},
{
"epoch": 0.77,
"grad_norm": 2.4883532524108887,
"learning_rate": 5.965926367361171e-05,
"loss": 0.1807,
"step": 403
},
{
"epoch": 0.77,
"grad_norm": 2.113600730895996,
"learning_rate": 5.965648180736217e-05,
"loss": 0.2351,
"step": 404
},
{
"epoch": 0.77,
"grad_norm": 2.6334376335144043,
"learning_rate": 5.965368869665656e-05,
"loss": 0.2462,
"step": 405
},
{
"epoch": 0.77,
"grad_norm": 2.371997833251953,
"learning_rate": 5.965088434255392e-05,
"loss": 0.2708,
"step": 406
},
{
"epoch": 0.77,
"grad_norm": 2.3920860290527344,
"learning_rate": 5.9648068746117525e-05,
"loss": 0.278,
"step": 407
},
{
"epoch": 0.78,
"grad_norm": 2.4014406204223633,
"learning_rate": 5.964524190841494e-05,
"loss": 0.2101,
"step": 408
},
{
"epoch": 0.78,
"grad_norm": 2.157572031021118,
"learning_rate": 5.964240383051797e-05,
"loss": 0.2731,
"step": 409
},
{
"epoch": 0.78,
"grad_norm": 2.7519772052764893,
"learning_rate": 5.96395545135027e-05,
"loss": 0.2941,
"step": 410
},
{
"epoch": 0.78,
"grad_norm": 2.4970381259918213,
"learning_rate": 5.9636693958449464e-05,
"loss": 0.2925,
"step": 411
},
{
"epoch": 0.78,
"grad_norm": 2.825547933578491,
"learning_rate": 5.963382216644285e-05,
"loss": 0.2282,
"step": 412
},
{
"epoch": 0.79,
"grad_norm": 2.247845411300659,
"learning_rate": 5.963093913857174e-05,
"loss": 0.2286,
"step": 413
},
{
"epoch": 0.79,
"grad_norm": 2.2771432399749756,
"learning_rate": 5.962804487592924e-05,
"loss": 0.2231,
"step": 414
},
{
"epoch": 0.79,
"grad_norm": 1.8853540420532227,
"learning_rate": 5.962513937961273e-05,
"loss": 0.1807,
"step": 415
},
{
"epoch": 0.79,
"grad_norm": 2.190530300140381,
"learning_rate": 5.9622222650723846e-05,
"loss": 0.2108,
"step": 416
},
{
"epoch": 0.79,
"grad_norm": 2.4473540782928467,
"learning_rate": 5.961929469036849e-05,
"loss": 0.2574,
"step": 417
},
{
"epoch": 0.79,
"grad_norm": 2.049542188644409,
"learning_rate": 5.9616355499656825e-05,
"loss": 0.1757,
"step": 418
},
{
"epoch": 0.8,
"grad_norm": 2.4526305198669434,
"learning_rate": 5.961340507970325e-05,
"loss": 0.3134,
"step": 419
},
{
"epoch": 0.8,
"grad_norm": 2.1873769760131836,
"learning_rate": 5.961044343162644e-05,
"loss": 0.2292,
"step": 420
},
{
"epoch": 0.8,
"grad_norm": 1.8719329833984375,
"learning_rate": 5.960747055654934e-05,
"loss": 0.1655,
"step": 421
},
{
"epoch": 0.8,
"grad_norm": 1.7540192604064941,
"learning_rate": 5.960448645559912e-05,
"loss": 0.1479,
"step": 422
},
{
"epoch": 0.8,
"grad_norm": 2.099355697631836,
"learning_rate": 5.960149112990721e-05,
"loss": 0.2077,
"step": 423
},
{
"epoch": 0.81,
"grad_norm": 1.910293459892273,
"learning_rate": 5.959848458060934e-05,
"loss": 0.2068,
"step": 424
},
{
"epoch": 0.81,
"grad_norm": 3.0821235179901123,
"learning_rate": 5.959546680884544e-05,
"loss": 0.3455,
"step": 425
},
{
"epoch": 0.81,
"grad_norm": 2.456092357635498,
"learning_rate": 5.959243781575972e-05,
"loss": 0.2303,
"step": 426
},
{
"epoch": 0.81,
"grad_norm": 2.8642914295196533,
"learning_rate": 5.9589397602500655e-05,
"loss": 0.3097,
"step": 427
},
{
"epoch": 0.81,
"grad_norm": 2.252239227294922,
"learning_rate": 5.958634617022096e-05,
"loss": 0.2202,
"step": 428
},
{
"epoch": 0.82,
"grad_norm": 2.231159210205078,
"learning_rate": 5.95832835200776e-05,
"loss": 0.2333,
"step": 429
},
{
"epoch": 0.82,
"grad_norm": 2.3209710121154785,
"learning_rate": 5.958020965323181e-05,
"loss": 0.2539,
"step": 430
},
{
"epoch": 0.82,
"grad_norm": 2.1590359210968018,
"learning_rate": 5.9577124570849054e-05,
"loss": 0.2185,
"step": 431
},
{
"epoch": 0.82,
"grad_norm": 2.3673133850097656,
"learning_rate": 5.9574028274099054e-05,
"loss": 0.2247,
"step": 432
},
{
"epoch": 0.82,
"grad_norm": 1.9484679698944092,
"learning_rate": 5.9570920764155826e-05,
"loss": 0.2214,
"step": 433
},
{
"epoch": 0.83,
"grad_norm": 2.3724923133850098,
"learning_rate": 5.956780204219757e-05,
"loss": 0.295,
"step": 434
},
{
"epoch": 0.83,
"grad_norm": 1.9988019466400146,
"learning_rate": 5.956467210940679e-05,
"loss": 0.2048,
"step": 435
},
{
"epoch": 0.83,
"grad_norm": 2.379077911376953,
"learning_rate": 5.956153096697021e-05,
"loss": 0.239,
"step": 436
},
{
"epoch": 0.83,
"grad_norm": 2.170581579208374,
"learning_rate": 5.955837861607882e-05,
"loss": 0.1738,
"step": 437
},
{
"epoch": 0.83,
"grad_norm": 2.4073967933654785,
"learning_rate": 5.9555215057927854e-05,
"loss": 0.2705,
"step": 438
},
{
"epoch": 0.83,
"grad_norm": 2.043124198913574,
"learning_rate": 5.955204029371681e-05,
"loss": 0.1895,
"step": 439
},
{
"epoch": 0.84,
"grad_norm": 2.126648426055908,
"learning_rate": 5.9548854324649395e-05,
"loss": 0.2426,
"step": 440
},
{
"epoch": 0.84,
"grad_norm": 2.125807285308838,
"learning_rate": 5.954565715193361e-05,
"loss": 0.1627,
"step": 441
},
{
"epoch": 0.84,
"grad_norm": 2.1922452449798584,
"learning_rate": 5.954244877678168e-05,
"loss": 0.1752,
"step": 442
},
{
"epoch": 0.84,
"grad_norm": 2.022874355316162,
"learning_rate": 5.953922920041009e-05,
"loss": 0.2435,
"step": 443
},
{
"epoch": 0.84,
"grad_norm": 2.176436424255371,
"learning_rate": 5.953599842403954e-05,
"loss": 0.2141,
"step": 444
},
{
"epoch": 0.85,
"grad_norm": 1.9334514141082764,
"learning_rate": 5.953275644889503e-05,
"loss": 0.2069,
"step": 445
},
{
"epoch": 0.85,
"grad_norm": 2.2492358684539795,
"learning_rate": 5.952950327620576e-05,
"loss": 0.3085,
"step": 446
},
{
"epoch": 0.85,
"grad_norm": 2.5290560722351074,
"learning_rate": 5.9526238907205195e-05,
"loss": 0.2851,
"step": 447
},
{
"epoch": 0.85,
"grad_norm": 2.284060478210449,
"learning_rate": 5.9522963343131046e-05,
"loss": 0.1957,
"step": 448
},
{
"epoch": 0.85,
"grad_norm": 2.3056936264038086,
"learning_rate": 5.951967658522527e-05,
"loss": 0.2644,
"step": 449
},
{
"epoch": 0.86,
"grad_norm": 2.088824510574341,
"learning_rate": 5.9516378634734044e-05,
"loss": 0.2298,
"step": 450
},
{
"epoch": 0.86,
"grad_norm": 2.2547812461853027,
"learning_rate": 5.9513069492907825e-05,
"loss": 0.1833,
"step": 451
},
{
"epoch": 0.86,
"grad_norm": 2.2166264057159424,
"learning_rate": 5.950974916100128e-05,
"loss": 0.2222,
"step": 452
},
{
"epoch": 0.86,
"grad_norm": 2.2373790740966797,
"learning_rate": 5.950641764027337e-05,
"loss": 0.2257,
"step": 453
},
{
"epoch": 0.86,
"grad_norm": 2.5090081691741943,
"learning_rate": 5.9503074931987226e-05,
"loss": 0.2315,
"step": 454
},
{
"epoch": 0.87,
"grad_norm": 1.8476955890655518,
"learning_rate": 5.949972103741028e-05,
"loss": 0.1999,
"step": 455
},
{
"epoch": 0.87,
"grad_norm": 2.076594114303589,
"learning_rate": 5.949635595781418e-05,
"loss": 0.1737,
"step": 456
},
{
"epoch": 0.87,
"grad_norm": 2.1583218574523926,
"learning_rate": 5.9492979694474816e-05,
"loss": 0.238,
"step": 457
},
{
"epoch": 0.87,
"grad_norm": 1.8073211908340454,
"learning_rate": 5.948959224867232e-05,
"loss": 0.1876,
"step": 458
},
{
"epoch": 0.87,
"grad_norm": 2.6077382564544678,
"learning_rate": 5.948619362169107e-05,
"loss": 0.2272,
"step": 459
},
{
"epoch": 0.87,
"grad_norm": 2.8505430221557617,
"learning_rate": 5.948278381481967e-05,
"loss": 0.2974,
"step": 460
},
{
"epoch": 0.88,
"grad_norm": 2.5609378814697266,
"learning_rate": 5.947936282935097e-05,
"loss": 0.2639,
"step": 461
},
{
"epoch": 0.88,
"grad_norm": 2.624199867248535,
"learning_rate": 5.947593066658208e-05,
"loss": 0.2133,
"step": 462
},
{
"epoch": 0.88,
"grad_norm": 2.1878724098205566,
"learning_rate": 5.947248732781431e-05,
"loss": 0.2583,
"step": 463
},
{
"epoch": 0.88,
"grad_norm": 1.8262358903884888,
"learning_rate": 5.946903281435323e-05,
"loss": 0.1758,
"step": 464
},
{
"epoch": 0.88,
"grad_norm": 1.6880618333816528,
"learning_rate": 5.946556712750864e-05,
"loss": 0.1487,
"step": 465
},
{
"epoch": 0.89,
"grad_norm": 1.9107998609542847,
"learning_rate": 5.946209026859458e-05,
"loss": 0.2322,
"step": 466
},
{
"epoch": 0.89,
"grad_norm": 2.437047004699707,
"learning_rate": 5.945860223892932e-05,
"loss": 0.2387,
"step": 467
},
{
"epoch": 0.89,
"grad_norm": 1.9823325872421265,
"learning_rate": 5.9455103039835384e-05,
"loss": 0.2535,
"step": 468
},
{
"epoch": 0.89,
"grad_norm": 2.270927906036377,
"learning_rate": 5.94515926726395e-05,
"loss": 0.2571,
"step": 469
},
{
"epoch": 0.89,
"grad_norm": 2.687070608139038,
"learning_rate": 5.944807113867266e-05,
"loss": 0.3052,
"step": 470
},
{
"epoch": 0.9,
"grad_norm": 1.6973243951797485,
"learning_rate": 5.9444538439270074e-05,
"loss": 0.1593,
"step": 471
},
{
"epoch": 0.9,
"grad_norm": 1.4848052263259888,
"learning_rate": 5.944099457577119e-05,
"loss": 0.1262,
"step": 472
},
{
"epoch": 0.9,
"grad_norm": 1.9856061935424805,
"learning_rate": 5.943743954951968e-05,
"loss": 0.1972,
"step": 473
},
{
"epoch": 0.9,
"grad_norm": 2.098254680633545,
"learning_rate": 5.943387336186346e-05,
"loss": 0.2286,
"step": 474
},
{
"epoch": 0.9,
"grad_norm": 1.9061243534088135,
"learning_rate": 5.943029601415468e-05,
"loss": 0.1651,
"step": 475
},
{
"epoch": 0.9,
"grad_norm": 2.570235252380371,
"learning_rate": 5.9426707507749716e-05,
"loss": 0.2066,
"step": 476
},
{
"epoch": 0.91,
"grad_norm": 2.494701385498047,
"learning_rate": 5.942310784400917e-05,
"loss": 0.2365,
"step": 477
},
{
"epoch": 0.91,
"grad_norm": 2.3150126934051514,
"learning_rate": 5.941949702429787e-05,
"loss": 0.2372,
"step": 478
},
{
"epoch": 0.91,
"grad_norm": 2.0876221656799316,
"learning_rate": 5.9415875049984905e-05,
"loss": 0.1873,
"step": 479
},
{
"epoch": 0.91,
"grad_norm": 2.255262851715088,
"learning_rate": 5.9412241922443544e-05,
"loss": 0.1744,
"step": 480
},
{
"epoch": 0.91,
"grad_norm": 2.3070077896118164,
"learning_rate": 5.9408597643051344e-05,
"loss": 0.2328,
"step": 481
},
{
"epoch": 0.92,
"grad_norm": 2.3618338108062744,
"learning_rate": 5.940494221319002e-05,
"loss": 0.2648,
"step": 482
},
{
"epoch": 0.92,
"grad_norm": 2.0942540168762207,
"learning_rate": 5.940127563424558e-05,
"loss": 0.1798,
"step": 483
},
{
"epoch": 0.92,
"grad_norm": 2.025592565536499,
"learning_rate": 5.939759790760823e-05,
"loss": 0.173,
"step": 484
},
{
"epoch": 0.92,
"grad_norm": 2.4102261066436768,
"learning_rate": 5.9393909034672405e-05,
"loss": 0.1941,
"step": 485
},
{
"epoch": 0.92,
"grad_norm": 2.0048739910125732,
"learning_rate": 5.939020901683676e-05,
"loss": 0.1735,
"step": 486
},
{
"epoch": 0.93,
"grad_norm": 2.4052751064300537,
"learning_rate": 5.9386497855504174e-05,
"loss": 0.2273,
"step": 487
},
{
"epoch": 0.93,
"grad_norm": 2.6534152030944824,
"learning_rate": 5.938277555208178e-05,
"loss": 0.1906,
"step": 488
},
{
"epoch": 0.93,
"grad_norm": 2.097266912460327,
"learning_rate": 5.93790421079809e-05,
"loss": 0.2387,
"step": 489
},
{
"epoch": 0.93,
"grad_norm": 2.1218717098236084,
"learning_rate": 5.9375297524617085e-05,
"loss": 0.1801,
"step": 490
},
{
"epoch": 0.93,
"grad_norm": 2.003192901611328,
"learning_rate": 5.9371541803410145e-05,
"loss": 0.2112,
"step": 491
},
{
"epoch": 0.94,
"grad_norm": 2.209049940109253,
"learning_rate": 5.936777494578407e-05,
"loss": 0.2082,
"step": 492
},
{
"epoch": 0.94,
"grad_norm": 2.2141027450561523,
"learning_rate": 5.9363996953167096e-05,
"loss": 0.2574,
"step": 493
},
{
"epoch": 0.94,
"grad_norm": 1.7429929971694946,
"learning_rate": 5.9360207826991674e-05,
"loss": 0.1914,
"step": 494
},
{
"epoch": 0.94,
"grad_norm": 2.0210278034210205,
"learning_rate": 5.935640756869448e-05,
"loss": 0.1993,
"step": 495
},
{
"epoch": 0.94,
"grad_norm": 2.0718135833740234,
"learning_rate": 5.935259617971639e-05,
"loss": 0.1668,
"step": 496
},
{
"epoch": 0.94,
"grad_norm": 1.92594313621521,
"learning_rate": 5.934877366150253e-05,
"loss": 0.1565,
"step": 497
},
{
"epoch": 0.95,
"grad_norm": 1.8142975568771362,
"learning_rate": 5.9344940015502246e-05,
"loss": 0.1653,
"step": 498
},
{
"epoch": 0.95,
"grad_norm": 2.125386953353882,
"learning_rate": 5.934109524316907e-05,
"loss": 0.2553,
"step": 499
},
{
"epoch": 0.95,
"grad_norm": 1.7284529209136963,
"learning_rate": 5.9337239345960794e-05,
"loss": 0.1517,
"step": 500
},
{
"epoch": 0.95,
"eval_blimp_filtered_avg": 0.7295522388059702,
"eval_blimp_filtered_std": 0.004923339720694572,
"step": 500
},
{
"epoch": 0.95,
"eval_blimp_supplement_avg": 0.7866379310344828,
"eval_blimp_supplement_std": 0.0178606573271889,
"step": 500
},
{
"epoch": 0.95,
"eval_vqa_filtered_avg": 0.29,
"eval_vqa_filtered_std": 0.045604802157206845,
"step": 500
},
{
"epoch": 0.95,
"eval_winoground_filtered_avg": 0.52,
"eval_winoground_filtered_std": 0.05021167315686779,
"step": 500
},
{
"epoch": 0.95,
"grad_norm": 2.1051669120788574,
"learning_rate": 5.9333372325339385e-05,
"loss": 0.1589,
"step": 501
},
{
"epoch": 0.95,
"grad_norm": 2.319821357727051,
"learning_rate": 5.932949418277107e-05,
"loss": 0.2285,
"step": 502
},
{
"epoch": 0.96,
"grad_norm": 1.9719042778015137,
"learning_rate": 5.932560491972626e-05,
"loss": 0.1733,
"step": 503
},
{
"epoch": 0.96,
"grad_norm": 2.091729164123535,
"learning_rate": 5.93217045376796e-05,
"loss": 0.1474,
"step": 504
},
{
"epoch": 0.96,
"grad_norm": 2.2337629795074463,
"learning_rate": 5.931779303810995e-05,
"loss": 0.2485,
"step": 505
},
{
"epoch": 0.96,
"grad_norm": 3.1285109519958496,
"learning_rate": 5.9313870422500365e-05,
"loss": 0.2765,
"step": 506
},
{
"epoch": 0.96,
"grad_norm": 2.0365238189697266,
"learning_rate": 5.930993669233815e-05,
"loss": 0.1852,
"step": 507
},
{
"epoch": 0.97,
"grad_norm": 2.5228734016418457,
"learning_rate": 5.9305991849114805e-05,
"loss": 0.2085,
"step": 508
},
{
"epoch": 0.97,
"grad_norm": 1.7304902076721191,
"learning_rate": 5.9302035894326034e-05,
"loss": 0.2048,
"step": 509
},
{
"epoch": 0.97,
"grad_norm": 1.7001547813415527,
"learning_rate": 5.929806882947177e-05,
"loss": 0.1563,
"step": 510
},
{
"epoch": 0.97,
"grad_norm": 1.505239486694336,
"learning_rate": 5.9294090656056154e-05,
"loss": 0.1298,
"step": 511
},
{
"epoch": 0.97,
"grad_norm": 2.2603979110717773,
"learning_rate": 5.929010137558754e-05,
"loss": 0.1761,
"step": 512
},
{
"epoch": 0.98,
"grad_norm": 2.2538270950317383,
"learning_rate": 5.928610098957847e-05,
"loss": 0.2949,
"step": 513
},
{
"epoch": 0.98,
"grad_norm": 2.165473699569702,
"learning_rate": 5.928208949954574e-05,
"loss": 0.2627,
"step": 514
},
{
"epoch": 0.98,
"grad_norm": 1.9142481088638306,
"learning_rate": 5.9278066907010325e-05,
"loss": 0.1915,
"step": 515
},
{
"epoch": 0.98,
"grad_norm": 1.6888093948364258,
"learning_rate": 5.927403321349742e-05,
"loss": 0.1484,
"step": 516
},
{
"epoch": 0.98,
"grad_norm": 1.9431672096252441,
"learning_rate": 5.926998842053643e-05,
"loss": 0.1707,
"step": 517
},
{
"epoch": 0.98,
"grad_norm": 2.3113322257995605,
"learning_rate": 5.926593252966096e-05,
"loss": 0.2926,
"step": 518
},
{
"epoch": 0.99,
"grad_norm": 1.8364462852478027,
"learning_rate": 5.926186554240884e-05,
"loss": 0.1522,
"step": 519
},
{
"epoch": 0.99,
"grad_norm": 1.97207510471344,
"learning_rate": 5.9257787460322066e-05,
"loss": 0.1902,
"step": 520
},
{
"epoch": 0.99,
"grad_norm": 1.756150722503662,
"learning_rate": 5.9253698284946904e-05,
"loss": 0.1446,
"step": 521
},
{
"epoch": 0.99,
"grad_norm": 1.6009244918823242,
"learning_rate": 5.924959801783378e-05,
"loss": 0.174,
"step": 522
},
{
"epoch": 0.99,
"grad_norm": 2.059051036834717,
"learning_rate": 5.924548666053734e-05,
"loss": 0.1708,
"step": 523
},
{
"epoch": 1.0,
"grad_norm": 2.155153512954712,
"learning_rate": 5.9241364214616424e-05,
"loss": 0.2418,
"step": 524
},
{
"epoch": 1.0,
"grad_norm": 2.3209428787231445,
"learning_rate": 5.92372306816341e-05,
"loss": 0.2019,
"step": 525
},
{
"epoch": 1.0,
"grad_norm": 1.634609341621399,
"learning_rate": 5.923308606315761e-05,
"loss": 0.1356,
"step": 526
},
{
"epoch": 1.0,
"grad_norm": 1.6899735927581787,
"learning_rate": 5.922893036075843e-05,
"loss": 0.0975,
"step": 527
},
{
"epoch": 1.0,
"grad_norm": 1.9456154108047485,
"learning_rate": 5.92247635760122e-05,
"loss": 0.1837,
"step": 528
},
{
"epoch": 1.01,
"grad_norm": 1.842386245727539,
"learning_rate": 5.922058571049882e-05,
"loss": 0.1264,
"step": 529
},
{
"epoch": 1.01,
"grad_norm": 1.7307193279266357,
"learning_rate": 5.921639676580231e-05,
"loss": 0.1086,
"step": 530
},
{
"epoch": 1.01,
"grad_norm": 2.1826059818267822,
"learning_rate": 5.921219674351097e-05,
"loss": 0.1373,
"step": 531
},
{
"epoch": 1.01,
"grad_norm": 1.9496084451675415,
"learning_rate": 5.920798564521727e-05,
"loss": 0.0946,
"step": 532
},
{
"epoch": 1.01,
"grad_norm": 1.6906980276107788,
"learning_rate": 5.9203763472517854e-05,
"loss": 0.1309,
"step": 533
},
{
"epoch": 1.02,
"grad_norm": 1.6446951627731323,
"learning_rate": 5.919953022701361e-05,
"loss": 0.1329,
"step": 534
},
{
"epoch": 1.02,
"grad_norm": 1.4856971502304077,
"learning_rate": 5.919528591030958e-05,
"loss": 0.0974,
"step": 535
},
{
"epoch": 1.02,
"grad_norm": 1.8769277334213257,
"learning_rate": 5.919103052401503e-05,
"loss": 0.1156,
"step": 536
},
{
"epoch": 1.02,
"grad_norm": 1.52096688747406,
"learning_rate": 5.9186764069743435e-05,
"loss": 0.0762,
"step": 537
},
{
"epoch": 1.02,
"grad_norm": 2.2493762969970703,
"learning_rate": 5.918248654911243e-05,
"loss": 0.1024,
"step": 538
},
{
"epoch": 1.02,
"grad_norm": 1.845508337020874,
"learning_rate": 5.9178197963743876e-05,
"loss": 0.1104,
"step": 539
},
{
"epoch": 1.03,
"grad_norm": 1.5087875127792358,
"learning_rate": 5.917389831526383e-05,
"loss": 0.1179,
"step": 540
},
{
"epoch": 1.03,
"grad_norm": 2.18507719039917,
"learning_rate": 5.916958760530252e-05,
"loss": 0.1552,
"step": 541
},
{
"epoch": 1.03,
"grad_norm": 1.9257490634918213,
"learning_rate": 5.916526583549437e-05,
"loss": 0.1123,
"step": 542
},
{
"epoch": 1.03,
"grad_norm": 1.7880960702896118,
"learning_rate": 5.916093300747802e-05,
"loss": 0.0889,
"step": 543
},
{
"epoch": 1.03,
"grad_norm": 2.032665252685547,
"learning_rate": 5.9156589122896294e-05,
"loss": 0.1517,
"step": 544
},
{
"epoch": 1.04,
"grad_norm": 1.5970219373703003,
"learning_rate": 5.915223418339621e-05,
"loss": 0.083,
"step": 545
},
{
"epoch": 1.04,
"grad_norm": 1.803774118423462,
"learning_rate": 5.9147868190628955e-05,
"loss": 0.1027,
"step": 546
},
{
"epoch": 1.04,
"grad_norm": 2.553575277328491,
"learning_rate": 5.9143491146249945e-05,
"loss": 0.1918,
"step": 547
},
{
"epoch": 1.04,
"grad_norm": 2.2445855140686035,
"learning_rate": 5.9139103051918755e-05,
"loss": 0.0949,
"step": 548
},
{
"epoch": 1.04,
"grad_norm": 1.7714852094650269,
"learning_rate": 5.913470390929916e-05,
"loss": 0.1174,
"step": 549
},
{
"epoch": 1.05,
"grad_norm": 1.9340238571166992,
"learning_rate": 5.913029372005913e-05,
"loss": 0.1229,
"step": 550
},
{
"epoch": 1.05,
"grad_norm": 2.006537675857544,
"learning_rate": 5.912587248587081e-05,
"loss": 0.1305,
"step": 551
},
{
"epoch": 1.05,
"grad_norm": 1.6777127981185913,
"learning_rate": 5.912144020841055e-05,
"loss": 0.0942,
"step": 552
},
{
"epoch": 1.05,
"grad_norm": 1.5644162893295288,
"learning_rate": 5.9116996889358864e-05,
"loss": 0.0863,
"step": 553
},
{
"epoch": 1.05,
"grad_norm": 1.6141797304153442,
"learning_rate": 5.9112542530400493e-05,
"loss": 0.0766,
"step": 554
},
{
"epoch": 1.06,
"grad_norm": 1.5457810163497925,
"learning_rate": 5.9108077133224304e-05,
"loss": 0.0959,
"step": 555
},
{
"epoch": 1.06,
"grad_norm": 2.018963098526001,
"learning_rate": 5.910360069952341e-05,
"loss": 0.1403,
"step": 556
},
{
"epoch": 1.06,
"grad_norm": 1.6411982774734497,
"learning_rate": 5.909911323099507e-05,
"loss": 0.1073,
"step": 557
},
{
"epoch": 1.06,
"grad_norm": 1.986802339553833,
"learning_rate": 5.9094614729340735e-05,
"loss": 0.1213,
"step": 558
},
{
"epoch": 1.06,
"grad_norm": 1.7201321125030518,
"learning_rate": 5.909010519626604e-05,
"loss": 0.0996,
"step": 559
},
{
"epoch": 1.06,
"grad_norm": 1.8231854438781738,
"learning_rate": 5.908558463348082e-05,
"loss": 0.0962,
"step": 560
},
{
"epoch": 1.07,
"grad_norm": 2.038626194000244,
"learning_rate": 5.908105304269907e-05,
"loss": 0.095,
"step": 561
},
{
"epoch": 1.07,
"grad_norm": 1.5228958129882812,
"learning_rate": 5.9076510425638965e-05,
"loss": 0.0685,
"step": 562
},
{
"epoch": 1.07,
"grad_norm": 1.7702863216400146,
"learning_rate": 5.907195678402288e-05,
"loss": 0.1217,
"step": 563
},
{
"epoch": 1.07,
"grad_norm": 2.3959579467773438,
"learning_rate": 5.906739211957734e-05,
"loss": 0.1228,
"step": 564
},
{
"epoch": 1.07,
"grad_norm": 1.5083688497543335,
"learning_rate": 5.906281643403311e-05,
"loss": 0.0926,
"step": 565
},
{
"epoch": 1.08,
"grad_norm": 1.9131320714950562,
"learning_rate": 5.905822972912505e-05,
"loss": 0.1194,
"step": 566
},
{
"epoch": 1.08,
"grad_norm": 2.5337512493133545,
"learning_rate": 5.9053632006592255e-05,
"loss": 0.1071,
"step": 567
},
{
"epoch": 1.08,
"grad_norm": 1.5562999248504639,
"learning_rate": 5.9049023268177976e-05,
"loss": 0.0595,
"step": 568
},
{
"epoch": 1.08,
"grad_norm": 2.2364683151245117,
"learning_rate": 5.904440351562967e-05,
"loss": 0.1204,
"step": 569
},
{
"epoch": 1.08,
"grad_norm": 1.9726566076278687,
"learning_rate": 5.903977275069892e-05,
"loss": 0.1071,
"step": 570
},
{
"epoch": 1.09,
"grad_norm": 1.9628205299377441,
"learning_rate": 5.903513097514153e-05,
"loss": 0.1216,
"step": 571
},
{
"epoch": 1.09,
"grad_norm": 2.110915184020996,
"learning_rate": 5.9030478190717454e-05,
"loss": 0.1198,
"step": 572
},
{
"epoch": 1.09,
"grad_norm": 2.0957467555999756,
"learning_rate": 5.9025814399190824e-05,
"loss": 0.1074,
"step": 573
},
{
"epoch": 1.09,
"grad_norm": 2.3168108463287354,
"learning_rate": 5.902113960232996e-05,
"loss": 0.1268,
"step": 574
},
{
"epoch": 1.09,
"grad_norm": 2.7989158630371094,
"learning_rate": 5.901645380190732e-05,
"loss": 0.1645,
"step": 575
},
{
"epoch": 1.1,
"grad_norm": 1.3799980878829956,
"learning_rate": 5.901175699969959e-05,
"loss": 0.0726,
"step": 576
},
{
"epoch": 1.1,
"grad_norm": 1.7545613050460815,
"learning_rate": 5.900704919748756e-05,
"loss": 0.1222,
"step": 577
},
{
"epoch": 1.1,
"grad_norm": 2.20133376121521,
"learning_rate": 5.900233039705625e-05,
"loss": 0.1508,
"step": 578
},
{
"epoch": 1.1,
"grad_norm": 1.9659409523010254,
"learning_rate": 5.899760060019481e-05,
"loss": 0.1209,
"step": 579
},
{
"epoch": 1.1,
"grad_norm": 1.4326027631759644,
"learning_rate": 5.899285980869658e-05,
"loss": 0.0869,
"step": 580
},
{
"epoch": 1.1,
"grad_norm": 1.6788190603256226,
"learning_rate": 5.898810802435907e-05,
"loss": 0.0941,
"step": 581
},
{
"epoch": 1.11,
"grad_norm": 1.7446272373199463,
"learning_rate": 5.8983345248983944e-05,
"loss": 0.0967,
"step": 582
},
{
"epoch": 1.11,
"grad_norm": 1.8604663610458374,
"learning_rate": 5.897857148437704e-05,
"loss": 0.1439,
"step": 583
},
{
"epoch": 1.11,
"grad_norm": 2.0718953609466553,
"learning_rate": 5.897378673234836e-05,
"loss": 0.1429,
"step": 584
},
{
"epoch": 1.11,
"grad_norm": 1.690424919128418,
"learning_rate": 5.896899099471209e-05,
"loss": 0.1089,
"step": 585
},
{
"epoch": 1.11,
"grad_norm": 1.7930703163146973,
"learning_rate": 5.8964184273286544e-05,
"loss": 0.1238,
"step": 586
},
{
"epoch": 1.12,
"grad_norm": 2.0013372898101807,
"learning_rate": 5.895936656989424e-05,
"loss": 0.1441,
"step": 587
},
{
"epoch": 1.12,
"grad_norm": 1.5639559030532837,
"learning_rate": 5.8954537886361846e-05,
"loss": 0.1033,
"step": 588
},
{
"epoch": 1.12,
"grad_norm": 1.951863169670105,
"learning_rate": 5.894969822452018e-05,
"loss": 0.1045,
"step": 589
},
{
"epoch": 1.12,
"grad_norm": 1.7152680158615112,
"learning_rate": 5.894484758620423e-05,
"loss": 0.0906,
"step": 590
},
{
"epoch": 1.12,
"grad_norm": 1.909698486328125,
"learning_rate": 5.8939985973253156e-05,
"loss": 0.1449,
"step": 591
},
{
"epoch": 1.13,
"grad_norm": 1.6972252130508423,
"learning_rate": 5.893511338751027e-05,
"loss": 0.0841,
"step": 592
},
{
"epoch": 1.13,
"grad_norm": 2.1710314750671387,
"learning_rate": 5.893022983082304e-05,
"loss": 0.1611,
"step": 593
},
{
"epoch": 1.13,
"grad_norm": 1.998836874961853,
"learning_rate": 5.8925335305043106e-05,
"loss": 0.1484,
"step": 594
},
{
"epoch": 1.13,
"grad_norm": 1.7370854616165161,
"learning_rate": 5.892042981202626e-05,
"loss": 0.0909,
"step": 595
},
{
"epoch": 1.13,
"grad_norm": 1.610866665840149,
"learning_rate": 5.891551335363245e-05,
"loss": 0.1036,
"step": 596
},
{
"epoch": 1.13,
"grad_norm": 2.373359441757202,
"learning_rate": 5.891058593172579e-05,
"loss": 0.1464,
"step": 597
},
{
"epoch": 1.14,
"grad_norm": 2.3502795696258545,
"learning_rate": 5.890564754817454e-05,
"loss": 0.1491,
"step": 598
},
{
"epoch": 1.14,
"grad_norm": 1.5936166048049927,
"learning_rate": 5.890069820485112e-05,
"loss": 0.0969,
"step": 599
},
{
"epoch": 1.14,
"grad_norm": 1.9798052310943604,
"learning_rate": 5.8895737903632116e-05,
"loss": 0.0926,
"step": 600
},
{
"epoch": 1.14,
"eval_blimp_filtered_avg": 0.7379104477611941,
"eval_blimp_filtered_std": 0.004802523876434939,
"step": 600
},
{
"epoch": 1.14,
"eval_blimp_supplement_avg": 0.7780172413793104,
"eval_blimp_supplement_std": 0.017860575174285586,
"step": 600
},
{
"epoch": 1.14,
"eval_vqa_filtered_avg": 0.31,
"eval_vqa_filtered_std": 0.04648231987117316,
"step": 600
},
{
"epoch": 1.14,
"eval_winoground_filtered_avg": 0.49,
"eval_winoground_filtered_std": 0.05024183937956912,
"step": 600
},
{
"epoch": 1.14,
"grad_norm": 1.7146224975585938,
"learning_rate": 5.889076664639825e-05,
"loss": 0.1084,
"step": 601
},
{
"epoch": 1.14,
"grad_norm": 1.5940072536468506,
"learning_rate": 5.888578443503442e-05,
"loss": 0.1037,
"step": 602
},
{
"epoch": 1.15,
"grad_norm": 1.6230546236038208,
"learning_rate": 5.888079127142965e-05,
"loss": 0.0883,
"step": 603
},
{
"epoch": 1.15,
"grad_norm": 1.545094609260559,
"learning_rate": 5.887578715747714e-05,
"loss": 0.0897,
"step": 604
},
{
"epoch": 1.15,
"grad_norm": 2.010274648666382,
"learning_rate": 5.8870772095074226e-05,
"loss": 0.1814,
"step": 605
},
{
"epoch": 1.15,
"grad_norm": 1.8277559280395508,
"learning_rate": 5.886574608612242e-05,
"loss": 0.1031,
"step": 606
},
{
"epoch": 1.15,
"grad_norm": 1.9344485998153687,
"learning_rate": 5.886070913252736e-05,
"loss": 0.1112,
"step": 607
},
{
"epoch": 1.16,
"grad_norm": 1.6639392375946045,
"learning_rate": 5.885566123619884e-05,
"loss": 0.0951,
"step": 608
},
{
"epoch": 1.16,
"grad_norm": 2.4886300563812256,
"learning_rate": 5.885060239905079e-05,
"loss": 0.1281,
"step": 609
},
{
"epoch": 1.16,
"grad_norm": 1.768975853919983,
"learning_rate": 5.884553262300132e-05,
"loss": 0.1181,
"step": 610
},
{
"epoch": 1.16,
"grad_norm": 1.6957968473434448,
"learning_rate": 5.8840451909972665e-05,
"loss": 0.085,
"step": 611
},
{
"epoch": 1.16,
"grad_norm": 2.130704879760742,
"learning_rate": 5.8835360261891216e-05,
"loss": 0.094,
"step": 612
},
{
"epoch": 1.17,
"grad_norm": 1.0947633981704712,
"learning_rate": 5.8830257680687495e-05,
"loss": 0.0756,
"step": 613
},
{
"epoch": 1.17,
"grad_norm": 2.610474109649658,
"learning_rate": 5.8825144168296184e-05,
"loss": 0.1298,
"step": 614
},
{
"epoch": 1.17,
"grad_norm": 2.138145685195923,
"learning_rate": 5.8820019726656116e-05,
"loss": 0.0953,
"step": 615
},
{
"epoch": 1.17,
"grad_norm": 2.270446300506592,
"learning_rate": 5.881488435771025e-05,
"loss": 0.1227,
"step": 616
},
{
"epoch": 1.17,
"grad_norm": 1.5228792428970337,
"learning_rate": 5.880973806340569e-05,
"loss": 0.0742,
"step": 617
},
{
"epoch": 1.17,
"grad_norm": 1.4687288999557495,
"learning_rate": 5.88045808456937e-05,
"loss": 0.0854,
"step": 618
},
{
"epoch": 1.18,
"grad_norm": 1.6738731861114502,
"learning_rate": 5.879941270652967e-05,
"loss": 0.0984,
"step": 619
},
{
"epoch": 1.18,
"grad_norm": 1.6337116956710815,
"learning_rate": 5.879423364787313e-05,
"loss": 0.0814,
"step": 620
},
{
"epoch": 1.18,
"grad_norm": 2.3074495792388916,
"learning_rate": 5.8789043671687756e-05,
"loss": 0.1308,
"step": 621
},
{
"epoch": 1.18,
"grad_norm": 2.1286215782165527,
"learning_rate": 5.8783842779941364e-05,
"loss": 0.1377,
"step": 622
},
{
"epoch": 1.18,
"grad_norm": 2.176161050796509,
"learning_rate": 5.8778630974605916e-05,
"loss": 0.1343,
"step": 623
},
{
"epoch": 1.19,
"grad_norm": 2.5389082431793213,
"learning_rate": 5.877340825765749e-05,
"loss": 0.0981,
"step": 624
},
{
"epoch": 1.19,
"grad_norm": 1.8792098760604858,
"learning_rate": 5.876817463107633e-05,
"loss": 0.1314,
"step": 625
},
{
"epoch": 1.19,
"grad_norm": 1.7783194780349731,
"learning_rate": 5.876293009684679e-05,
"loss": 0.1039,
"step": 626
},
{
"epoch": 1.19,
"grad_norm": 1.7445591688156128,
"learning_rate": 5.875767465695737e-05,
"loss": 0.094,
"step": 627
},
{
"epoch": 1.19,
"grad_norm": 1.762431025505066,
"learning_rate": 5.87524083134007e-05,
"loss": 0.0938,
"step": 628
},
{
"epoch": 1.2,
"grad_norm": 1.9671847820281982,
"learning_rate": 5.8747131068173565e-05,
"loss": 0.0836,
"step": 629
},
{
"epoch": 1.2,
"grad_norm": 1.8194429874420166,
"learning_rate": 5.874184292327686e-05,
"loss": 0.0907,
"step": 630
},
{
"epoch": 1.2,
"grad_norm": 1.8796792030334473,
"learning_rate": 5.873654388071562e-05,
"loss": 0.1015,
"step": 631
},
{
"epoch": 1.2,
"grad_norm": 1.7125799655914307,
"learning_rate": 5.873123394249902e-05,
"loss": 0.1087,
"step": 632
},
{
"epoch": 1.2,
"grad_norm": 1.5964893102645874,
"learning_rate": 5.872591311064035e-05,
"loss": 0.0888,
"step": 633
},
{
"epoch": 1.21,
"grad_norm": 1.7498795986175537,
"learning_rate": 5.872058138715704e-05,
"loss": 0.1216,
"step": 634
},
{
"epoch": 1.21,
"grad_norm": 1.674188494682312,
"learning_rate": 5.871523877407067e-05,
"loss": 0.1077,
"step": 635
},
{
"epoch": 1.21,
"grad_norm": 1.6416884660720825,
"learning_rate": 5.870988527340689e-05,
"loss": 0.0951,
"step": 636
},
{
"epoch": 1.21,
"grad_norm": 1.90082585811615,
"learning_rate": 5.8704520887195545e-05,
"loss": 0.0936,
"step": 637
},
{
"epoch": 1.21,
"grad_norm": 2.18945050239563,
"learning_rate": 5.8699145617470564e-05,
"loss": 0.1309,
"step": 638
},
{
"epoch": 1.21,
"grad_norm": 1.935381293296814,
"learning_rate": 5.869375946627004e-05,
"loss": 0.1461,
"step": 639
},
{
"epoch": 1.22,
"grad_norm": 2.6404571533203125,
"learning_rate": 5.868836243563613e-05,
"loss": 0.1008,
"step": 640
},
{
"epoch": 1.22,
"grad_norm": 2.290581226348877,
"learning_rate": 5.868295452761519e-05,
"loss": 0.1814,
"step": 641
},
{
"epoch": 1.22,
"grad_norm": 1.9355552196502686,
"learning_rate": 5.8677535744257647e-05,
"loss": 0.1329,
"step": 642
},
{
"epoch": 1.22,
"grad_norm": 1.8912508487701416,
"learning_rate": 5.867210608761808e-05,
"loss": 0.1113,
"step": 643
},
{
"epoch": 1.22,
"grad_norm": 1.3006269931793213,
"learning_rate": 5.866666555975517e-05,
"loss": 0.0915,
"step": 644
},
{
"epoch": 1.23,
"grad_norm": 1.5112371444702148,
"learning_rate": 5.866121416273173e-05,
"loss": 0.0954,
"step": 645
},
{
"epoch": 1.23,
"grad_norm": 1.2579724788665771,
"learning_rate": 5.86557518986147e-05,
"loss": 0.0836,
"step": 646
},
{
"epoch": 1.23,
"grad_norm": 1.463584303855896,
"learning_rate": 5.865027876947514e-05,
"loss": 0.1377,
"step": 647
},
{
"epoch": 1.23,
"grad_norm": 1.8201812505722046,
"learning_rate": 5.86447947773882e-05,
"loss": 0.1394,
"step": 648
},
{
"epoch": 1.23,
"grad_norm": 1.8606367111206055,
"learning_rate": 5.863929992443321e-05,
"loss": 0.1372,
"step": 649
},
{
"epoch": 1.24,
"grad_norm": 1.5007331371307373,
"learning_rate": 5.863379421269354e-05,
"loss": 0.1081,
"step": 650
},
{
"epoch": 1.24,
"grad_norm": 1.7608414888381958,
"learning_rate": 5.862827764425675e-05,
"loss": 0.1359,
"step": 651
},
{
"epoch": 1.24,
"grad_norm": 1.5184133052825928,
"learning_rate": 5.8622750221214466e-05,
"loss": 0.1004,
"step": 652
},
{
"epoch": 1.24,
"grad_norm": 1.3966097831726074,
"learning_rate": 5.861721194566246e-05,
"loss": 0.0804,
"step": 653
},
{
"epoch": 1.24,
"grad_norm": 1.3854674100875854,
"learning_rate": 5.86116628197006e-05,
"loss": 0.1103,
"step": 654
},
{
"epoch": 1.25,
"grad_norm": 1.8577762842178345,
"learning_rate": 5.8606102845432875e-05,
"loss": 0.0975,
"step": 655
},
{
"epoch": 1.25,
"grad_norm": 1.5021672248840332,
"learning_rate": 5.8600532024967384e-05,
"loss": 0.0991,
"step": 656
},
{
"epoch": 1.25,
"grad_norm": 2.092601776123047,
"learning_rate": 5.859495036041635e-05,
"loss": 0.1731,
"step": 657
},
{
"epoch": 1.25,
"grad_norm": 1.9624136686325073,
"learning_rate": 5.858935785389609e-05,
"loss": 0.1278,
"step": 658
},
{
"epoch": 1.25,
"grad_norm": 1.6786785125732422,
"learning_rate": 5.8583754507527045e-05,
"loss": 0.1049,
"step": 659
},
{
"epoch": 1.25,
"grad_norm": 1.4157971143722534,
"learning_rate": 5.857814032343376e-05,
"loss": 0.0797,
"step": 660
},
{
"epoch": 1.26,
"grad_norm": 1.7263004779815674,
"learning_rate": 5.85725153037449e-05,
"loss": 0.0911,
"step": 661
},
{
"epoch": 1.26,
"grad_norm": 1.5280500650405884,
"learning_rate": 5.8566879450593216e-05,
"loss": 0.1019,
"step": 662
},
{
"epoch": 1.26,
"grad_norm": 1.906541109085083,
"learning_rate": 5.856123276611559e-05,
"loss": 0.1267,
"step": 663
},
{
"epoch": 1.26,
"grad_norm": 1.7652462720870972,
"learning_rate": 5.855557525245298e-05,
"loss": 0.1138,
"step": 664
},
{
"epoch": 1.26,
"grad_norm": 2.2186853885650635,
"learning_rate": 5.85499069117505e-05,
"loss": 0.1141,
"step": 665
},
{
"epoch": 1.27,
"grad_norm": 1.95181405544281,
"learning_rate": 5.8544227746157335e-05,
"loss": 0.1285,
"step": 666
},
{
"epoch": 1.27,
"grad_norm": 1.2935830354690552,
"learning_rate": 5.853853775782677e-05,
"loss": 0.0591,
"step": 667
},
{
"epoch": 1.27,
"grad_norm": 1.7702007293701172,
"learning_rate": 5.8532836948916186e-05,
"loss": 0.1244,
"step": 668
},
{
"epoch": 1.27,
"grad_norm": 1.6261893510818481,
"learning_rate": 5.85271253215871e-05,
"loss": 0.0804,
"step": 669
},
{
"epoch": 1.27,
"grad_norm": 1.394527792930603,
"learning_rate": 5.852140287800513e-05,
"loss": 0.0774,
"step": 670
},
{
"epoch": 1.28,
"grad_norm": 1.6781400442123413,
"learning_rate": 5.851566962033996e-05,
"loss": 0.1058,
"step": 671
},
{
"epoch": 1.28,
"grad_norm": 1.9328844547271729,
"learning_rate": 5.85099255507654e-05,
"loss": 0.1166,
"step": 672
},
{
"epoch": 1.28,
"grad_norm": 2.1044747829437256,
"learning_rate": 5.850417067145935e-05,
"loss": 0.123,
"step": 673
},
{
"epoch": 1.28,
"grad_norm": 1.8533934354782104,
"learning_rate": 5.84984049846038e-05,
"loss": 0.109,
"step": 674
},
{
"epoch": 1.28,
"grad_norm": 1.8706427812576294,
"learning_rate": 5.849262849238487e-05,
"loss": 0.116,
"step": 675
},
{
"epoch": 1.29,
"grad_norm": 1.244273066520691,
"learning_rate": 5.848684119699275e-05,
"loss": 0.0628,
"step": 676
},
{
"epoch": 1.29,
"grad_norm": 2.3993990421295166,
"learning_rate": 5.848104310062173e-05,
"loss": 0.1321,
"step": 677
},
{
"epoch": 1.29,
"grad_norm": 1.7353196144104004,
"learning_rate": 5.8475234205470195e-05,
"loss": 0.0818,
"step": 678
},
{
"epoch": 1.29,
"grad_norm": 1.837346076965332,
"learning_rate": 5.846941451374063e-05,
"loss": 0.0846,
"step": 679
},
{
"epoch": 1.29,
"grad_norm": 2.0827691555023193,
"learning_rate": 5.846358402763962e-05,
"loss": 0.0845,
"step": 680
},
{
"epoch": 1.29,
"grad_norm": 1.6402720212936401,
"learning_rate": 5.845774274937781e-05,
"loss": 0.1102,
"step": 681
},
{
"epoch": 1.3,
"grad_norm": 1.4540116786956787,
"learning_rate": 5.845189068116997e-05,
"loss": 0.0888,
"step": 682
},
{
"epoch": 1.3,
"grad_norm": 1.5979608297348022,
"learning_rate": 5.844602782523496e-05,
"loss": 0.1121,
"step": 683
},
{
"epoch": 1.3,
"grad_norm": 1.3434520959854126,
"learning_rate": 5.844015418379572e-05,
"loss": 0.0916,
"step": 684
},
{
"epoch": 1.3,
"grad_norm": 1.9180506467819214,
"learning_rate": 5.8434269759079274e-05,
"loss": 0.0977,
"step": 685
},
{
"epoch": 1.3,
"grad_norm": 2.338634490966797,
"learning_rate": 5.842837455331674e-05,
"loss": 0.1307,
"step": 686
},
{
"epoch": 1.31,
"grad_norm": 1.3670991659164429,
"learning_rate": 5.842246856874333e-05,
"loss": 0.0846,
"step": 687
},
{
"epoch": 1.31,
"grad_norm": 1.357681393623352,
"learning_rate": 5.841655180759835e-05,
"loss": 0.0742,
"step": 688
},
{
"epoch": 1.31,
"grad_norm": 1.5757499933242798,
"learning_rate": 5.8410624272125155e-05,
"loss": 0.0751,
"step": 689
},
{
"epoch": 1.31,
"grad_norm": 1.3674101829528809,
"learning_rate": 5.8404685964571236e-05,
"loss": 0.0814,
"step": 690
},
{
"epoch": 1.31,
"grad_norm": 1.9906283617019653,
"learning_rate": 5.839873688718813e-05,
"loss": 0.1513,
"step": 691
},
{
"epoch": 1.32,
"grad_norm": 2.0459611415863037,
"learning_rate": 5.8392777042231455e-05,
"loss": 0.0998,
"step": 692
},
{
"epoch": 1.32,
"grad_norm": 2.1719119548797607,
"learning_rate": 5.838680643196095e-05,
"loss": 0.1146,
"step": 693
},
{
"epoch": 1.32,
"grad_norm": 1.6034185886383057,
"learning_rate": 5.8380825058640406e-05,
"loss": 0.0884,
"step": 694
},
{
"epoch": 1.32,
"grad_norm": 1.620283603668213,
"learning_rate": 5.83748329245377e-05,
"loss": 0.0959,
"step": 695
},
{
"epoch": 1.32,
"grad_norm": 1.809004545211792,
"learning_rate": 5.8368830031924785e-05,
"loss": 0.0992,
"step": 696
},
{
"epoch": 1.33,
"grad_norm": 1.47053861618042,
"learning_rate": 5.836281638307771e-05,
"loss": 0.0652,
"step": 697
},
{
"epoch": 1.33,
"grad_norm": 2.576958656311035,
"learning_rate": 5.835679198027657e-05,
"loss": 0.1107,
"step": 698
},
{
"epoch": 1.33,
"grad_norm": 1.4772312641143799,
"learning_rate": 5.835075682580558e-05,
"loss": 0.083,
"step": 699
},
{
"epoch": 1.33,
"grad_norm": 1.9658222198486328,
"learning_rate": 5.8344710921953e-05,
"loss": 0.1419,
"step": 700
},
{
"epoch": 1.33,
"eval_blimp_filtered_avg": 0.7346268656716418,
"eval_blimp_filtered_std": 0.004856245940037265,
"step": 700
},
{
"epoch": 1.33,
"eval_blimp_supplement_avg": 0.7758620689655172,
"eval_blimp_supplement_std": 0.018196301348519527,
"step": 700
},
{
"epoch": 1.33,
"eval_vqa_filtered_avg": 0.26,
"eval_vqa_filtered_std": 0.04408440022768078,
"step": 700
},
{
"epoch": 1.33,
"eval_winoground_filtered_avg": 0.5,
"eval_winoground_filtered_std": 0.050251890762960605,
"step": 700
},
{
"epoch": 1.33,
"grad_norm": 2.036735773086548,
"learning_rate": 5.833865427101118e-05,
"loss": 0.0746,
"step": 701
},
{
"epoch": 1.33,
"grad_norm": 1.4296282529830933,
"learning_rate": 5.8332586875276535e-05,
"loss": 0.0755,
"step": 702
},
{
"epoch": 1.34,
"grad_norm": 2.426517963409424,
"learning_rate": 5.832650873704957e-05,
"loss": 0.1652,
"step": 703
},
{
"epoch": 1.34,
"grad_norm": 1.483805775642395,
"learning_rate": 5.832041985863483e-05,
"loss": 0.1037,
"step": 704
},
{
"epoch": 1.34,
"grad_norm": 2.052328586578369,
"learning_rate": 5.8314320242340964e-05,
"loss": 0.108,
"step": 705
},
{
"epoch": 1.34,
"grad_norm": 1.7976720333099365,
"learning_rate": 5.8308209890480706e-05,
"loss": 0.1293,
"step": 706
},
{
"epoch": 1.34,
"grad_norm": 1.8044564723968506,
"learning_rate": 5.8302088805370804e-05,
"loss": 0.1003,
"step": 707
},
{
"epoch": 1.35,
"grad_norm": 1.2838388681411743,
"learning_rate": 5.829595698933213e-05,
"loss": 0.0811,
"step": 708
},
{
"epoch": 1.35,
"grad_norm": 1.6621758937835693,
"learning_rate": 5.828981444468959e-05,
"loss": 0.1144,
"step": 709
},
{
"epoch": 1.35,
"grad_norm": 1.3415857553482056,
"learning_rate": 5.828366117377218e-05,
"loss": 0.0781,
"step": 710
},
{
"epoch": 1.35,
"grad_norm": 1.6506767272949219,
"learning_rate": 5.827749717891296e-05,
"loss": 0.1297,
"step": 711
},
{
"epoch": 1.35,
"grad_norm": 1.8295927047729492,
"learning_rate": 5.8271322462449034e-05,
"loss": 0.1013,
"step": 712
},
{
"epoch": 1.36,
"grad_norm": 1.6573423147201538,
"learning_rate": 5.8265137026721606e-05,
"loss": 0.0889,
"step": 713
},
{
"epoch": 1.36,
"grad_norm": 1.723860502243042,
"learning_rate": 5.825894087407591e-05,
"loss": 0.1078,
"step": 714
},
{
"epoch": 1.36,
"grad_norm": 1.5366859436035156,
"learning_rate": 5.825273400686127e-05,
"loss": 0.1074,
"step": 715
},
{
"epoch": 1.36,
"grad_norm": 1.6328061819076538,
"learning_rate": 5.8246516427431056e-05,
"loss": 0.1339,
"step": 716
},
{
"epoch": 1.36,
"grad_norm": 1.6423825025558472,
"learning_rate": 5.8240288138142704e-05,
"loss": 0.1141,
"step": 717
},
{
"epoch": 1.37,
"grad_norm": 1.6472768783569336,
"learning_rate": 5.823404914135771e-05,
"loss": 0.0936,
"step": 718
},
{
"epoch": 1.37,
"grad_norm": 1.6298952102661133,
"learning_rate": 5.822779943944165e-05,
"loss": 0.1231,
"step": 719
},
{
"epoch": 1.37,
"grad_norm": 1.3124911785125732,
"learning_rate": 5.822153903476412e-05,
"loss": 0.0761,
"step": 720
},
{
"epoch": 1.37,
"grad_norm": 1.8522390127182007,
"learning_rate": 5.8215267929698804e-05,
"loss": 0.102,
"step": 721
},
{
"epoch": 1.37,
"grad_norm": 1.7672533988952637,
"learning_rate": 5.820898612662343e-05,
"loss": 0.0914,
"step": 722
},
{
"epoch": 1.37,
"grad_norm": 1.6756874322891235,
"learning_rate": 5.820269362791979e-05,
"loss": 0.0912,
"step": 723
},
{
"epoch": 1.38,
"grad_norm": 1.658960223197937,
"learning_rate": 5.819639043597372e-05,
"loss": 0.0834,
"step": 724
},
{
"epoch": 1.38,
"grad_norm": 2.218472957611084,
"learning_rate": 5.8190076553175146e-05,
"loss": 0.1134,
"step": 725
},
{
"epoch": 1.38,
"grad_norm": 1.535555362701416,
"learning_rate": 5.818375198191798e-05,
"loss": 0.0869,
"step": 726
},
{
"epoch": 1.38,
"grad_norm": 2.039428234100342,
"learning_rate": 5.817741672460025e-05,
"loss": 0.1009,
"step": 727
},
{
"epoch": 1.38,
"grad_norm": 2.052220344543457,
"learning_rate": 5.8171070783624e-05,
"loss": 0.0972,
"step": 728
},
{
"epoch": 1.39,
"grad_norm": 1.3047471046447754,
"learning_rate": 5.8164714161395346e-05,
"loss": 0.08,
"step": 729
},
{
"epoch": 1.39,
"grad_norm": 1.403720498085022,
"learning_rate": 5.815834686032444e-05,
"loss": 0.1098,
"step": 730
},
{
"epoch": 1.39,
"grad_norm": 2.8895301818847656,
"learning_rate": 5.815196888282549e-05,
"loss": 0.1279,
"step": 731
},
{
"epoch": 1.39,
"grad_norm": 1.6551228761672974,
"learning_rate": 5.814558023131674e-05,
"loss": 0.0873,
"step": 732
},
{
"epoch": 1.39,
"grad_norm": 2.0921037197113037,
"learning_rate": 5.81391809082205e-05,
"loss": 0.1337,
"step": 733
},
{
"epoch": 1.4,
"grad_norm": 1.652608871459961,
"learning_rate": 5.813277091596311e-05,
"loss": 0.1034,
"step": 734
},
{
"epoch": 1.4,
"grad_norm": 1.74064040184021,
"learning_rate": 5.812635025697498e-05,
"loss": 0.1131,
"step": 735
},
{
"epoch": 1.4,
"grad_norm": 1.613586187362671,
"learning_rate": 5.811991893369053e-05,
"loss": 0.1108,
"step": 736
},
{
"epoch": 1.4,
"grad_norm": 2.0762531757354736,
"learning_rate": 5.811347694854823e-05,
"loss": 0.1323,
"step": 737
},
{
"epoch": 1.4,
"grad_norm": 2.0046579837799072,
"learning_rate": 5.810702430399063e-05,
"loss": 0.1261,
"step": 738
},
{
"epoch": 1.4,
"grad_norm": 1.8749340772628784,
"learning_rate": 5.810056100246427e-05,
"loss": 0.1269,
"step": 739
},
{
"epoch": 1.41,
"grad_norm": 1.8523480892181396,
"learning_rate": 5.8094087046419765e-05,
"loss": 0.1186,
"step": 740
},
{
"epoch": 1.41,
"grad_norm": 1.709141492843628,
"learning_rate": 5.808760243831176e-05,
"loss": 0.1177,
"step": 741
},
{
"epoch": 1.41,
"grad_norm": 1.6003118753433228,
"learning_rate": 5.808110718059894e-05,
"loss": 0.1039,
"step": 742
},
{
"epoch": 1.41,
"grad_norm": 1.8025463819503784,
"learning_rate": 5.807460127574401e-05,
"loss": 0.0908,
"step": 743
},
{
"epoch": 1.41,
"grad_norm": 1.9997127056121826,
"learning_rate": 5.806808472621374e-05,
"loss": 0.1342,
"step": 744
},
{
"epoch": 1.42,
"grad_norm": 1.577642798423767,
"learning_rate": 5.806155753447894e-05,
"loss": 0.11,
"step": 745
},
{
"epoch": 1.42,
"grad_norm": 1.633766531944275,
"learning_rate": 5.8055019703014406e-05,
"loss": 0.0976,
"step": 746
},
{
"epoch": 1.42,
"grad_norm": 1.7557709217071533,
"learning_rate": 5.804847123429902e-05,
"loss": 0.1053,
"step": 747
},
{
"epoch": 1.42,
"grad_norm": 1.7026622295379639,
"learning_rate": 5.804191213081569e-05,
"loss": 0.0886,
"step": 748
},
{
"epoch": 1.42,
"grad_norm": 1.639654278755188,
"learning_rate": 5.803534239505132e-05,
"loss": 0.1595,
"step": 749
},
{
"epoch": 1.43,
"grad_norm": 2.0295403003692627,
"learning_rate": 5.8028762029496884e-05,
"loss": 0.1457,
"step": 750
},
{
"epoch": 1.43,
"grad_norm": 1.9312379360198975,
"learning_rate": 5.802217103664736e-05,
"loss": 0.1061,
"step": 751
},
{
"epoch": 1.43,
"grad_norm": 1.4533945322036743,
"learning_rate": 5.8015569419001784e-05,
"loss": 0.0743,
"step": 752
},
{
"epoch": 1.43,
"grad_norm": 2.051530122756958,
"learning_rate": 5.8008957179063194e-05,
"loss": 0.1022,
"step": 753
},
{
"epoch": 1.43,
"grad_norm": 1.8426611423492432,
"learning_rate": 5.800233431933867e-05,
"loss": 0.1141,
"step": 754
},
{
"epoch": 1.44,
"grad_norm": 1.7045934200286865,
"learning_rate": 5.799570084233931e-05,
"loss": 0.1022,
"step": 755
},
{
"epoch": 1.44,
"grad_norm": 1.4449961185455322,
"learning_rate": 5.7989056750580245e-05,
"loss": 0.0726,
"step": 756
},
{
"epoch": 1.44,
"grad_norm": 2.2863986492156982,
"learning_rate": 5.7982402046580627e-05,
"loss": 0.1072,
"step": 757
},
{
"epoch": 1.44,
"grad_norm": 1.748349666595459,
"learning_rate": 5.7975736732863626e-05,
"loss": 0.099,
"step": 758
},
{
"epoch": 1.44,
"grad_norm": 1.9685564041137695,
"learning_rate": 5.796906081195645e-05,
"loss": 0.115,
"step": 759
},
{
"epoch": 1.44,
"grad_norm": 2.135883092880249,
"learning_rate": 5.796237428639032e-05,
"loss": 0.1499,
"step": 760
},
{
"epoch": 1.45,
"grad_norm": 1.4097079038619995,
"learning_rate": 5.795567715870047e-05,
"loss": 0.0873,
"step": 761
},
{
"epoch": 1.45,
"grad_norm": 1.5826094150543213,
"learning_rate": 5.794896943142616e-05,
"loss": 0.0884,
"step": 762
},
{
"epoch": 1.45,
"grad_norm": 1.795210838317871,
"learning_rate": 5.794225110711068e-05,
"loss": 0.1032,
"step": 763
},
{
"epoch": 1.45,
"grad_norm": 1.369030237197876,
"learning_rate": 5.793552218830132e-05,
"loss": 0.0756,
"step": 764
},
{
"epoch": 1.45,
"grad_norm": 1.6064612865447998,
"learning_rate": 5.7928782677549404e-05,
"loss": 0.1058,
"step": 765
},
{
"epoch": 1.46,
"grad_norm": 2.2267277240753174,
"learning_rate": 5.792203257741025e-05,
"loss": 0.1117,
"step": 766
},
{
"epoch": 1.46,
"grad_norm": 1.7661080360412598,
"learning_rate": 5.791527189044322e-05,
"loss": 0.1087,
"step": 767
},
{
"epoch": 1.46,
"grad_norm": 2.068272352218628,
"learning_rate": 5.790850061921166e-05,
"loss": 0.1124,
"step": 768
},
{
"epoch": 1.46,
"grad_norm": 1.5895425081253052,
"learning_rate": 5.790171876628295e-05,
"loss": 0.1029,
"step": 769
},
{
"epoch": 1.46,
"grad_norm": 1.7508352994918823,
"learning_rate": 5.789492633422848e-05,
"loss": 0.0985,
"step": 770
},
{
"epoch": 1.47,
"grad_norm": 1.385257601737976,
"learning_rate": 5.788812332562364e-05,
"loss": 0.0713,
"step": 771
},
{
"epoch": 1.47,
"grad_norm": 2.0284268856048584,
"learning_rate": 5.788130974304783e-05,
"loss": 0.1523,
"step": 772
},
{
"epoch": 1.47,
"grad_norm": 1.7096168994903564,
"learning_rate": 5.787448558908449e-05,
"loss": 0.0956,
"step": 773
},
{
"epoch": 1.47,
"grad_norm": 1.6719380617141724,
"learning_rate": 5.786765086632102e-05,
"loss": 0.096,
"step": 774
},
{
"epoch": 1.47,
"grad_norm": 1.8120020627975464,
"learning_rate": 5.786080557734885e-05,
"loss": 0.1126,
"step": 775
},
{
"epoch": 1.48,
"grad_norm": 2.207967519760132,
"learning_rate": 5.7853949724763436e-05,
"loss": 0.1043,
"step": 776
},
{
"epoch": 1.48,
"grad_norm": 2.180513381958008,
"learning_rate": 5.784708331116421e-05,
"loss": 0.1086,
"step": 777
},
{
"epoch": 1.48,
"grad_norm": 2.1063060760498047,
"learning_rate": 5.784020633915463e-05,
"loss": 0.1345,
"step": 778
},
{
"epoch": 1.48,
"grad_norm": 1.4697365760803223,
"learning_rate": 5.783331881134213e-05,
"loss": 0.076,
"step": 779
},
{
"epoch": 1.48,
"grad_norm": 1.8634710311889648,
"learning_rate": 5.782642073033818e-05,
"loss": 0.1519,
"step": 780
},
{
"epoch": 1.48,
"grad_norm": 1.8060096502304077,
"learning_rate": 5.781951209875822e-05,
"loss": 0.0789,
"step": 781
},
{
"epoch": 1.49,
"grad_norm": 1.8313156366348267,
"learning_rate": 5.781259291922171e-05,
"loss": 0.0893,
"step": 782
},
{
"epoch": 1.49,
"grad_norm": 2.105073928833008,
"learning_rate": 5.78056631943521e-05,
"loss": 0.1314,
"step": 783
},
{
"epoch": 1.49,
"grad_norm": 11.344228744506836,
"learning_rate": 5.779872292677686e-05,
"loss": 0.1159,
"step": 784
},
{
"epoch": 1.49,
"grad_norm": 2.1268560886383057,
"learning_rate": 5.779177211912741e-05,
"loss": 0.1628,
"step": 785
},
{
"epoch": 1.49,
"grad_norm": 2.3884365558624268,
"learning_rate": 5.7784810774039216e-05,
"loss": 0.1586,
"step": 786
},
{
"epoch": 1.5,
"grad_norm": 2.1637678146362305,
"learning_rate": 5.7777838894151715e-05,
"loss": 0.1313,
"step": 787
},
{
"epoch": 1.5,
"grad_norm": 2.380366563796997,
"learning_rate": 5.777085648210835e-05,
"loss": 0.1111,
"step": 788
},
{
"epoch": 1.5,
"grad_norm": 1.4485961198806763,
"learning_rate": 5.776386354055653e-05,
"loss": 0.0923,
"step": 789
},
{
"epoch": 1.5,
"grad_norm": 2.169799327850342,
"learning_rate": 5.77568600721477e-05,
"loss": 0.1256,
"step": 790
},
{
"epoch": 1.5,
"grad_norm": 1.6244041919708252,
"learning_rate": 5.774984607953725e-05,
"loss": 0.1233,
"step": 791
},
{
"epoch": 1.51,
"grad_norm": 3.648930788040161,
"learning_rate": 5.77428215653846e-05,
"loss": 0.1014,
"step": 792
},
{
"epoch": 1.51,
"grad_norm": 1.9726791381835938,
"learning_rate": 5.7735786532353125e-05,
"loss": 0.1334,
"step": 793
},
{
"epoch": 1.51,
"grad_norm": 2.4793238639831543,
"learning_rate": 5.772874098311023e-05,
"loss": 0.1391,
"step": 794
},
{
"epoch": 1.51,
"grad_norm": 1.6665037870407104,
"learning_rate": 5.772168492032726e-05,
"loss": 0.1201,
"step": 795
},
{
"epoch": 1.51,
"grad_norm": 2.233578681945801,
"learning_rate": 5.771461834667958e-05,
"loss": 0.1932,
"step": 796
},
{
"epoch": 1.52,
"grad_norm": 2.538400888442993,
"learning_rate": 5.770754126484652e-05,
"loss": 0.1481,
"step": 797
},
{
"epoch": 1.52,
"grad_norm": 1.6307553052902222,
"learning_rate": 5.770045367751142e-05,
"loss": 0.1061,
"step": 798
},
{
"epoch": 1.52,
"grad_norm": 2.5792276859283447,
"learning_rate": 5.7693355587361576e-05,
"loss": 0.1276,
"step": 799
},
{
"epoch": 1.52,
"grad_norm": 2.149144411087036,
"learning_rate": 5.768624699708827e-05,
"loss": 0.0926,
"step": 800
},
{
"epoch": 1.52,
"eval_blimp_filtered_avg": 0.7179104477611941,
"eval_blimp_filtered_std": 0.004999938119960643,
"step": 800
},
{
"epoch": 1.52,
"eval_blimp_supplement_avg": 0.7758620689655172,
"eval_blimp_supplement_std": 0.018304357108194665,
"step": 800
},
{
"epoch": 1.52,
"eval_vqa_filtered_avg": 0.35,
"eval_vqa_filtered_std": 0.047937248544110196,
"step": 800
},
{
"epoch": 1.52,
"eval_winoground_filtered_avg": 0.57,
"eval_winoground_filtered_std": 0.049756985195624284,
"step": 800
},
{
"epoch": 1.52,
"grad_norm": 6.146359920501709,
"learning_rate": 5.7679127909386785e-05,
"loss": 0.1788,
"step": 801
},
{
"epoch": 1.52,
"grad_norm": 3.2608888149261475,
"learning_rate": 5.767199832695637e-05,
"loss": 0.1954,
"step": 802
},
{
"epoch": 1.53,
"grad_norm": 3.0002334117889404,
"learning_rate": 5.766485825250025e-05,
"loss": 0.1433,
"step": 803
},
{
"epoch": 1.53,
"grad_norm": 3.0024967193603516,
"learning_rate": 5.7657707688725635e-05,
"loss": 0.1565,
"step": 804
},
{
"epoch": 1.53,
"grad_norm": 2.2252197265625,
"learning_rate": 5.765054663834371e-05,
"loss": 0.1418,
"step": 805
},
{
"epoch": 1.53,
"grad_norm": 2.04168438911438,
"learning_rate": 5.7643375104069635e-05,
"loss": 0.1262,
"step": 806
},
{
"epoch": 1.53,
"grad_norm": 1.8204281330108643,
"learning_rate": 5.763619308862254e-05,
"loss": 0.1278,
"step": 807
},
{
"epoch": 1.54,
"grad_norm": 2.2390406131744385,
"learning_rate": 5.7629000594725536e-05,
"loss": 0.1718,
"step": 808
},
{
"epoch": 1.54,
"grad_norm": 2.245917558670044,
"learning_rate": 5.7621797625105724e-05,
"loss": 0.1169,
"step": 809
},
{
"epoch": 1.54,
"grad_norm": 2.36124324798584,
"learning_rate": 5.7614584182494135e-05,
"loss": 0.1543,
"step": 810
},
{
"epoch": 1.54,
"grad_norm": 1.9892337322235107,
"learning_rate": 5.7607360269625804e-05,
"loss": 0.139,
"step": 811
},
{
"epoch": 1.54,
"grad_norm": 1.6522183418273926,
"learning_rate": 5.760012588923973e-05,
"loss": 0.1063,
"step": 812
},
{
"epoch": 1.55,
"grad_norm": 2.3076581954956055,
"learning_rate": 5.759288104407887e-05,
"loss": 0.1498,
"step": 813
},
{
"epoch": 1.55,
"grad_norm": 1.800788164138794,
"learning_rate": 5.7585625736890165e-05,
"loss": 0.1231,
"step": 814
},
{
"epoch": 1.55,
"grad_norm": 2.0470104217529297,
"learning_rate": 5.757835997042451e-05,
"loss": 0.1169,
"step": 815
},
{
"epoch": 1.55,
"grad_norm": 1.4313958883285522,
"learning_rate": 5.757108374743677e-05,
"loss": 0.0932,
"step": 816
},
{
"epoch": 1.55,
"grad_norm": 1.9654278755187988,
"learning_rate": 5.7563797070685765e-05,
"loss": 0.1177,
"step": 817
},
{
"epoch": 1.56,
"grad_norm": 1.8501168489456177,
"learning_rate": 5.75564999429343e-05,
"loss": 0.1071,
"step": 818
},
{
"epoch": 1.56,
"grad_norm": 2.0504648685455322,
"learning_rate": 5.7549192366949126e-05,
"loss": 0.1522,
"step": 819
},
{
"epoch": 1.56,
"grad_norm": 2.14029598236084,
"learning_rate": 5.754187434550096e-05,
"loss": 0.1398,
"step": 820
},
{
"epoch": 1.56,
"grad_norm": 2.0163114070892334,
"learning_rate": 5.7534545881364486e-05,
"loss": 0.1061,
"step": 821
},
{
"epoch": 1.56,
"grad_norm": 1.744583010673523,
"learning_rate": 5.752720697731833e-05,
"loss": 0.0896,
"step": 822
},
{
"epoch": 1.56,
"grad_norm": 2.097350597381592,
"learning_rate": 5.751985763614509e-05,
"loss": 0.1394,
"step": 823
},
{
"epoch": 1.57,
"grad_norm": 1.7897913455963135,
"learning_rate": 5.7512497860631316e-05,
"loss": 0.1256,
"step": 824
},
{
"epoch": 1.57,
"grad_norm": 1.8421887159347534,
"learning_rate": 5.750512765356753e-05,
"loss": 0.1157,
"step": 825
},
{
"epoch": 1.57,
"grad_norm": 2.095339059829712,
"learning_rate": 5.7497747017748185e-05,
"loss": 0.1321,
"step": 826
},
{
"epoch": 1.57,
"grad_norm": 1.8623578548431396,
"learning_rate": 5.7490355955971696e-05,
"loss": 0.0858,
"step": 827
},
{
"epoch": 1.57,
"grad_norm": 1.8074578046798706,
"learning_rate": 5.748295447104044e-05,
"loss": 0.1107,
"step": 828
},
{
"epoch": 1.58,
"grad_norm": 1.9986863136291504,
"learning_rate": 5.747554256576075e-05,
"loss": 0.1111,
"step": 829
},
{
"epoch": 1.58,
"grad_norm": 2.1539547443389893,
"learning_rate": 5.746812024294288e-05,
"loss": 0.105,
"step": 830
},
{
"epoch": 1.58,
"grad_norm": 2.100527286529541,
"learning_rate": 5.746068750540106e-05,
"loss": 0.0963,
"step": 831
},
{
"epoch": 1.58,
"grad_norm": 2.106220245361328,
"learning_rate": 5.745324435595348e-05,
"loss": 0.1536,
"step": 832
},
{
"epoch": 1.58,
"grad_norm": 1.7538392543792725,
"learning_rate": 5.744579079742224e-05,
"loss": 0.1218,
"step": 833
},
{
"epoch": 1.59,
"grad_norm": 1.7100622653961182,
"learning_rate": 5.7438326832633414e-05,
"loss": 0.1182,
"step": 834
},
{
"epoch": 1.59,
"grad_norm": 2.4272661209106445,
"learning_rate": 5.7430852464417006e-05,
"loss": 0.1268,
"step": 835
},
{
"epoch": 1.59,
"grad_norm": 2.110100030899048,
"learning_rate": 5.7423367695607e-05,
"loss": 0.1171,
"step": 836
},
{
"epoch": 1.59,
"grad_norm": 1.6885584592819214,
"learning_rate": 5.741587252904127e-05,
"loss": 0.1146,
"step": 837
},
{
"epoch": 1.59,
"grad_norm": 2.280898332595825,
"learning_rate": 5.740836696756167e-05,
"loss": 0.1342,
"step": 838
},
{
"epoch": 1.6,
"grad_norm": 2.4058070182800293,
"learning_rate": 5.740085101401397e-05,
"loss": 0.1563,
"step": 839
},
{
"epoch": 1.6,
"grad_norm": 2.0147788524627686,
"learning_rate": 5.7393324671247926e-05,
"loss": 0.1346,
"step": 840
},
{
"epoch": 1.6,
"grad_norm": 2.8497121334075928,
"learning_rate": 5.738578794211718e-05,
"loss": 0.1432,
"step": 841
},
{
"epoch": 1.6,
"grad_norm": 2.0407824516296387,
"learning_rate": 5.737824082947933e-05,
"loss": 0.1425,
"step": 842
},
{
"epoch": 1.6,
"grad_norm": 1.8581652641296387,
"learning_rate": 5.7370683336195925e-05,
"loss": 0.0946,
"step": 843
},
{
"epoch": 1.6,
"grad_norm": 1.6785504817962646,
"learning_rate": 5.736311546513244e-05,
"loss": 0.116,
"step": 844
},
{
"epoch": 1.61,
"grad_norm": 1.8900675773620605,
"learning_rate": 5.735553721915828e-05,
"loss": 0.1238,
"step": 845
},
{
"epoch": 1.61,
"grad_norm": 1.6115425825119019,
"learning_rate": 5.734794860114679e-05,
"loss": 0.1096,
"step": 846
},
{
"epoch": 1.61,
"grad_norm": 2.0189406871795654,
"learning_rate": 5.7340349613975245e-05,
"loss": 0.1658,
"step": 847
},
{
"epoch": 1.61,
"grad_norm": 1.8827282190322876,
"learning_rate": 5.7332740260524855e-05,
"loss": 0.119,
"step": 848
},
{
"epoch": 1.61,
"grad_norm": 1.633846640586853,
"learning_rate": 5.7325120543680754e-05,
"loss": 0.0922,
"step": 849
},
{
"epoch": 1.62,
"grad_norm": 1.8015508651733398,
"learning_rate": 5.731749046633201e-05,
"loss": 0.1425,
"step": 850
},
{
"epoch": 1.62,
"grad_norm": 2.207883596420288,
"learning_rate": 5.730985003137163e-05,
"loss": 0.1148,
"step": 851
},
{
"epoch": 1.62,
"grad_norm": 2.0439696311950684,
"learning_rate": 5.730219924169652e-05,
"loss": 0.0797,
"step": 852
},
{
"epoch": 1.62,
"grad_norm": 1.7458409070968628,
"learning_rate": 5.729453810020753e-05,
"loss": 0.1047,
"step": 853
},
{
"epoch": 1.62,
"grad_norm": 1.9059698581695557,
"learning_rate": 5.728686660980945e-05,
"loss": 0.1216,
"step": 854
},
{
"epoch": 1.63,
"grad_norm": 2.0925211906433105,
"learning_rate": 5.7279184773410966e-05,
"loss": 0.1084,
"step": 855
},
{
"epoch": 1.63,
"grad_norm": 1.835500955581665,
"learning_rate": 5.727149259392469e-05,
"loss": 0.1087,
"step": 856
},
{
"epoch": 1.63,
"grad_norm": 1.6592069864273071,
"learning_rate": 5.726379007426718e-05,
"loss": 0.1094,
"step": 857
},
{
"epoch": 1.63,
"grad_norm": 1.9059797525405884,
"learning_rate": 5.725607721735889e-05,
"loss": 0.1281,
"step": 858
},
{
"epoch": 1.63,
"grad_norm": 1.655903935432434,
"learning_rate": 5.724835402612422e-05,
"loss": 0.1017,
"step": 859
},
{
"epoch": 1.63,
"grad_norm": 1.948338508605957,
"learning_rate": 5.724062050349143e-05,
"loss": 0.1564,
"step": 860
},
{
"epoch": 1.64,
"grad_norm": 1.9377073049545288,
"learning_rate": 5.723287665239277e-05,
"loss": 0.1494,
"step": 861
},
{
"epoch": 1.64,
"grad_norm": 1.5197291374206543,
"learning_rate": 5.722512247576436e-05,
"loss": 0.1081,
"step": 862
},
{
"epoch": 1.64,
"grad_norm": 1.9834606647491455,
"learning_rate": 5.7217357976546265e-05,
"loss": 0.1217,
"step": 863
},
{
"epoch": 1.64,
"grad_norm": 1.7538058757781982,
"learning_rate": 5.720958315768243e-05,
"loss": 0.1613,
"step": 864
},
{
"epoch": 1.64,
"grad_norm": 1.4824787378311157,
"learning_rate": 5.720179802212074e-05,
"loss": 0.0908,
"step": 865
},
{
"epoch": 1.65,
"grad_norm": 1.3277506828308105,
"learning_rate": 5.7194002572812983e-05,
"loss": 0.0882,
"step": 866
},
{
"epoch": 1.65,
"grad_norm": 1.612392544746399,
"learning_rate": 5.718619681271484e-05,
"loss": 0.1221,
"step": 867
},
{
"epoch": 1.65,
"grad_norm": 1.8367537260055542,
"learning_rate": 5.717838074478593e-05,
"loss": 0.1067,
"step": 868
},
{
"epoch": 1.65,
"grad_norm": 1.5855368375778198,
"learning_rate": 5.717055437198977e-05,
"loss": 0.0794,
"step": 869
},
{
"epoch": 1.65,
"grad_norm": 1.3612979650497437,
"learning_rate": 5.716271769729379e-05,
"loss": 0.076,
"step": 870
},
{
"epoch": 1.66,
"grad_norm": 1.6545333862304688,
"learning_rate": 5.71548707236693e-05,
"loss": 0.1099,
"step": 871
},
{
"epoch": 1.66,
"grad_norm": 1.4889581203460693,
"learning_rate": 5.714701345409155e-05,
"loss": 0.0943,
"step": 872
},
{
"epoch": 1.66,
"grad_norm": 7.103367328643799,
"learning_rate": 5.7139145891539656e-05,
"loss": 0.2061,
"step": 873
},
{
"epoch": 1.66,
"grad_norm": 2.9664530754089355,
"learning_rate": 5.7131268038996684e-05,
"loss": 0.11,
"step": 874
},
{
"epoch": 1.66,
"grad_norm": 1.8900036811828613,
"learning_rate": 5.7123379899449554e-05,
"loss": 0.1629,
"step": 875
},
{
"epoch": 1.67,
"grad_norm": 1.831352949142456,
"learning_rate": 5.711548147588911e-05,
"loss": 0.1081,
"step": 876
},
{
"epoch": 1.67,
"grad_norm": 2.254328966140747,
"learning_rate": 5.710757277131011e-05,
"loss": 0.1374,
"step": 877
},
{
"epoch": 1.67,
"grad_norm": 2.1936252117156982,
"learning_rate": 5.709965378871117e-05,
"loss": 0.1549,
"step": 878
},
{
"epoch": 1.67,
"grad_norm": 1.710563063621521,
"learning_rate": 5.709172453109484e-05,
"loss": 0.0942,
"step": 879
},
{
"epoch": 1.67,
"grad_norm": 1.8462220430374146,
"learning_rate": 5.7083785001467545e-05,
"loss": 0.1078,
"step": 880
},
{
"epoch": 1.67,
"grad_norm": 2.092045783996582,
"learning_rate": 5.7075835202839626e-05,
"loss": 0.1595,
"step": 881
},
{
"epoch": 1.68,
"grad_norm": 1.627698540687561,
"learning_rate": 5.706787513822528e-05,
"loss": 0.1094,
"step": 882
},
{
"epoch": 1.68,
"grad_norm": 2.0072085857391357,
"learning_rate": 5.705990481064264e-05,
"loss": 0.1263,
"step": 883
},
{
"epoch": 1.68,
"grad_norm": 1.6844136714935303,
"learning_rate": 5.7051924223113704e-05,
"loss": 0.0964,
"step": 884
},
{
"epoch": 1.68,
"grad_norm": 1.6155954599380493,
"learning_rate": 5.704393337866436e-05,
"loss": 0.1185,
"step": 885
},
{
"epoch": 1.68,
"grad_norm": 2.2645368576049805,
"learning_rate": 5.703593228032439e-05,
"loss": 0.1115,
"step": 886
},
{
"epoch": 1.69,
"grad_norm": 1.909521460533142,
"learning_rate": 5.702792093112747e-05,
"loss": 0.1318,
"step": 887
},
{
"epoch": 1.69,
"grad_norm": 1.7702713012695312,
"learning_rate": 5.701989933411116e-05,
"loss": 0.116,
"step": 888
},
{
"epoch": 1.69,
"grad_norm": 1.8525720834732056,
"learning_rate": 5.701186749231689e-05,
"loss": 0.094,
"step": 889
},
{
"epoch": 1.69,
"grad_norm": 1.9026999473571777,
"learning_rate": 5.700382540879e-05,
"loss": 0.1224,
"step": 890
},
{
"epoch": 1.69,
"grad_norm": 1.3891780376434326,
"learning_rate": 5.69957730865797e-05,
"loss": 0.0888,
"step": 891
},
{
"epoch": 1.7,
"grad_norm": 1.6546486616134644,
"learning_rate": 5.698771052873908e-05,
"loss": 0.1115,
"step": 892
},
{
"epoch": 1.7,
"grad_norm": 1.6816213130950928,
"learning_rate": 5.697963773832511e-05,
"loss": 0.118,
"step": 893
},
{
"epoch": 1.7,
"grad_norm": 2.3506553173065186,
"learning_rate": 5.6971554718398644e-05,
"loss": 0.1988,
"step": 894
},
{
"epoch": 1.7,
"grad_norm": 1.9294542074203491,
"learning_rate": 5.6963461472024415e-05,
"loss": 0.1081,
"step": 895
},
{
"epoch": 1.7,
"grad_norm": 1.8704988956451416,
"learning_rate": 5.6955358002271036e-05,
"loss": 0.0996,
"step": 896
},
{
"epoch": 1.71,
"grad_norm": 1.1524009704589844,
"learning_rate": 5.6947244312210994e-05,
"loss": 0.059,
"step": 897
},
{
"epoch": 1.71,
"grad_norm": 1.779471755027771,
"learning_rate": 5.693912040492063e-05,
"loss": 0.1333,
"step": 898
},
{
"epoch": 1.71,
"grad_norm": 1.6911739110946655,
"learning_rate": 5.6930986283480205e-05,
"loss": 0.1189,
"step": 899
},
{
"epoch": 1.71,
"grad_norm": 2.1767501831054688,
"learning_rate": 5.692284195097381e-05,
"loss": 0.148,
"step": 900
},
{
"epoch": 1.71,
"eval_blimp_filtered_avg": 0.7194029850746269,
"eval_blimp_filtered_std": 0.004984523008432103,
"step": 900
},
{
"epoch": 1.71,
"eval_blimp_supplement_avg": 0.7564655172413793,
"eval_blimp_supplement_std": 0.018366696623239677,
"step": 900
},
{
"epoch": 1.71,
"eval_vqa_filtered_avg": 0.37,
"eval_vqa_filtered_std": 0.04852365870939099,
"step": 900
},
{
"epoch": 1.71,
"eval_winoground_filtered_avg": 0.49,
"eval_winoground_filtered_std": 0.05024183937956912,
"step": 900
},
{
"epoch": 1.71,
"grad_norm": 1.4335408210754395,
"learning_rate": 5.691468741048943e-05,
"loss": 0.0931,
"step": 901
},
{
"epoch": 1.71,
"grad_norm": 1.8907419443130493,
"learning_rate": 5.6906522665118915e-05,
"loss": 0.1118,
"step": 902
},
{
"epoch": 1.72,
"grad_norm": 1.1042141914367676,
"learning_rate": 5.6898347717957986e-05,
"loss": 0.0717,
"step": 903
},
{
"epoch": 1.72,
"grad_norm": 1.4695698022842407,
"learning_rate": 5.6890162572106214e-05,
"loss": 0.0865,
"step": 904
},
{
"epoch": 1.72,
"grad_norm": 2.144843339920044,
"learning_rate": 5.688196723066707e-05,
"loss": 0.1702,
"step": 905
},
{
"epoch": 1.72,
"grad_norm": 1.8307011127471924,
"learning_rate": 5.687376169674786e-05,
"loss": 0.1256,
"step": 906
},
{
"epoch": 1.72,
"grad_norm": 1.7023776769638062,
"learning_rate": 5.686554597345978e-05,
"loss": 0.1262,
"step": 907
},
{
"epoch": 1.73,
"grad_norm": 1.3729526996612549,
"learning_rate": 5.685732006391787e-05,
"loss": 0.1015,
"step": 908
},
{
"epoch": 1.73,
"grad_norm": 1.6433268785476685,
"learning_rate": 5.684908397124104e-05,
"loss": 0.0988,
"step": 909
},
{
"epoch": 1.73,
"grad_norm": 1.4760923385620117,
"learning_rate": 5.6840837698552064e-05,
"loss": 0.1208,
"step": 910
},
{
"epoch": 1.73,
"grad_norm": 1.6141918897628784,
"learning_rate": 5.6832581248977566e-05,
"loss": 0.116,
"step": 911
},
{
"epoch": 1.73,
"grad_norm": 1.8932074308395386,
"learning_rate": 5.682431462564805e-05,
"loss": 0.1107,
"step": 912
},
{
"epoch": 1.74,
"grad_norm": 1.5878970623016357,
"learning_rate": 5.6816037831697837e-05,
"loss": 0.0945,
"step": 913
},
{
"epoch": 1.74,
"grad_norm": 2.176020860671997,
"learning_rate": 5.680775087026514e-05,
"loss": 0.1175,
"step": 914
},
{
"epoch": 1.74,
"grad_norm": 1.5189670324325562,
"learning_rate": 5.6799453744492035e-05,
"loss": 0.0831,
"step": 915
},
{
"epoch": 1.74,
"grad_norm": 1.6576449871063232,
"learning_rate": 5.6791146457524404e-05,
"loss": 0.0832,
"step": 916
},
{
"epoch": 1.74,
"grad_norm": 1.9738123416900635,
"learning_rate": 5.678282901251203e-05,
"loss": 0.1202,
"step": 917
},
{
"epoch": 1.75,
"grad_norm": 2.5064868927001953,
"learning_rate": 5.6774501412608525e-05,
"loss": 0.1275,
"step": 918
},
{
"epoch": 1.75,
"grad_norm": 1.9396264553070068,
"learning_rate": 5.6766163660971356e-05,
"loss": 0.1015,
"step": 919
},
{
"epoch": 1.75,
"grad_norm": 1.6867655515670776,
"learning_rate": 5.675781576076183e-05,
"loss": 0.0886,
"step": 920
},
{
"epoch": 1.75,
"grad_norm": 1.4513479471206665,
"learning_rate": 5.6749457715145125e-05,
"loss": 0.0815,
"step": 921
},
{
"epoch": 1.75,
"grad_norm": 1.6450719833374023,
"learning_rate": 5.6741089527290235e-05,
"loss": 0.0808,
"step": 922
},
{
"epoch": 1.75,
"grad_norm": 1.648695468902588,
"learning_rate": 5.6732711200370024e-05,
"loss": 0.1027,
"step": 923
},
{
"epoch": 1.76,
"grad_norm": 1.569627046585083,
"learning_rate": 5.6724322737561185e-05,
"loss": 0.1032,
"step": 924
},
{
"epoch": 1.76,
"grad_norm": 1.813715934753418,
"learning_rate": 5.671592414204427e-05,
"loss": 0.1025,
"step": 925
},
{
"epoch": 1.76,
"grad_norm": 2.0189218521118164,
"learning_rate": 5.670751541700366e-05,
"loss": 0.1101,
"step": 926
},
{
"epoch": 1.76,
"grad_norm": 2.329876184463501,
"learning_rate": 5.669909656562757e-05,
"loss": 0.1633,
"step": 927
},
{
"epoch": 1.76,
"grad_norm": 2.0899405479431152,
"learning_rate": 5.669066759110808e-05,
"loss": 0.1483,
"step": 928
},
{
"epoch": 1.77,
"grad_norm": 1.8417226076126099,
"learning_rate": 5.668222849664108e-05,
"loss": 0.0982,
"step": 929
},
{
"epoch": 1.77,
"grad_norm": 1.7111105918884277,
"learning_rate": 5.6673779285426314e-05,
"loss": 0.1266,
"step": 930
},
{
"epoch": 1.77,
"grad_norm": 1.8011417388916016,
"learning_rate": 5.6665319960667363e-05,
"loss": 0.1065,
"step": 931
},
{
"epoch": 1.77,
"grad_norm": 2.249350070953369,
"learning_rate": 5.665685052557164e-05,
"loss": 0.1283,
"step": 932
},
{
"epoch": 1.77,
"grad_norm": 2.1038548946380615,
"learning_rate": 5.6648370983350367e-05,
"loss": 0.0984,
"step": 933
},
{
"epoch": 1.78,
"grad_norm": 1.581061840057373,
"learning_rate": 5.663988133721864e-05,
"loss": 0.0778,
"step": 934
},
{
"epoch": 1.78,
"grad_norm": 1.6780366897583008,
"learning_rate": 5.663138159039537e-05,
"loss": 0.1177,
"step": 935
},
{
"epoch": 1.78,
"grad_norm": 1.5220457315444946,
"learning_rate": 5.6622871746103276e-05,
"loss": 0.0984,
"step": 936
},
{
"epoch": 1.78,
"grad_norm": 1.4900442361831665,
"learning_rate": 5.6614351807568934e-05,
"loss": 0.0865,
"step": 937
},
{
"epoch": 1.78,
"grad_norm": 1.5529632568359375,
"learning_rate": 5.660582177802273e-05,
"loss": 0.1041,
"step": 938
},
{
"epoch": 1.79,
"grad_norm": 1.3834776878356934,
"learning_rate": 5.659728166069889e-05,
"loss": 0.0859,
"step": 939
},
{
"epoch": 1.79,
"grad_norm": 1.615221619606018,
"learning_rate": 5.658873145883545e-05,
"loss": 0.1095,
"step": 940
},
{
"epoch": 1.79,
"grad_norm": 1.6735745668411255,
"learning_rate": 5.6580171175674286e-05,
"loss": 0.0929,
"step": 941
},
{
"epoch": 1.79,
"grad_norm": 1.6453355550765991,
"learning_rate": 5.657160081446108e-05,
"loss": 0.0952,
"step": 942
},
{
"epoch": 1.79,
"grad_norm": 1.6123896837234497,
"learning_rate": 5.6563020378445344e-05,
"loss": 0.09,
"step": 943
},
{
"epoch": 1.79,
"grad_norm": 1.8371021747589111,
"learning_rate": 5.6554429870880416e-05,
"loss": 0.1302,
"step": 944
},
{
"epoch": 1.8,
"grad_norm": 1.2052795886993408,
"learning_rate": 5.654582929502343e-05,
"loss": 0.0558,
"step": 945
},
{
"epoch": 1.8,
"grad_norm": 2.37248158454895,
"learning_rate": 5.653721865413536e-05,
"loss": 0.0861,
"step": 946
},
{
"epoch": 1.8,
"grad_norm": 2.0881247520446777,
"learning_rate": 5.6528597951481e-05,
"loss": 0.1416,
"step": 947
},
{
"epoch": 1.8,
"grad_norm": 1.6185557842254639,
"learning_rate": 5.6519967190328924e-05,
"loss": 0.095,
"step": 948
},
{
"epoch": 1.8,
"grad_norm": 1.6969891786575317,
"learning_rate": 5.651132637395157e-05,
"loss": 0.098,
"step": 949
},
{
"epoch": 1.81,
"grad_norm": 1.5117006301879883,
"learning_rate": 5.6502675505625146e-05,
"loss": 0.0943,
"step": 950
},
{
"epoch": 1.81,
"grad_norm": 1.8414154052734375,
"learning_rate": 5.6494014588629693e-05,
"loss": 0.1523,
"step": 951
},
{
"epoch": 1.81,
"grad_norm": 2.3739333152770996,
"learning_rate": 5.648534362624905e-05,
"loss": 0.1233,
"step": 952
},
{
"epoch": 1.81,
"grad_norm": 1.4182101488113403,
"learning_rate": 5.647666262177087e-05,
"loss": 0.1103,
"step": 953
},
{
"epoch": 1.81,
"grad_norm": 1.7330520153045654,
"learning_rate": 5.646797157848663e-05,
"loss": 0.0943,
"step": 954
},
{
"epoch": 1.82,
"grad_norm": 1.7026429176330566,
"learning_rate": 5.645927049969158e-05,
"loss": 0.1359,
"step": 955
},
{
"epoch": 1.82,
"grad_norm": 1.6526436805725098,
"learning_rate": 5.6450559388684804e-05,
"loss": 0.1221,
"step": 956
},
{
"epoch": 1.82,
"grad_norm": 1.3357737064361572,
"learning_rate": 5.644183824876917e-05,
"loss": 0.079,
"step": 957
},
{
"epoch": 1.82,
"grad_norm": 1.8663547039031982,
"learning_rate": 5.643310708325137e-05,
"loss": 0.1118,
"step": 958
},
{
"epoch": 1.82,
"grad_norm": 1.4101762771606445,
"learning_rate": 5.642436589544188e-05,
"loss": 0.0829,
"step": 959
},
{
"epoch": 1.83,
"grad_norm": 1.6070988178253174,
"learning_rate": 5.641561468865498e-05,
"loss": 0.109,
"step": 960
},
{
"epoch": 1.83,
"grad_norm": 1.4873356819152832,
"learning_rate": 5.640685346620874e-05,
"loss": 0.1031,
"step": 961
},
{
"epoch": 1.83,
"grad_norm": 1.7090345621109009,
"learning_rate": 5.639808223142504e-05,
"loss": 0.1272,
"step": 962
},
{
"epoch": 1.83,
"grad_norm": 1.8463568687438965,
"learning_rate": 5.638930098762957e-05,
"loss": 0.1055,
"step": 963
},
{
"epoch": 1.83,
"grad_norm": 1.4266470670700073,
"learning_rate": 5.638050973815177e-05,
"loss": 0.0898,
"step": 964
},
{
"epoch": 1.83,
"grad_norm": 1.2953797578811646,
"learning_rate": 5.637170848632494e-05,
"loss": 0.0809,
"step": 965
},
{
"epoch": 1.84,
"grad_norm": 1.5040665864944458,
"learning_rate": 5.636289723548609e-05,
"loss": 0.0954,
"step": 966
},
{
"epoch": 1.84,
"grad_norm": 1.552450180053711,
"learning_rate": 5.63540759889761e-05,
"loss": 0.109,
"step": 967
},
{
"epoch": 1.84,
"grad_norm": 1.5159934759140015,
"learning_rate": 5.6345244750139586e-05,
"loss": 0.0861,
"step": 968
},
{
"epoch": 1.84,
"grad_norm": 1.6420884132385254,
"learning_rate": 5.633640352232498e-05,
"loss": 0.1144,
"step": 969
},
{
"epoch": 1.84,
"grad_norm": 1.4441750049591064,
"learning_rate": 5.632755230888449e-05,
"loss": 0.0988,
"step": 970
},
{
"epoch": 1.85,
"grad_norm": 2.2879650592803955,
"learning_rate": 5.631869111317412e-05,
"loss": 0.1072,
"step": 971
},
{
"epoch": 1.85,
"grad_norm": 2.002598285675049,
"learning_rate": 5.630981993855365e-05,
"loss": 0.1219,
"step": 972
},
{
"epoch": 1.85,
"grad_norm": 1.6596964597702026,
"learning_rate": 5.6300938788386635e-05,
"loss": 0.1065,
"step": 973
},
{
"epoch": 1.85,
"grad_norm": 8.159488677978516,
"learning_rate": 5.629204766604044e-05,
"loss": 0.0876,
"step": 974
},
{
"epoch": 1.85,
"grad_norm": 1.467795729637146,
"learning_rate": 5.628314657488618e-05,
"loss": 0.0851,
"step": 975
},
{
"epoch": 1.86,
"grad_norm": 1.8075873851776123,
"learning_rate": 5.627423551829878e-05,
"loss": 0.1118,
"step": 976
},
{
"epoch": 1.86,
"grad_norm": 1.8339486122131348,
"learning_rate": 5.626531449965691e-05,
"loss": 0.0934,
"step": 977
},
{
"epoch": 1.86,
"grad_norm": 1.5868639945983887,
"learning_rate": 5.625638352234305e-05,
"loss": 0.0732,
"step": 978
},
{
"epoch": 1.86,
"grad_norm": 1.363390564918518,
"learning_rate": 5.624744258974343e-05,
"loss": 0.1015,
"step": 979
},
{
"epoch": 1.86,
"grad_norm": 1.8534549474716187,
"learning_rate": 5.623849170524807e-05,
"loss": 0.1415,
"step": 980
},
{
"epoch": 1.87,
"grad_norm": 1.9296644926071167,
"learning_rate": 5.6229530872250754e-05,
"loss": 0.1367,
"step": 981
},
{
"epoch": 1.87,
"grad_norm": 2.0634493827819824,
"learning_rate": 5.6220560094149046e-05,
"loss": 0.1474,
"step": 982
},
{
"epoch": 1.87,
"grad_norm": 1.7213340997695923,
"learning_rate": 5.621157937434427e-05,
"loss": 0.1179,
"step": 983
},
{
"epoch": 1.87,
"grad_norm": 1.7458151578903198,
"learning_rate": 5.620258871624155e-05,
"loss": 0.1036,
"step": 984
},
{
"epoch": 1.87,
"grad_norm": 1.6724697351455688,
"learning_rate": 5.619358812324973e-05,
"loss": 0.126,
"step": 985
},
{
"epoch": 1.87,
"grad_norm": 1.7286278009414673,
"learning_rate": 5.6184577598781456e-05,
"loss": 0.1143,
"step": 986
},
{
"epoch": 1.88,
"grad_norm": 1.5519840717315674,
"learning_rate": 5.617555714625312e-05,
"loss": 0.0937,
"step": 987
},
{
"epoch": 1.88,
"grad_norm": 1.7170854806900024,
"learning_rate": 5.6166526769084906e-05,
"loss": 0.1306,
"step": 988
},
{
"epoch": 1.88,
"grad_norm": 1.4816299676895142,
"learning_rate": 5.615748647070073e-05,
"loss": 0.0795,
"step": 989
},
{
"epoch": 1.88,
"grad_norm": 2.021548271179199,
"learning_rate": 5.6148436254528276e-05,
"loss": 0.161,
"step": 990
},
{
"epoch": 1.88,
"grad_norm": 1.5989229679107666,
"learning_rate": 5.613937612399901e-05,
"loss": 0.1263,
"step": 991
},
{
"epoch": 1.89,
"grad_norm": 1.8732894659042358,
"learning_rate": 5.6130306082548135e-05,
"loss": 0.1031,
"step": 992
},
{
"epoch": 1.89,
"grad_norm": 1.8975111246109009,
"learning_rate": 5.612122613361462e-05,
"loss": 0.1115,
"step": 993
},
{
"epoch": 1.89,
"grad_norm": 1.8112422227859497,
"learning_rate": 5.611213628064119e-05,
"loss": 0.1341,
"step": 994
},
{
"epoch": 1.89,
"grad_norm": 1.5903676748275757,
"learning_rate": 5.610303652707432e-05,
"loss": 0.0789,
"step": 995
},
{
"epoch": 1.89,
"grad_norm": 1.4763625860214233,
"learning_rate": 5.609392687636425e-05,
"loss": 0.0837,
"step": 996
},
{
"epoch": 1.9,
"grad_norm": 1.8155887126922607,
"learning_rate": 5.608480733196495e-05,
"loss": 0.134,
"step": 997
},
{
"epoch": 1.9,
"grad_norm": 2.0768091678619385,
"learning_rate": 5.607567789733419e-05,
"loss": 0.1089,
"step": 998
},
{
"epoch": 1.9,
"grad_norm": 1.7753323316574097,
"learning_rate": 5.606653857593342e-05,
"loss": 0.1128,
"step": 999
},
{
"epoch": 1.9,
"grad_norm": 2.3460538387298584,
"learning_rate": 5.605738937122789e-05,
"loss": 0.1222,
"step": 1000
},
{
"epoch": 1.9,
"eval_blimp_filtered_avg": 0.7005970149253732,
"eval_blimp_filtered_std": 0.005092553538367661,
"step": 1000
},
{
"epoch": 1.9,
"eval_blimp_supplement_avg": 0.75,
"eval_blimp_supplement_std": 0.018469307335754352,
"step": 1000
},
{
"epoch": 1.9,
"eval_vqa_filtered_avg": 0.33,
"eval_vqa_filtered_std": 0.047258156262526045,
"step": 1000
},
{
"epoch": 1.9,
"eval_winoground_filtered_avg": 0.5,
"eval_winoground_filtered_std": 0.050251890762960605,
"step": 1000
},
{
"epoch": 1.9,
"grad_norm": 1.8593547344207764,
"learning_rate": 5.6048230286686594e-05,
"loss": 0.1132,
"step": 1001
},
{
"epoch": 1.9,
"grad_norm": 2.0598199367523193,
"learning_rate": 5.6039061325782245e-05,
"loss": 0.1593,
"step": 1002
},
{
"epoch": 1.91,
"grad_norm": 1.9057772159576416,
"learning_rate": 5.602988249199132e-05,
"loss": 0.1385,
"step": 1003
},
{
"epoch": 1.91,
"grad_norm": 2.0247206687927246,
"learning_rate": 5.602069378879404e-05,
"loss": 0.1463,
"step": 1004
},
{
"epoch": 1.91,
"grad_norm": 1.6614131927490234,
"learning_rate": 5.6011495219674366e-05,
"loss": 0.092,
"step": 1005
},
{
"epoch": 1.91,
"grad_norm": 2.0429210662841797,
"learning_rate": 5.600228678812e-05,
"loss": 0.1044,
"step": 1006
},
{
"epoch": 1.91,
"grad_norm": 1.6447049379348755,
"learning_rate": 5.599306849762237e-05,
"loss": 0.0711,
"step": 1007
},
{
"epoch": 1.92,
"grad_norm": 2.083996534347534,
"learning_rate": 5.598384035167666e-05,
"loss": 0.123,
"step": 1008
},
{
"epoch": 1.92,
"grad_norm": 2.117016315460205,
"learning_rate": 5.5974602353781786e-05,
"loss": 0.1317,
"step": 1009
},
{
"epoch": 1.92,
"grad_norm": 1.2511467933654785,
"learning_rate": 5.596535450744039e-05,
"loss": 0.0887,
"step": 1010
},
{
"epoch": 1.92,
"grad_norm": 1.9032139778137207,
"learning_rate": 5.595609681615887e-05,
"loss": 0.1337,
"step": 1011
},
{
"epoch": 1.92,
"grad_norm": 1.7955689430236816,
"learning_rate": 5.594682928344732e-05,
"loss": 0.1237,
"step": 1012
},
{
"epoch": 1.93,
"grad_norm": 2.033277988433838,
"learning_rate": 5.593755191281961e-05,
"loss": 0.124,
"step": 1013
},
{
"epoch": 1.93,
"grad_norm": 1.5172165632247925,
"learning_rate": 5.592826470779331e-05,
"loss": 0.0944,
"step": 1014
},
{
"epoch": 1.93,
"grad_norm": 2.0833628177642822,
"learning_rate": 5.591896767188972e-05,
"loss": 0.1375,
"step": 1015
},
{
"epoch": 1.93,
"grad_norm": 1.6703460216522217,
"learning_rate": 5.590966080863388e-05,
"loss": 0.122,
"step": 1016
},
{
"epoch": 1.93,
"grad_norm": 1.7678513526916504,
"learning_rate": 5.590034412155456e-05,
"loss": 0.1218,
"step": 1017
},
{
"epoch": 1.94,
"grad_norm": 2.1431422233581543,
"learning_rate": 5.5891017614184224e-05,
"loss": 0.1393,
"step": 1018
},
{
"epoch": 1.94,
"grad_norm": 1.8071457147598267,
"learning_rate": 5.58816812900591e-05,
"loss": 0.157,
"step": 1019
},
{
"epoch": 1.94,
"grad_norm": 1.5262678861618042,
"learning_rate": 5.58723351527191e-05,
"loss": 0.1119,
"step": 1020
},
{
"epoch": 1.94,
"grad_norm": 1.7534315586090088,
"learning_rate": 5.5862979205707894e-05,
"loss": 0.1484,
"step": 1021
},
{
"epoch": 1.94,
"grad_norm": 1.7104535102844238,
"learning_rate": 5.585361345257285e-05,
"loss": 0.1038,
"step": 1022
},
{
"epoch": 1.94,
"grad_norm": 1.5410526990890503,
"learning_rate": 5.5844237896865025e-05,
"loss": 0.0978,
"step": 1023
},
{
"epoch": 1.95,
"grad_norm": 1.2848670482635498,
"learning_rate": 5.5834852542139276e-05,
"loss": 0.0786,
"step": 1024
},
{
"epoch": 1.95,
"grad_norm": 1.4537817239761353,
"learning_rate": 5.5825457391954086e-05,
"loss": 0.088,
"step": 1025
},
{
"epoch": 1.95,
"grad_norm": 1.6278260946273804,
"learning_rate": 5.5816052449871705e-05,
"loss": 0.1065,
"step": 1026
},
{
"epoch": 1.95,
"grad_norm": 0.9688180088996887,
"learning_rate": 5.580663771945808e-05,
"loss": 0.0668,
"step": 1027
},
{
"epoch": 1.95,
"grad_norm": 1.4491488933563232,
"learning_rate": 5.579721320428286e-05,
"loss": 0.1065,
"step": 1028
},
{
"epoch": 1.96,
"grad_norm": 1.5145771503448486,
"learning_rate": 5.578777890791944e-05,
"loss": 0.1078,
"step": 1029
},
{
"epoch": 1.96,
"grad_norm": 1.2848808765411377,
"learning_rate": 5.5778334833944866e-05,
"loss": 0.0884,
"step": 1030
},
{
"epoch": 1.96,
"grad_norm": 1.3872559070587158,
"learning_rate": 5.576888098593993e-05,
"loss": 0.0617,
"step": 1031
},
{
"epoch": 1.96,
"grad_norm": 1.5433127880096436,
"learning_rate": 5.575941736748914e-05,
"loss": 0.1409,
"step": 1032
},
{
"epoch": 1.96,
"grad_norm": 1.448178768157959,
"learning_rate": 5.574994398218067e-05,
"loss": 0.092,
"step": 1033
},
{
"epoch": 1.97,
"grad_norm": 1.8241462707519531,
"learning_rate": 5.574046083360643e-05,
"loss": 0.0862,
"step": 1034
},
{
"epoch": 1.97,
"grad_norm": 1.285172700881958,
"learning_rate": 5.5730967925362016e-05,
"loss": 0.0665,
"step": 1035
},
{
"epoch": 1.97,
"grad_norm": 1.4705709218978882,
"learning_rate": 5.572146526104673e-05,
"loss": 0.082,
"step": 1036
},
{
"epoch": 1.97,
"grad_norm": 1.857764720916748,
"learning_rate": 5.571195284426357e-05,
"loss": 0.1188,
"step": 1037
},
{
"epoch": 1.97,
"grad_norm": 1.928681492805481,
"learning_rate": 5.570243067861924e-05,
"loss": 0.1078,
"step": 1038
},
{
"epoch": 1.98,
"grad_norm": 1.860390305519104,
"learning_rate": 5.569289876772412e-05,
"loss": 0.0918,
"step": 1039
},
{
"epoch": 1.98,
"grad_norm": 1.1791576147079468,
"learning_rate": 5.568335711519231e-05,
"loss": 0.0691,
"step": 1040
},
{
"epoch": 1.98,
"grad_norm": 1.9752148389816284,
"learning_rate": 5.5673805724641595e-05,
"loss": 0.1437,
"step": 1041
},
{
"epoch": 1.98,
"grad_norm": 1.901798129081726,
"learning_rate": 5.566424459969344e-05,
"loss": 0.1568,
"step": 1042
},
{
"epoch": 1.98,
"grad_norm": 1.538602590560913,
"learning_rate": 5.565467374397301e-05,
"loss": 0.0966,
"step": 1043
},
{
"epoch": 1.98,
"grad_norm": 1.417629361152649,
"learning_rate": 5.564509316110917e-05,
"loss": 0.0811,
"step": 1044
},
{
"epoch": 1.99,
"grad_norm": 1.7578818798065186,
"learning_rate": 5.5635502854734454e-05,
"loss": 0.1179,
"step": 1045
},
{
"epoch": 1.99,
"grad_norm": 1.829776406288147,
"learning_rate": 5.56259028284851e-05,
"loss": 0.122,
"step": 1046
},
{
"epoch": 1.99,
"grad_norm": 1.6768912076950073,
"learning_rate": 5.561629308600101e-05,
"loss": 0.084,
"step": 1047
},
{
"epoch": 1.99,
"grad_norm": 1.126956582069397,
"learning_rate": 5.5606673630925796e-05,
"loss": 0.0617,
"step": 1048
},
{
"epoch": 1.99,
"grad_norm": 2.0299723148345947,
"learning_rate": 5.559704446690672e-05,
"loss": 0.1188,
"step": 1049
},
{
"epoch": 2.0,
"grad_norm": 1.7828278541564941,
"learning_rate": 5.558740559759477e-05,
"loss": 0.135,
"step": 1050
},
{
"epoch": 2.0,
"grad_norm": 1.5410562753677368,
"learning_rate": 5.557775702664457e-05,
"loss": 0.1106,
"step": 1051
},
{
"epoch": 2.0,
"grad_norm": 1.5325226783752441,
"learning_rate": 5.5568098757714463e-05,
"loss": 0.075,
"step": 1052
},
{
"epoch": 2.0,
"grad_norm": 2.8212289810180664,
"learning_rate": 5.5558430794466417e-05,
"loss": 0.051,
"step": 1053
},
{
"epoch": 2.0,
"grad_norm": 1.4168522357940674,
"learning_rate": 5.554875314056614e-05,
"loss": 0.0582,
"step": 1054
},
{
"epoch": 2.01,
"grad_norm": 1.3949567079544067,
"learning_rate": 5.553906579968294e-05,
"loss": 0.0858,
"step": 1055
},
{
"epoch": 2.01,
"grad_norm": 1.1691689491271973,
"learning_rate": 5.552936877548987e-05,
"loss": 0.0659,
"step": 1056
},
{
"epoch": 2.01,
"grad_norm": 1.4187984466552734,
"learning_rate": 5.551966207166361e-05,
"loss": 0.0739,
"step": 1057
},
{
"epoch": 2.01,
"grad_norm": 1.49600088596344,
"learning_rate": 5.550994569188452e-05,
"loss": 0.0831,
"step": 1058
},
{
"epoch": 2.01,
"grad_norm": 1.2913434505462646,
"learning_rate": 5.550021963983664e-05,
"loss": 0.0757,
"step": 1059
},
{
"epoch": 2.02,
"grad_norm": 1.5615456104278564,
"learning_rate": 5.549048391920767e-05,
"loss": 0.0546,
"step": 1060
},
{
"epoch": 2.02,
"grad_norm": 1.3788610696792603,
"learning_rate": 5.548073853368896e-05,
"loss": 0.0805,
"step": 1061
},
{
"epoch": 2.02,
"grad_norm": 0.9863640069961548,
"learning_rate": 5.547098348697555e-05,
"loss": 0.0516,
"step": 1062
},
{
"epoch": 2.02,
"grad_norm": 1.2528656721115112,
"learning_rate": 5.546121878276613e-05,
"loss": 0.0666,
"step": 1063
},
{
"epoch": 2.02,
"grad_norm": 1.469199776649475,
"learning_rate": 5.545144442476305e-05,
"loss": 0.0649,
"step": 1064
},
{
"epoch": 2.02,
"grad_norm": 1.1972655057907104,
"learning_rate": 5.544166041667234e-05,
"loss": 0.046,
"step": 1065
},
{
"epoch": 2.03,
"grad_norm": 1.3445485830307007,
"learning_rate": 5.543186676220364e-05,
"loss": 0.0501,
"step": 1066
},
{
"epoch": 2.03,
"grad_norm": 1.2494211196899414,
"learning_rate": 5.5422063465070316e-05,
"loss": 0.0404,
"step": 1067
},
{
"epoch": 2.03,
"grad_norm": 1.1826874017715454,
"learning_rate": 5.541225052898934e-05,
"loss": 0.0479,
"step": 1068
},
{
"epoch": 2.03,
"grad_norm": 1.4201607704162598,
"learning_rate": 5.5402427957681346e-05,
"loss": 0.0607,
"step": 1069
},
{
"epoch": 2.03,
"grad_norm": 1.115506649017334,
"learning_rate": 5.539259575487064e-05,
"loss": 0.0409,
"step": 1070
},
{
"epoch": 2.04,
"grad_norm": 1.4415713548660278,
"learning_rate": 5.538275392428515e-05,
"loss": 0.0638,
"step": 1071
},
{
"epoch": 2.04,
"grad_norm": 1.226027011871338,
"learning_rate": 5.5372902469656505e-05,
"loss": 0.0569,
"step": 1072
},
{
"epoch": 2.04,
"grad_norm": 1.6660706996917725,
"learning_rate": 5.536304139471992e-05,
"loss": 0.0605,
"step": 1073
},
{
"epoch": 2.04,
"grad_norm": 0.9817166328430176,
"learning_rate": 5.5353170703214303e-05,
"loss": 0.036,
"step": 1074
},
{
"epoch": 2.04,
"grad_norm": 1.8314621448516846,
"learning_rate": 5.5343290398882196e-05,
"loss": 0.0782,
"step": 1075
},
{
"epoch": 2.05,
"grad_norm": 1.8119754791259766,
"learning_rate": 5.533340048546977e-05,
"loss": 0.1169,
"step": 1076
},
{
"epoch": 2.05,
"grad_norm": 1.2777998447418213,
"learning_rate": 5.532350096672687e-05,
"loss": 0.0543,
"step": 1077
},
{
"epoch": 2.05,
"grad_norm": 1.0604852437973022,
"learning_rate": 5.531359184640695e-05,
"loss": 0.0386,
"step": 1078
},
{
"epoch": 2.05,
"grad_norm": 1.6358789205551147,
"learning_rate": 5.530367312826713e-05,
"loss": 0.0928,
"step": 1079
},
{
"epoch": 2.05,
"grad_norm": 1.40937077999115,
"learning_rate": 5.529374481606815e-05,
"loss": 0.0733,
"step": 1080
},
{
"epoch": 2.06,
"grad_norm": 1.0809097290039062,
"learning_rate": 5.528380691357441e-05,
"loss": 0.0579,
"step": 1081
},
{
"epoch": 2.06,
"grad_norm": 1.3440784215927124,
"learning_rate": 5.527385942455392e-05,
"loss": 0.0439,
"step": 1082
},
{
"epoch": 2.06,
"grad_norm": 1.1631584167480469,
"learning_rate": 5.526390235277835e-05,
"loss": 0.0455,
"step": 1083
},
{
"epoch": 2.06,
"grad_norm": 1.5910366773605347,
"learning_rate": 5.525393570202298e-05,
"loss": 0.0715,
"step": 1084
},
{
"epoch": 2.06,
"grad_norm": 1.5799205303192139,
"learning_rate": 5.5243959476066746e-05,
"loss": 0.0624,
"step": 1085
},
{
"epoch": 2.06,
"grad_norm": 1.4309570789337158,
"learning_rate": 5.523397367869218e-05,
"loss": 0.0561,
"step": 1086
},
{
"epoch": 2.07,
"grad_norm": 1.3139511346817017,
"learning_rate": 5.522397831368549e-05,
"loss": 0.0589,
"step": 1087
},
{
"epoch": 2.07,
"grad_norm": 1.5517388582229614,
"learning_rate": 5.5213973384836466e-05,
"loss": 0.0636,
"step": 1088
},
{
"epoch": 2.07,
"grad_norm": 1.6563770771026611,
"learning_rate": 5.520395889593856e-05,
"loss": 0.0542,
"step": 1089
},
{
"epoch": 2.07,
"grad_norm": 1.204545259475708,
"learning_rate": 5.519393485078883e-05,
"loss": 0.0606,
"step": 1090
},
{
"epoch": 2.07,
"grad_norm": 0.9802208542823792,
"learning_rate": 5.5183901253187955e-05,
"loss": 0.0401,
"step": 1091
},
{
"epoch": 2.08,
"grad_norm": 1.8141857385635376,
"learning_rate": 5.517385810694024e-05,
"loss": 0.0583,
"step": 1092
},
{
"epoch": 2.08,
"grad_norm": 1.3271673917770386,
"learning_rate": 5.516380541585362e-05,
"loss": 0.056,
"step": 1093
},
{
"epoch": 2.08,
"grad_norm": 1.597602367401123,
"learning_rate": 5.515374318373964e-05,
"loss": 0.0595,
"step": 1094
},
{
"epoch": 2.08,
"grad_norm": 0.9415888786315918,
"learning_rate": 5.514367141441345e-05,
"loss": 0.043,
"step": 1095
},
{
"epoch": 2.08,
"grad_norm": 1.1622306108474731,
"learning_rate": 5.513359011169385e-05,
"loss": 0.0365,
"step": 1096
},
{
"epoch": 2.09,
"grad_norm": 1.570786476135254,
"learning_rate": 5.512349927940322e-05,
"loss": 0.0697,
"step": 1097
},
{
"epoch": 2.09,
"grad_norm": 1.571063756942749,
"learning_rate": 5.511339892136757e-05,
"loss": 0.0737,
"step": 1098
},
{
"epoch": 2.09,
"grad_norm": 1.3366670608520508,
"learning_rate": 5.5103289041416524e-05,
"loss": 0.0683,
"step": 1099
},
{
"epoch": 2.09,
"grad_norm": 1.2015153169631958,
"learning_rate": 5.509316964338331e-05,
"loss": 0.0549,
"step": 1100
},
{
"epoch": 2.09,
"eval_blimp_filtered_avg": 0.7064179104477611,
"eval_blimp_filtered_std": 0.005036281728721821,
"step": 1100
},
{
"epoch": 2.09,
"eval_blimp_supplement_avg": 0.7586206896551724,
"eval_blimp_supplement_std": 0.017777571987554128,
"step": 1100
},
{
"epoch": 2.09,
"eval_vqa_filtered_avg": 0.39,
"eval_vqa_filtered_std": 0.04902071300001974,
"step": 1100
},
{
"epoch": 2.09,
"eval_winoground_filtered_avg": 0.5,
"eval_winoground_filtered_std": 0.050251890762960605,
"step": 1100
},
{
"epoch": 2.09,
"grad_norm": 1.3971072435379028,
"learning_rate": 5.508304073110476e-05,
"loss": 0.0595,
"step": 1101
},
{
"epoch": 2.1,
"grad_norm": 1.5023480653762817,
"learning_rate": 5.507290230842133e-05,
"loss": 0.0742,
"step": 1102
},
{
"epoch": 2.1,
"grad_norm": 1.6642764806747437,
"learning_rate": 5.506275437917706e-05,
"loss": 0.0605,
"step": 1103
},
{
"epoch": 2.1,
"grad_norm": 1.6743719577789307,
"learning_rate": 5.505259694721962e-05,
"loss": 0.0727,
"step": 1104
},
{
"epoch": 2.1,
"grad_norm": 1.1604280471801758,
"learning_rate": 5.504243001640026e-05,
"loss": 0.0359,
"step": 1105
},
{
"epoch": 2.1,
"grad_norm": 1.4510600566864014,
"learning_rate": 5.5032253590573844e-05,
"loss": 0.0569,
"step": 1106
},
{
"epoch": 2.1,
"grad_norm": 1.2207281589508057,
"learning_rate": 5.5022067673598824e-05,
"loss": 0.0406,
"step": 1107
},
{
"epoch": 2.11,
"grad_norm": 1.5505845546722412,
"learning_rate": 5.501187226933727e-05,
"loss": 0.0459,
"step": 1108
},
{
"epoch": 2.11,
"grad_norm": 1.176196575164795,
"learning_rate": 5.5001667381654833e-05,
"loss": 0.0466,
"step": 1109
},
{
"epoch": 2.11,
"grad_norm": 1.110958218574524,
"learning_rate": 5.499145301442078e-05,
"loss": 0.0359,
"step": 1110
},
{
"epoch": 2.11,
"grad_norm": 1.5733847618103027,
"learning_rate": 5.498122917150793e-05,
"loss": 0.0452,
"step": 1111
},
{
"epoch": 2.11,
"grad_norm": 1.2085074186325073,
"learning_rate": 5.4970995856792744e-05,
"loss": 0.0443,
"step": 1112
},
{
"epoch": 2.12,
"grad_norm": 1.4007294178009033,
"learning_rate": 5.4960753074155236e-05,
"loss": 0.0443,
"step": 1113
},
{
"epoch": 2.12,
"grad_norm": 1.6465131044387817,
"learning_rate": 5.4950500827479044e-05,
"loss": 0.0594,
"step": 1114
},
{
"epoch": 2.12,
"grad_norm": 1.4671810865402222,
"learning_rate": 5.494023912065137e-05,
"loss": 0.0542,
"step": 1115
},
{
"epoch": 2.12,
"grad_norm": 1.4673278331756592,
"learning_rate": 5.4929967957563016e-05,
"loss": 0.0484,
"step": 1116
},
{
"epoch": 2.12,
"grad_norm": 1.1502727270126343,
"learning_rate": 5.491968734210835e-05,
"loss": 0.0561,
"step": 1117
},
{
"epoch": 2.13,
"grad_norm": 1.1386257410049438,
"learning_rate": 5.4909397278185344e-05,
"loss": 0.0368,
"step": 1118
},
{
"epoch": 2.13,
"grad_norm": 1.944210410118103,
"learning_rate": 5.4899097769695535e-05,
"loss": 0.0751,
"step": 1119
},
{
"epoch": 2.13,
"grad_norm": 1.3016364574432373,
"learning_rate": 5.4888788820544075e-05,
"loss": 0.0802,
"step": 1120
},
{
"epoch": 2.13,
"grad_norm": 1.6970877647399902,
"learning_rate": 5.487847043463965e-05,
"loss": 0.0719,
"step": 1121
},
{
"epoch": 2.13,
"grad_norm": 1.4683481454849243,
"learning_rate": 5.4868142615894555e-05,
"loss": 0.0572,
"step": 1122
},
{
"epoch": 2.13,
"grad_norm": 1.2096418142318726,
"learning_rate": 5.485780536822466e-05,
"loss": 0.0671,
"step": 1123
},
{
"epoch": 2.14,
"grad_norm": 1.2248331308364868,
"learning_rate": 5.4847458695549376e-05,
"loss": 0.0415,
"step": 1124
},
{
"epoch": 2.14,
"grad_norm": 1.493685007095337,
"learning_rate": 5.4837102601791745e-05,
"loss": 0.0366,
"step": 1125
},
{
"epoch": 2.14,
"grad_norm": 1.2747687101364136,
"learning_rate": 5.482673709087833e-05,
"loss": 0.0442,
"step": 1126
},
{
"epoch": 2.14,
"grad_norm": 1.2811286449432373,
"learning_rate": 5.4816362166739305e-05,
"loss": 0.0563,
"step": 1127
},
{
"epoch": 2.14,
"grad_norm": 1.5202643871307373,
"learning_rate": 5.480597783330837e-05,
"loss": 0.0393,
"step": 1128
},
{
"epoch": 2.15,
"grad_norm": 1.4967796802520752,
"learning_rate": 5.479558409452282e-05,
"loss": 0.0611,
"step": 1129
},
{
"epoch": 2.15,
"grad_norm": 2.445615768432617,
"learning_rate": 5.4785180954323525e-05,
"loss": 0.0814,
"step": 1130
},
{
"epoch": 2.15,
"grad_norm": 1.4135996103286743,
"learning_rate": 5.47747684166549e-05,
"loss": 0.0596,
"step": 1131
},
{
"epoch": 2.15,
"grad_norm": 1.5554150342941284,
"learning_rate": 5.476434648546492e-05,
"loss": 0.0632,
"step": 1132
},
{
"epoch": 2.15,
"grad_norm": 1.3189866542816162,
"learning_rate": 5.4753915164705135e-05,
"loss": 0.0531,
"step": 1133
},
{
"epoch": 2.16,
"grad_norm": 1.2746988534927368,
"learning_rate": 5.474347445833066e-05,
"loss": 0.055,
"step": 1134
},
{
"epoch": 2.16,
"grad_norm": 1.7298076152801514,
"learning_rate": 5.4733024370300156e-05,
"loss": 0.0533,
"step": 1135
},
{
"epoch": 2.16,
"grad_norm": 1.267601728439331,
"learning_rate": 5.472256490457583e-05,
"loss": 0.0488,
"step": 1136
},
{
"epoch": 2.16,
"grad_norm": 1.3614825010299683,
"learning_rate": 5.471209606512348e-05,
"loss": 0.0491,
"step": 1137
},
{
"epoch": 2.16,
"grad_norm": 0.9298860430717468,
"learning_rate": 5.4701617855912426e-05,
"loss": 0.0502,
"step": 1138
},
{
"epoch": 2.17,
"grad_norm": 1.3963333368301392,
"learning_rate": 5.4691130280915554e-05,
"loss": 0.0603,
"step": 1139
},
{
"epoch": 2.17,
"grad_norm": 0.9974714517593384,
"learning_rate": 5.4680633344109306e-05,
"loss": 0.0309,
"step": 1140
},
{
"epoch": 2.17,
"grad_norm": 1.4522044658660889,
"learning_rate": 5.4670127049473654e-05,
"loss": 0.0526,
"step": 1141
},
{
"epoch": 2.17,
"grad_norm": 1.6446279287338257,
"learning_rate": 5.465961140099214e-05,
"loss": 0.0541,
"step": 1142
},
{
"epoch": 2.17,
"grad_norm": 1.9810283184051514,
"learning_rate": 5.464908640265184e-05,
"loss": 0.0554,
"step": 1143
},
{
"epoch": 2.17,
"grad_norm": 1.5595523118972778,
"learning_rate": 5.463855205844338e-05,
"loss": 0.0679,
"step": 1144
},
{
"epoch": 2.18,
"grad_norm": 1.419427752494812,
"learning_rate": 5.462800837236092e-05,
"loss": 0.0881,
"step": 1145
},
{
"epoch": 2.18,
"grad_norm": 1.2112529277801514,
"learning_rate": 5.4617455348402195e-05,
"loss": 0.0572,
"step": 1146
},
{
"epoch": 2.18,
"grad_norm": 1.2218973636627197,
"learning_rate": 5.460689299056842e-05,
"loss": 0.083,
"step": 1147
},
{
"epoch": 2.18,
"grad_norm": 1.1981514692306519,
"learning_rate": 5.459632130286441e-05,
"loss": 0.0477,
"step": 1148
},
{
"epoch": 2.18,
"grad_norm": 1.1551488637924194,
"learning_rate": 5.458574028929848e-05,
"loss": 0.0508,
"step": 1149
},
{
"epoch": 2.19,
"grad_norm": 1.3468579053878784,
"learning_rate": 5.45751499538825e-05,
"loss": 0.067,
"step": 1150
},
{
"epoch": 2.19,
"grad_norm": 1.2943669557571411,
"learning_rate": 5.4564550300631874e-05,
"loss": 0.0606,
"step": 1151
},
{
"epoch": 2.19,
"grad_norm": 1.7822179794311523,
"learning_rate": 5.455394133356552e-05,
"loss": 0.0526,
"step": 1152
},
{
"epoch": 2.19,
"grad_norm": 1.206272840499878,
"learning_rate": 5.454332305670589e-05,
"loss": 0.0633,
"step": 1153
},
{
"epoch": 2.19,
"grad_norm": 1.1472529172897339,
"learning_rate": 5.4532695474078994e-05,
"loss": 0.0616,
"step": 1154
},
{
"epoch": 2.2,
"grad_norm": 1.2524665594100952,
"learning_rate": 5.452205858971435e-05,
"loss": 0.0543,
"step": 1155
},
{
"epoch": 2.2,
"grad_norm": 1.115168571472168,
"learning_rate": 5.451141240764501e-05,
"loss": 0.047,
"step": 1156
},
{
"epoch": 2.2,
"grad_norm": 1.1044842004776,
"learning_rate": 5.4500756931907524e-05,
"loss": 0.056,
"step": 1157
},
{
"epoch": 2.2,
"grad_norm": 1.4551676511764526,
"learning_rate": 5.4490092166542005e-05,
"loss": 0.0538,
"step": 1158
},
{
"epoch": 2.2,
"grad_norm": 1.6292243003845215,
"learning_rate": 5.447941811559207e-05,
"loss": 0.0583,
"step": 1159
},
{
"epoch": 2.21,
"grad_norm": 0.8083898425102234,
"learning_rate": 5.446873478310485e-05,
"loss": 0.0299,
"step": 1160
},
{
"epoch": 2.21,
"grad_norm": 1.2083748579025269,
"learning_rate": 5.4458042173131015e-05,
"loss": 0.0629,
"step": 1161
},
{
"epoch": 2.21,
"grad_norm": 0.8567453622817993,
"learning_rate": 5.4447340289724725e-05,
"loss": 0.035,
"step": 1162
},
{
"epoch": 2.21,
"grad_norm": 1.2181017398834229,
"learning_rate": 5.443662913694369e-05,
"loss": 0.0398,
"step": 1163
},
{
"epoch": 2.21,
"grad_norm": 1.5463600158691406,
"learning_rate": 5.44259087188491e-05,
"loss": 0.0475,
"step": 1164
},
{
"epoch": 2.21,
"grad_norm": 0.9666810035705566,
"learning_rate": 5.441517903950568e-05,
"loss": 0.0418,
"step": 1165
},
{
"epoch": 2.22,
"grad_norm": 1.4218964576721191,
"learning_rate": 5.440444010298166e-05,
"loss": 0.0862,
"step": 1166
},
{
"epoch": 2.22,
"grad_norm": 1.3489267826080322,
"learning_rate": 5.4393691913348786e-05,
"loss": 0.0497,
"step": 1167
},
{
"epoch": 2.22,
"grad_norm": 1.229305386543274,
"learning_rate": 5.43829344746823e-05,
"loss": 0.048,
"step": 1168
},
{
"epoch": 2.22,
"grad_norm": 1.451535940170288,
"learning_rate": 5.437216779106096e-05,
"loss": 0.0512,
"step": 1169
},
{
"epoch": 2.22,
"grad_norm": 1.3999005556106567,
"learning_rate": 5.436139186656703e-05,
"loss": 0.0659,
"step": 1170
},
{
"epoch": 2.23,
"grad_norm": 1.137280821800232,
"learning_rate": 5.435060670528628e-05,
"loss": 0.0451,
"step": 1171
},
{
"epoch": 2.23,
"grad_norm": 1.6075098514556885,
"learning_rate": 5.4339812311307956e-05,
"loss": 0.0553,
"step": 1172
},
{
"epoch": 2.23,
"grad_norm": 1.3414723873138428,
"learning_rate": 5.432900868872484e-05,
"loss": 0.0719,
"step": 1173
},
{
"epoch": 2.23,
"grad_norm": 1.5067349672317505,
"learning_rate": 5.4318195841633204e-05,
"loss": 0.0532,
"step": 1174
},
{
"epoch": 2.23,
"grad_norm": 1.4898079633712769,
"learning_rate": 5.43073737741328e-05,
"loss": 0.0719,
"step": 1175
},
{
"epoch": 2.24,
"grad_norm": 1.2872443199157715,
"learning_rate": 5.42965424903269e-05,
"loss": 0.0369,
"step": 1176
},
{
"epoch": 2.24,
"grad_norm": 2.6055374145507812,
"learning_rate": 5.428570199432226e-05,
"loss": 0.0706,
"step": 1177
},
{
"epoch": 2.24,
"grad_norm": 1.3079980611801147,
"learning_rate": 5.4274852290229104e-05,
"loss": 0.0452,
"step": 1178
},
{
"epoch": 2.24,
"grad_norm": 1.2309643030166626,
"learning_rate": 5.4263993382161196e-05,
"loss": 0.054,
"step": 1179
},
{
"epoch": 2.24,
"grad_norm": 1.1787432432174683,
"learning_rate": 5.425312527423575e-05,
"loss": 0.0453,
"step": 1180
},
{
"epoch": 2.25,
"grad_norm": 1.3654181957244873,
"learning_rate": 5.42422479705735e-05,
"loss": 0.0873,
"step": 1181
},
{
"epoch": 2.25,
"grad_norm": 1.7388787269592285,
"learning_rate": 5.4231361475298624e-05,
"loss": 0.0572,
"step": 1182
},
{
"epoch": 2.25,
"grad_norm": 1.2375620603561401,
"learning_rate": 5.422046579253883e-05,
"loss": 0.0439,
"step": 1183
},
{
"epoch": 2.25,
"grad_norm": 1.162243127822876,
"learning_rate": 5.4209560926425285e-05,
"loss": 0.0629,
"step": 1184
},
{
"epoch": 2.25,
"grad_norm": 1.3965047597885132,
"learning_rate": 5.419864688109265e-05,
"loss": 0.0597,
"step": 1185
},
{
"epoch": 2.25,
"grad_norm": 1.1871970891952515,
"learning_rate": 5.4187723660679044e-05,
"loss": 0.0668,
"step": 1186
},
{
"epoch": 2.26,
"grad_norm": 1.621813178062439,
"learning_rate": 5.417679126932608e-05,
"loss": 0.0846,
"step": 1187
},
{
"epoch": 2.26,
"grad_norm": 1.172818899154663,
"learning_rate": 5.416584971117887e-05,
"loss": 0.0411,
"step": 1188
},
{
"epoch": 2.26,
"grad_norm": 1.1815779209136963,
"learning_rate": 5.4154898990385955e-05,
"loss": 0.0576,
"step": 1189
},
{
"epoch": 2.26,
"grad_norm": 1.3724629878997803,
"learning_rate": 5.414393911109939e-05,
"loss": 0.044,
"step": 1190
},
{
"epoch": 2.26,
"grad_norm": 1.2136143445968628,
"learning_rate": 5.4132970077474685e-05,
"loss": 0.0583,
"step": 1191
},
{
"epoch": 2.27,
"grad_norm": 1.6299148797988892,
"learning_rate": 5.412199189367082e-05,
"loss": 0.0869,
"step": 1192
},
{
"epoch": 2.27,
"grad_norm": 2.1303787231445312,
"learning_rate": 5.411100456385025e-05,
"loss": 0.0907,
"step": 1193
},
{
"epoch": 2.27,
"grad_norm": 1.1059167385101318,
"learning_rate": 5.41000080921789e-05,
"loss": 0.0476,
"step": 1194
},
{
"epoch": 2.27,
"grad_norm": 1.7504640817642212,
"learning_rate": 5.4089002482826136e-05,
"loss": 0.0555,
"step": 1195
},
{
"epoch": 2.27,
"grad_norm": 1.4289520978927612,
"learning_rate": 5.407798773996484e-05,
"loss": 0.0604,
"step": 1196
},
{
"epoch": 2.28,
"grad_norm": 1.1072924137115479,
"learning_rate": 5.4066963867771303e-05,
"loss": 0.0395,
"step": 1197
},
{
"epoch": 2.28,
"grad_norm": 0.9000833034515381,
"learning_rate": 5.405593087042532e-05,
"loss": 0.0367,
"step": 1198
},
{
"epoch": 2.28,
"grad_norm": 1.3626693487167358,
"learning_rate": 5.4044888752110114e-05,
"loss": 0.0684,
"step": 1199
},
{
"epoch": 2.28,
"grad_norm": 1.3157615661621094,
"learning_rate": 5.403383751701237e-05,
"loss": 0.0529,
"step": 1200
},
{
"epoch": 2.28,
"eval_blimp_filtered_avg": 0.7140298507462687,
"eval_blimp_filtered_std": 0.005036661487424995,
"step": 1200
},
{
"epoch": 2.28,
"eval_blimp_supplement_avg": 0.7650862068965517,
"eval_blimp_supplement_std": 0.018100899184111896,
"step": 1200
},
{
"epoch": 2.28,
"eval_vqa_filtered_avg": 0.37,
"eval_vqa_filtered_std": 0.048523658709391,
"step": 1200
},
{
"epoch": 2.28,
"eval_winoground_filtered_avg": 0.5,
"eval_winoground_filtered_std": 0.050251890762960605,
"step": 1200
},
{
"epoch": 2.28,
"grad_norm": 1.6770339012145996,
"learning_rate": 5.4022777169322256e-05,
"loss": 0.0798,
"step": 1201
},
{
"epoch": 2.29,
"grad_norm": 0.9976529479026794,
"learning_rate": 5.4011707713233376e-05,
"loss": 0.0304,
"step": 1202
},
{
"epoch": 2.29,
"grad_norm": 1.3133912086486816,
"learning_rate": 5.400062915294278e-05,
"loss": 0.0691,
"step": 1203
},
{
"epoch": 2.29,
"grad_norm": 1.0383552312850952,
"learning_rate": 5.398954149265099e-05,
"loss": 0.0438,
"step": 1204
},
{
"epoch": 2.29,
"grad_norm": 1.1348787546157837,
"learning_rate": 5.3978444736561946e-05,
"loss": 0.0369,
"step": 1205
},
{
"epoch": 2.29,
"grad_norm": 1.4327248334884644,
"learning_rate": 5.3967338888883095e-05,
"loss": 0.0525,
"step": 1206
},
{
"epoch": 2.29,
"grad_norm": 1.5268528461456299,
"learning_rate": 5.395622395382526e-05,
"loss": 0.0599,
"step": 1207
},
{
"epoch": 2.3,
"grad_norm": 1.2855076789855957,
"learning_rate": 5.394509993560276e-05,
"loss": 0.0487,
"step": 1208
},
{
"epoch": 2.3,
"grad_norm": 1.0276800394058228,
"learning_rate": 5.393396683843334e-05,
"loss": 0.0327,
"step": 1209
},
{
"epoch": 2.3,
"grad_norm": 1.1450759172439575,
"learning_rate": 5.392282466653819e-05,
"loss": 0.0413,
"step": 1210
},
{
"epoch": 2.3,
"grad_norm": 1.2271578311920166,
"learning_rate": 5.3911673424141943e-05,
"loss": 0.0478,
"step": 1211
},
{
"epoch": 2.3,
"grad_norm": 1.80759596824646,
"learning_rate": 5.390051311547265e-05,
"loss": 0.0761,
"step": 1212
},
{
"epoch": 2.31,
"grad_norm": 1.4193974733352661,
"learning_rate": 5.388934374476184e-05,
"loss": 0.0427,
"step": 1213
},
{
"epoch": 2.31,
"grad_norm": 1.6652902364730835,
"learning_rate": 5.387816531624445e-05,
"loss": 0.0878,
"step": 1214
},
{
"epoch": 2.31,
"grad_norm": 1.171245813369751,
"learning_rate": 5.3866977834158854e-05,
"loss": 0.0703,
"step": 1215
},
{
"epoch": 2.31,
"grad_norm": 1.2354869842529297,
"learning_rate": 5.385578130274685e-05,
"loss": 0.0595,
"step": 1216
},
{
"epoch": 2.31,
"grad_norm": 1.0901631116867065,
"learning_rate": 5.384457572625371e-05,
"loss": 0.052,
"step": 1217
},
{
"epoch": 2.32,
"grad_norm": 0.9187933206558228,
"learning_rate": 5.383336110892806e-05,
"loss": 0.0417,
"step": 1218
},
{
"epoch": 2.32,
"grad_norm": 0.8184903860092163,
"learning_rate": 5.382213745502204e-05,
"loss": 0.0326,
"step": 1219
},
{
"epoch": 2.32,
"grad_norm": 1.6810932159423828,
"learning_rate": 5.381090476879115e-05,
"loss": 0.0466,
"step": 1220
},
{
"epoch": 2.32,
"grad_norm": 1.2054543495178223,
"learning_rate": 5.379966305449435e-05,
"loss": 0.0635,
"step": 1221
},
{
"epoch": 2.32,
"grad_norm": 1.1691229343414307,
"learning_rate": 5.378841231639401e-05,
"loss": 0.0442,
"step": 1222
},
{
"epoch": 2.33,
"grad_norm": 1.514968752861023,
"learning_rate": 5.3777152558755915e-05,
"loss": 0.0628,
"step": 1223
},
{
"epoch": 2.33,
"grad_norm": 1.187780737876892,
"learning_rate": 5.3765883785849284e-05,
"loss": 0.0669,
"step": 1224
},
{
"epoch": 2.33,
"grad_norm": 2.8802411556243896,
"learning_rate": 5.375460600194675e-05,
"loss": 0.0798,
"step": 1225
},
{
"epoch": 2.33,
"grad_norm": 1.1711046695709229,
"learning_rate": 5.3743319211324376e-05,
"loss": 0.0579,
"step": 1226
},
{
"epoch": 2.33,
"grad_norm": 6.481640338897705,
"learning_rate": 5.373202341826159e-05,
"loss": 0.0872,
"step": 1227
},
{
"epoch": 2.33,
"grad_norm": 1.3612349033355713,
"learning_rate": 5.3720718627041304e-05,
"loss": 0.0602,
"step": 1228
},
{
"epoch": 2.34,
"grad_norm": 1.073043704032898,
"learning_rate": 5.370940484194978e-05,
"loss": 0.0574,
"step": 1229
},
{
"epoch": 2.34,
"grad_norm": 1.5003401041030884,
"learning_rate": 5.369808206727674e-05,
"loss": 0.0696,
"step": 1230
},
{
"epoch": 2.34,
"grad_norm": 1.1137248277664185,
"learning_rate": 5.3686750307315275e-05,
"loss": 0.0545,
"step": 1231
},
{
"epoch": 2.34,
"grad_norm": 1.4048364162445068,
"learning_rate": 5.36754095663619e-05,
"loss": 0.0813,
"step": 1232
},
{
"epoch": 2.34,
"grad_norm": 1.118627905845642,
"learning_rate": 5.366405984871654e-05,
"loss": 0.0482,
"step": 1233
},
{
"epoch": 2.35,
"grad_norm": 1.3862615823745728,
"learning_rate": 5.365270115868249e-05,
"loss": 0.0446,
"step": 1234
},
{
"epoch": 2.35,
"grad_norm": 1.369816541671753,
"learning_rate": 5.364133350056652e-05,
"loss": 0.0552,
"step": 1235
},
{
"epoch": 2.35,
"grad_norm": 1.7138818502426147,
"learning_rate": 5.362995687867873e-05,
"loss": 0.0738,
"step": 1236
},
{
"epoch": 2.35,
"grad_norm": 1.8603782653808594,
"learning_rate": 5.3618571297332636e-05,
"loss": 0.0627,
"step": 1237
},
{
"epoch": 2.35,
"grad_norm": 1.095945119857788,
"learning_rate": 5.360717676084518e-05,
"loss": 0.0539,
"step": 1238
},
{
"epoch": 2.36,
"grad_norm": 1.5818086862564087,
"learning_rate": 5.359577327353665e-05,
"loss": 0.0645,
"step": 1239
},
{
"epoch": 2.36,
"grad_norm": 1.6076397895812988,
"learning_rate": 5.3584360839730777e-05,
"loss": 0.0705,
"step": 1240
},
{
"epoch": 2.36,
"grad_norm": 1.6995656490325928,
"learning_rate": 5.357293946375466e-05,
"loss": 0.0771,
"step": 1241
},
{
"epoch": 2.36,
"grad_norm": 1.2431119680404663,
"learning_rate": 5.356150914993879e-05,
"loss": 0.0347,
"step": 1242
},
{
"epoch": 2.36,
"grad_norm": 1.5639264583587646,
"learning_rate": 5.3550069902617034e-05,
"loss": 0.0744,
"step": 1243
},
{
"epoch": 2.37,
"grad_norm": 1.7416237592697144,
"learning_rate": 5.3538621726126685e-05,
"loss": 0.0774,
"step": 1244
},
{
"epoch": 2.37,
"grad_norm": 1.8791722059249878,
"learning_rate": 5.352716462480837e-05,
"loss": 0.0708,
"step": 1245
},
{
"epoch": 2.37,
"grad_norm": 1.911062479019165,
"learning_rate": 5.351569860300614e-05,
"loss": 0.0871,
"step": 1246
},
{
"epoch": 2.37,
"grad_norm": 1.3535583019256592,
"learning_rate": 5.3504223665067425e-05,
"loss": 0.0527,
"step": 1247
},
{
"epoch": 2.37,
"grad_norm": 1.8695369958877563,
"learning_rate": 5.3492739815343016e-05,
"loss": 0.0831,
"step": 1248
},
{
"epoch": 2.37,
"grad_norm": 2.1056954860687256,
"learning_rate": 5.348124705818708e-05,
"loss": 0.0539,
"step": 1249
},
{
"epoch": 2.38,
"grad_norm": 1.4229819774627686,
"learning_rate": 5.346974539795719e-05,
"loss": 0.046,
"step": 1250
},
{
"epoch": 2.38,
"grad_norm": 1.5071381330490112,
"learning_rate": 5.345823483901428e-05,
"loss": 0.0531,
"step": 1251
},
{
"epoch": 2.38,
"grad_norm": 1.5696215629577637,
"learning_rate": 5.3446715385722635e-05,
"loss": 0.0668,
"step": 1252
},
{
"epoch": 2.38,
"grad_norm": 1.4545366764068604,
"learning_rate": 5.343518704244997e-05,
"loss": 0.0697,
"step": 1253
},
{
"epoch": 2.38,
"grad_norm": 2.4107248783111572,
"learning_rate": 5.3423649813567294e-05,
"loss": 0.0828,
"step": 1254
},
{
"epoch": 2.39,
"grad_norm": 1.6417913436889648,
"learning_rate": 5.341210370344905e-05,
"loss": 0.0649,
"step": 1255
},
{
"epoch": 2.39,
"grad_norm": 1.5612318515777588,
"learning_rate": 5.3400548716473026e-05,
"loss": 0.0556,
"step": 1256
},
{
"epoch": 2.39,
"grad_norm": 1.9091428518295288,
"learning_rate": 5.338898485702036e-05,
"loss": 0.0954,
"step": 1257
},
{
"epoch": 2.39,
"grad_norm": 1.5666626691818237,
"learning_rate": 5.337741212947558e-05,
"loss": 0.0891,
"step": 1258
},
{
"epoch": 2.39,
"grad_norm": 1.3324806690216064,
"learning_rate": 5.336583053822655e-05,
"loss": 0.074,
"step": 1259
},
{
"epoch": 2.4,
"grad_norm": 1.1940336227416992,
"learning_rate": 5.335424008766452e-05,
"loss": 0.0533,
"step": 1260
},
{
"epoch": 2.4,
"grad_norm": 1.9031062126159668,
"learning_rate": 5.334264078218409e-05,
"loss": 0.0871,
"step": 1261
},
{
"epoch": 2.4,
"grad_norm": 1.2414464950561523,
"learning_rate": 5.33310326261832e-05,
"loss": 0.0358,
"step": 1262
},
{
"epoch": 2.4,
"grad_norm": 1.1730962991714478,
"learning_rate": 5.331941562406319e-05,
"loss": 0.0681,
"step": 1263
},
{
"epoch": 2.4,
"grad_norm": 0.9594612121582031,
"learning_rate": 5.330778978022869e-05,
"loss": 0.0694,
"step": 1264
},
{
"epoch": 2.4,
"grad_norm": 1.0870344638824463,
"learning_rate": 5.329615509908774e-05,
"loss": 0.0582,
"step": 1265
},
{
"epoch": 2.41,
"grad_norm": 1.6430423259735107,
"learning_rate": 5.3284511585051706e-05,
"loss": 0.0546,
"step": 1266
},
{
"epoch": 2.41,
"grad_norm": 1.4444844722747803,
"learning_rate": 5.327285924253529e-05,
"loss": 0.0598,
"step": 1267
},
{
"epoch": 2.41,
"grad_norm": 1.1068073511123657,
"learning_rate": 5.3261198075956574e-05,
"loss": 0.0499,
"step": 1268
},
{
"epoch": 2.41,
"grad_norm": 1.190974473953247,
"learning_rate": 5.3249528089736956e-05,
"loss": 0.0662,
"step": 1269
},
{
"epoch": 2.41,
"grad_norm": 1.2237499952316284,
"learning_rate": 5.323784928830121e-05,
"loss": 0.0533,
"step": 1270
},
{
"epoch": 2.42,
"grad_norm": 1.327459454536438,
"learning_rate": 5.322616167607741e-05,
"loss": 0.054,
"step": 1271
},
{
"epoch": 2.42,
"grad_norm": 1.5490281581878662,
"learning_rate": 5.3214465257497014e-05,
"loss": 0.0712,
"step": 1272
},
{
"epoch": 2.42,
"grad_norm": 1.4030593633651733,
"learning_rate": 5.320276003699477e-05,
"loss": 0.0786,
"step": 1273
},
{
"epoch": 2.42,
"grad_norm": 1.4227803945541382,
"learning_rate": 5.3191046019008814e-05,
"loss": 0.0553,
"step": 1274
},
{
"epoch": 2.42,
"grad_norm": 1.0864344835281372,
"learning_rate": 5.3179323207980596e-05,
"loss": 0.0323,
"step": 1275
},
{
"epoch": 2.43,
"grad_norm": 1.2288326025009155,
"learning_rate": 5.316759160835489e-05,
"loss": 0.0535,
"step": 1276
},
{
"epoch": 2.43,
"grad_norm": 1.552797794342041,
"learning_rate": 5.315585122457981e-05,
"loss": 0.0658,
"step": 1277
},
{
"epoch": 2.43,
"grad_norm": 1.4772969484329224,
"learning_rate": 5.314410206110682e-05,
"loss": 0.0534,
"step": 1278
},
{
"epoch": 2.43,
"grad_norm": 1.4350897073745728,
"learning_rate": 5.313234412239068e-05,
"loss": 0.068,
"step": 1279
},
{
"epoch": 2.43,
"grad_norm": 1.3091074228286743,
"learning_rate": 5.3120577412889474e-05,
"loss": 0.0501,
"step": 1280
},
{
"epoch": 2.44,
"grad_norm": 1.4543299674987793,
"learning_rate": 5.310880193706466e-05,
"loss": 0.0616,
"step": 1281
},
{
"epoch": 2.44,
"grad_norm": 1.505179524421692,
"learning_rate": 5.3097017699380986e-05,
"loss": 0.0587,
"step": 1282
},
{
"epoch": 2.44,
"grad_norm": 0.9245684146881104,
"learning_rate": 5.308522470430652e-05,
"loss": 0.0513,
"step": 1283
},
{
"epoch": 2.44,
"grad_norm": 1.3417291641235352,
"learning_rate": 5.307342295631265e-05,
"loss": 0.0636,
"step": 1284
},
{
"epoch": 2.44,
"grad_norm": 1.411302089691162,
"learning_rate": 5.306161245987409e-05,
"loss": 0.0554,
"step": 1285
},
{
"epoch": 2.44,
"grad_norm": 1.425968885421753,
"learning_rate": 5.3049793219468875e-05,
"loss": 0.0665,
"step": 1286
},
{
"epoch": 2.45,
"grad_norm": 0.9820340871810913,
"learning_rate": 5.303796523957835e-05,
"loss": 0.0353,
"step": 1287
},
{
"epoch": 2.45,
"grad_norm": 1.7293184995651245,
"learning_rate": 5.3026128524687167e-05,
"loss": 0.0664,
"step": 1288
},
{
"epoch": 2.45,
"grad_norm": 1.2528949975967407,
"learning_rate": 5.301428307928331e-05,
"loss": 0.0455,
"step": 1289
},
{
"epoch": 2.45,
"grad_norm": 1.545236349105835,
"learning_rate": 5.3002428907858054e-05,
"loss": 0.0653,
"step": 1290
},
{
"epoch": 2.45,
"grad_norm": 1.4326586723327637,
"learning_rate": 5.299056601490597e-05,
"loss": 0.0588,
"step": 1291
},
{
"epoch": 2.46,
"grad_norm": 1.125520944595337,
"learning_rate": 5.297869440492499e-05,
"loss": 0.0548,
"step": 1292
},
{
"epoch": 2.46,
"grad_norm": 1.2554142475128174,
"learning_rate": 5.296681408241628e-05,
"loss": 0.0617,
"step": 1293
},
{
"epoch": 2.46,
"grad_norm": 1.580428957939148,
"learning_rate": 5.295492505188437e-05,
"loss": 0.0621,
"step": 1294
},
{
"epoch": 2.46,
"grad_norm": 1.7090961933135986,
"learning_rate": 5.294302731783706e-05,
"loss": 0.0769,
"step": 1295
},
{
"epoch": 2.46,
"grad_norm": 1.518317699432373,
"learning_rate": 5.293112088478544e-05,
"loss": 0.0666,
"step": 1296
},
{
"epoch": 2.47,
"grad_norm": 2.0082004070281982,
"learning_rate": 5.291920575724394e-05,
"loss": 0.0673,
"step": 1297
},
{
"epoch": 2.47,
"grad_norm": 1.127295970916748,
"learning_rate": 5.290728193973024e-05,
"loss": 0.0709,
"step": 1298
},
{
"epoch": 2.47,
"grad_norm": 1.3025023937225342,
"learning_rate": 5.289534943676535e-05,
"loss": 0.0527,
"step": 1299
},
{
"epoch": 2.47,
"grad_norm": 1.429404377937317,
"learning_rate": 5.288340825287356e-05,
"loss": 0.0702,
"step": 1300
},
{
"epoch": 2.47,
"eval_blimp_filtered_avg": 0.6974626865671641,
"eval_blimp_filtered_std": 0.005069813522353524,
"step": 1300
},
{
"epoch": 2.47,
"eval_blimp_supplement_avg": 0.7521551724137931,
"eval_blimp_supplement_std": 0.018330310976722223,
"step": 1300
},
{
"epoch": 2.47,
"eval_vqa_filtered_avg": 0.41,
"eval_vqa_filtered_std": 0.04943110704237102,
"step": 1300
},
{
"epoch": 2.47,
"eval_winoground_filtered_avg": 0.45,
"eval_winoground_filtered_std": 0.05,
"step": 1300
},
{
"epoch": 2.47,
"grad_norm": 1.2422250509262085,
"learning_rate": 5.2871458392582446e-05,
"loss": 0.0652,
"step": 1301
},
{
"epoch": 2.48,
"grad_norm": 1.3641676902770996,
"learning_rate": 5.285949986042287e-05,
"loss": 0.0616,
"step": 1302
},
{
"epoch": 2.48,
"grad_norm": 1.4342631101608276,
"learning_rate": 5.2847532660929e-05,
"loss": 0.0535,
"step": 1303
},
{
"epoch": 2.48,
"grad_norm": 1.0171459913253784,
"learning_rate": 5.283555679863829e-05,
"loss": 0.048,
"step": 1304
},
{
"epoch": 2.48,
"grad_norm": 1.1762173175811768,
"learning_rate": 5.282357227809145e-05,
"loss": 0.0712,
"step": 1305
},
{
"epoch": 2.48,
"grad_norm": 1.074406385421753,
"learning_rate": 5.28115791038325e-05,
"loss": 0.0427,
"step": 1306
},
{
"epoch": 2.48,
"grad_norm": 1.2737088203430176,
"learning_rate": 5.2799577280408745e-05,
"loss": 0.0339,
"step": 1307
},
{
"epoch": 2.49,
"grad_norm": 1.31791090965271,
"learning_rate": 5.278756681237074e-05,
"loss": 0.0703,
"step": 1308
},
{
"epoch": 2.49,
"grad_norm": 1.4465645551681519,
"learning_rate": 5.277554770427234e-05,
"loss": 0.0532,
"step": 1309
},
{
"epoch": 2.49,
"grad_norm": 1.6235612630844116,
"learning_rate": 5.276351996067069e-05,
"loss": 0.0719,
"step": 1310
},
{
"epoch": 2.49,
"grad_norm": 1.2076858282089233,
"learning_rate": 5.275148358612616e-05,
"loss": 0.0666,
"step": 1311
},
{
"epoch": 2.49,
"grad_norm": 1.3544360399246216,
"learning_rate": 5.273943858520243e-05,
"loss": 0.0743,
"step": 1312
},
{
"epoch": 2.5,
"grad_norm": 1.182690978050232,
"learning_rate": 5.2727384962466464e-05,
"loss": 0.0552,
"step": 1313
},
{
"epoch": 2.5,
"grad_norm": 1.0207767486572266,
"learning_rate": 5.271532272248846e-05,
"loss": 0.0378,
"step": 1314
},
{
"epoch": 2.5,
"grad_norm": 1.3264497518539429,
"learning_rate": 5.2703251869841905e-05,
"loss": 0.0516,
"step": 1315
},
{
"epoch": 2.5,
"grad_norm": 0.9967998266220093,
"learning_rate": 5.269117240910353e-05,
"loss": 0.0382,
"step": 1316
},
{
"epoch": 2.5,
"grad_norm": 1.1323212385177612,
"learning_rate": 5.2679084344853355e-05,
"loss": 0.0416,
"step": 1317
},
{
"epoch": 2.51,
"grad_norm": 1.1304579973220825,
"learning_rate": 5.2666987681674654e-05,
"loss": 0.0384,
"step": 1318
},
{
"epoch": 2.51,
"grad_norm": 1.9932969808578491,
"learning_rate": 5.265488242415396e-05,
"loss": 0.0654,
"step": 1319
},
{
"epoch": 2.51,
"grad_norm": 1.3726540803909302,
"learning_rate": 5.264276857688105e-05,
"loss": 0.064,
"step": 1320
},
{
"epoch": 2.51,
"grad_norm": 1.280343770980835,
"learning_rate": 5.2630646144448984e-05,
"loss": 0.0586,
"step": 1321
},
{
"epoch": 2.51,
"grad_norm": 1.581494927406311,
"learning_rate": 5.2618515131454064e-05,
"loss": 0.0716,
"step": 1322
},
{
"epoch": 2.52,
"grad_norm": 1.235398292541504,
"learning_rate": 5.2606375542495834e-05,
"loss": 0.0422,
"step": 1323
},
{
"epoch": 2.52,
"grad_norm": 1.6975948810577393,
"learning_rate": 5.259422738217712e-05,
"loss": 0.0782,
"step": 1324
},
{
"epoch": 2.52,
"grad_norm": 1.1720950603485107,
"learning_rate": 5.258207065510396e-05,
"loss": 0.0592,
"step": 1325
},
{
"epoch": 2.52,
"grad_norm": 1.0369981527328491,
"learning_rate": 5.2569905365885666e-05,
"loss": 0.0472,
"step": 1326
},
{
"epoch": 2.52,
"grad_norm": 1.2626200914382935,
"learning_rate": 5.25577315191348e-05,
"loss": 0.0607,
"step": 1327
},
{
"epoch": 2.52,
"grad_norm": 1.0311836004257202,
"learning_rate": 5.254554911946715e-05,
"loss": 0.0522,
"step": 1328
},
{
"epoch": 2.53,
"grad_norm": 1.4834760427474976,
"learning_rate": 5.2533358171501755e-05,
"loss": 0.0722,
"step": 1329
},
{
"epoch": 2.53,
"grad_norm": 1.2561490535736084,
"learning_rate": 5.25211586798609e-05,
"loss": 0.0558,
"step": 1330
},
{
"epoch": 2.53,
"grad_norm": 1.3886915445327759,
"learning_rate": 5.2508950649170094e-05,
"loss": 0.0585,
"step": 1331
},
{
"epoch": 2.53,
"grad_norm": 1.3595167398452759,
"learning_rate": 5.24967340840581e-05,
"loss": 0.0453,
"step": 1332
},
{
"epoch": 2.53,
"grad_norm": 1.8665841817855835,
"learning_rate": 5.248450898915693e-05,
"loss": 0.0801,
"step": 1333
},
{
"epoch": 2.54,
"grad_norm": 1.4385470151901245,
"learning_rate": 5.247227536910178e-05,
"loss": 0.0676,
"step": 1334
},
{
"epoch": 2.54,
"grad_norm": 1.4748903512954712,
"learning_rate": 5.2460033228531135e-05,
"loss": 0.0552,
"step": 1335
},
{
"epoch": 2.54,
"grad_norm": 1.2043030261993408,
"learning_rate": 5.244778257208667e-05,
"loss": 0.0414,
"step": 1336
},
{
"epoch": 2.54,
"grad_norm": 1.2462152242660522,
"learning_rate": 5.2435523404413305e-05,
"loss": 0.0609,
"step": 1337
},
{
"epoch": 2.54,
"grad_norm": 1.0698647499084473,
"learning_rate": 5.24232557301592e-05,
"loss": 0.0362,
"step": 1338
},
{
"epoch": 2.55,
"grad_norm": 1.4026707410812378,
"learning_rate": 5.2410979553975715e-05,
"loss": 0.0658,
"step": 1339
},
{
"epoch": 2.55,
"grad_norm": 1.2336044311523438,
"learning_rate": 5.2398694880517454e-05,
"loss": 0.0384,
"step": 1340
},
{
"epoch": 2.55,
"grad_norm": 1.1586161851882935,
"learning_rate": 5.2386401714442225e-05,
"loss": 0.0553,
"step": 1341
},
{
"epoch": 2.55,
"grad_norm": 1.6403613090515137,
"learning_rate": 5.2374100060411065e-05,
"loss": 0.0716,
"step": 1342
},
{
"epoch": 2.55,
"grad_norm": 1.1715034246444702,
"learning_rate": 5.2361789923088246e-05,
"loss": 0.0605,
"step": 1343
},
{
"epoch": 2.56,
"grad_norm": 1.3810871839523315,
"learning_rate": 5.234947130714122e-05,
"loss": 0.0608,
"step": 1344
},
{
"epoch": 2.56,
"grad_norm": 1.5715049505233765,
"learning_rate": 5.233714421724068e-05,
"loss": 0.0757,
"step": 1345
},
{
"epoch": 2.56,
"grad_norm": 1.0748265981674194,
"learning_rate": 5.232480865806053e-05,
"loss": 0.0493,
"step": 1346
},
{
"epoch": 2.56,
"grad_norm": 1.2202306985855103,
"learning_rate": 5.2312464634277874e-05,
"loss": 0.0398,
"step": 1347
},
{
"epoch": 2.56,
"grad_norm": 1.4714404344558716,
"learning_rate": 5.230011215057304e-05,
"loss": 0.0823,
"step": 1348
},
{
"epoch": 2.56,
"grad_norm": 1.197867751121521,
"learning_rate": 5.228775121162956e-05,
"loss": 0.0592,
"step": 1349
},
{
"epoch": 2.57,
"grad_norm": 1.361423134803772,
"learning_rate": 5.227538182213414e-05,
"loss": 0.0379,
"step": 1350
},
{
"epoch": 2.57,
"grad_norm": 1.8070042133331299,
"learning_rate": 5.2263003986776754e-05,
"loss": 0.0719,
"step": 1351
},
{
"epoch": 2.57,
"grad_norm": 0.8731653094291687,
"learning_rate": 5.225061771025053e-05,
"loss": 0.0354,
"step": 1352
},
{
"epoch": 2.57,
"grad_norm": 1.6901946067810059,
"learning_rate": 5.223822299725179e-05,
"loss": 0.0601,
"step": 1353
},
{
"epoch": 2.57,
"grad_norm": 1.3738157749176025,
"learning_rate": 5.2225819852480104e-05,
"loss": 0.056,
"step": 1354
},
{
"epoch": 2.58,
"grad_norm": 1.1929755210876465,
"learning_rate": 5.2213408280638176e-05,
"loss": 0.0503,
"step": 1355
},
{
"epoch": 2.58,
"grad_norm": 1.0309116840362549,
"learning_rate": 5.220098828643197e-05,
"loss": 0.046,
"step": 1356
},
{
"epoch": 2.58,
"grad_norm": 1.6570714712142944,
"learning_rate": 5.218855987457059e-05,
"loss": 0.0698,
"step": 1357
},
{
"epoch": 2.58,
"grad_norm": 1.0366765260696411,
"learning_rate": 5.217612304976635e-05,
"loss": 0.0368,
"step": 1358
},
{
"epoch": 2.58,
"grad_norm": 1.8921147584915161,
"learning_rate": 5.216367781673477e-05,
"loss": 0.0606,
"step": 1359
},
{
"epoch": 2.59,
"grad_norm": 1.8033537864685059,
"learning_rate": 5.215122418019453e-05,
"loss": 0.076,
"step": 1360
},
{
"epoch": 2.59,
"grad_norm": 1.4140779972076416,
"learning_rate": 5.2138762144867536e-05,
"loss": 0.0582,
"step": 1361
},
{
"epoch": 2.59,
"grad_norm": 1.3936694860458374,
"learning_rate": 5.212629171547883e-05,
"loss": 0.0598,
"step": 1362
},
{
"epoch": 2.59,
"grad_norm": 1.3207770586013794,
"learning_rate": 5.211381289675666e-05,
"loss": 0.0673,
"step": 1363
},
{
"epoch": 2.59,
"grad_norm": 1.1927788257598877,
"learning_rate": 5.210132569343247e-05,
"loss": 0.0566,
"step": 1364
},
{
"epoch": 2.6,
"grad_norm": 1.4058153629302979,
"learning_rate": 5.208883011024085e-05,
"loss": 0.0515,
"step": 1365
},
{
"epoch": 2.6,
"grad_norm": 1.4498968124389648,
"learning_rate": 5.20763261519196e-05,
"loss": 0.0582,
"step": 1366
},
{
"epoch": 2.6,
"grad_norm": 0.9277570247650146,
"learning_rate": 5.2063813823209675e-05,
"loss": 0.0506,
"step": 1367
},
{
"epoch": 2.6,
"grad_norm": 1.0205438137054443,
"learning_rate": 5.205129312885521e-05,
"loss": 0.0413,
"step": 1368
},
{
"epoch": 2.6,
"grad_norm": 1.264689564704895,
"learning_rate": 5.203876407360349e-05,
"loss": 0.0725,
"step": 1369
},
{
"epoch": 2.6,
"grad_norm": 1.5223809480667114,
"learning_rate": 5.202622666220503e-05,
"loss": 0.0765,
"step": 1370
},
{
"epoch": 2.61,
"grad_norm": 1.5418691635131836,
"learning_rate": 5.2013680899413456e-05,
"loss": 0.0835,
"step": 1371
},
{
"epoch": 2.61,
"grad_norm": 1.369511365890503,
"learning_rate": 5.200112678998557e-05,
"loss": 0.0774,
"step": 1372
},
{
"epoch": 2.61,
"grad_norm": 1.2308417558670044,
"learning_rate": 5.198856433868136e-05,
"loss": 0.0618,
"step": 1373
},
{
"epoch": 2.61,
"grad_norm": 1.0067722797393799,
"learning_rate": 5.197599355026397e-05,
"loss": 0.0388,
"step": 1374
},
{
"epoch": 2.61,
"grad_norm": 1.4092450141906738,
"learning_rate": 5.196341442949968e-05,
"loss": 0.0668,
"step": 1375
},
{
"epoch": 2.62,
"grad_norm": 1.2684311866760254,
"learning_rate": 5.195082698115795e-05,
"loss": 0.0471,
"step": 1376
},
{
"epoch": 2.62,
"grad_norm": 1.1197936534881592,
"learning_rate": 5.193823121001142e-05,
"loss": 0.0701,
"step": 1377
},
{
"epoch": 2.62,
"grad_norm": 1.491631269454956,
"learning_rate": 5.192562712083584e-05,
"loss": 0.0573,
"step": 1378
},
{
"epoch": 2.62,
"grad_norm": 1.3656389713287354,
"learning_rate": 5.191301471841015e-05,
"loss": 0.066,
"step": 1379
},
{
"epoch": 2.62,
"grad_norm": 1.1795417070388794,
"learning_rate": 5.1900394007516413e-05,
"loss": 0.0714,
"step": 1380
},
{
"epoch": 2.63,
"grad_norm": 1.173470139503479,
"learning_rate": 5.188776499293987e-05,
"loss": 0.0516,
"step": 1381
},
{
"epoch": 2.63,
"grad_norm": 1.1300203800201416,
"learning_rate": 5.18751276794689e-05,
"loss": 0.0541,
"step": 1382
},
{
"epoch": 2.63,
"grad_norm": 1.058279275894165,
"learning_rate": 5.186248207189501e-05,
"loss": 0.0637,
"step": 1383
},
{
"epoch": 2.63,
"grad_norm": 1.2247236967086792,
"learning_rate": 5.184982817501287e-05,
"loss": 0.0531,
"step": 1384
},
{
"epoch": 2.63,
"grad_norm": 0.9494766592979431,
"learning_rate": 5.18371659936203e-05,
"loss": 0.0312,
"step": 1385
},
{
"epoch": 2.63,
"grad_norm": 1.072037696838379,
"learning_rate": 5.182449553251827e-05,
"loss": 0.0431,
"step": 1386
},
{
"epoch": 2.64,
"grad_norm": 1.385895013809204,
"learning_rate": 5.181181679651083e-05,
"loss": 0.0509,
"step": 1387
},
{
"epoch": 2.64,
"grad_norm": 1.3322844505310059,
"learning_rate": 5.1799129790405226e-05,
"loss": 0.0578,
"step": 1388
},
{
"epoch": 2.64,
"grad_norm": 1.1688324213027954,
"learning_rate": 5.178643451901184e-05,
"loss": 0.04,
"step": 1389
},
{
"epoch": 2.64,
"grad_norm": 1.5262318849563599,
"learning_rate": 5.177373098714414e-05,
"loss": 0.0711,
"step": 1390
},
{
"epoch": 2.64,
"grad_norm": 1.2698298692703247,
"learning_rate": 5.1761019199618784e-05,
"loss": 0.0425,
"step": 1391
},
{
"epoch": 2.65,
"grad_norm": 1.0538215637207031,
"learning_rate": 5.174829916125552e-05,
"loss": 0.0604,
"step": 1392
},
{
"epoch": 2.65,
"grad_norm": 1.406567096710205,
"learning_rate": 5.1735570876877236e-05,
"loss": 0.0492,
"step": 1393
},
{
"epoch": 2.65,
"grad_norm": 1.4662601947784424,
"learning_rate": 5.172283435130994e-05,
"loss": 0.0558,
"step": 1394
},
{
"epoch": 2.65,
"grad_norm": 1.1350047588348389,
"learning_rate": 5.1710089589382785e-05,
"loss": 0.0509,
"step": 1395
},
{
"epoch": 2.65,
"grad_norm": 1.1119331121444702,
"learning_rate": 5.1697336595928033e-05,
"loss": 0.0479,
"step": 1396
},
{
"epoch": 2.66,
"grad_norm": 1.3725008964538574,
"learning_rate": 5.168457537578106e-05,
"loss": 0.066,
"step": 1397
},
{
"epoch": 2.66,
"grad_norm": 1.3826853036880493,
"learning_rate": 5.167180593378037e-05,
"loss": 0.0425,
"step": 1398
},
{
"epoch": 2.66,
"grad_norm": 0.9081801772117615,
"learning_rate": 5.165902827476759e-05,
"loss": 0.0522,
"step": 1399
},
{
"epoch": 2.66,
"grad_norm": 1.5069169998168945,
"learning_rate": 5.164624240358746e-05,
"loss": 0.0713,
"step": 1400
},
{
"epoch": 2.66,
"eval_blimp_filtered_avg": 0.7105970149253731,
"eval_blimp_filtered_std": 0.005034985865816392,
"step": 1400
},
{
"epoch": 2.66,
"eval_blimp_supplement_avg": 0.7737068965517241,
"eval_blimp_supplement_std": 0.017633040526537185,
"step": 1400
},
{
"epoch": 2.66,
"eval_vqa_filtered_avg": 0.39,
"eval_vqa_filtered_std": 0.04902071300001975,
"step": 1400
},
{
"epoch": 2.66,
"eval_winoground_filtered_avg": 0.49,
"eval_winoground_filtered_std": 0.05024183937956912,
"step": 1400
},
{
"epoch": 2.66,
"grad_norm": 1.4957689046859741,
"learning_rate": 5.1633448325087814e-05,
"loss": 0.0661,
"step": 1401
},
{
"epoch": 2.67,
"grad_norm": 1.0628505945205688,
"learning_rate": 5.162064604411962e-05,
"loss": 0.0442,
"step": 1402
},
{
"epoch": 2.67,
"grad_norm": 1.2679592370986938,
"learning_rate": 5.160783556553696e-05,
"loss": 0.0461,
"step": 1403
},
{
"epoch": 2.67,
"grad_norm": 1.5347133874893188,
"learning_rate": 5.1595016894197e-05,
"loss": 0.0548,
"step": 1404
},
{
"epoch": 2.67,
"grad_norm": 2.1136763095855713,
"learning_rate": 5.158219003496003e-05,
"loss": 0.056,
"step": 1405
},
{
"epoch": 2.67,
"grad_norm": 1.404384970664978,
"learning_rate": 5.156935499268944e-05,
"loss": 0.038,
"step": 1406
},
{
"epoch": 2.67,
"grad_norm": 1.3219765424728394,
"learning_rate": 5.1556511772251726e-05,
"loss": 0.0372,
"step": 1407
},
{
"epoch": 2.68,
"grad_norm": 1.3650156259536743,
"learning_rate": 5.154366037851649e-05,
"loss": 0.068,
"step": 1408
},
{
"epoch": 2.68,
"grad_norm": 1.226851224899292,
"learning_rate": 5.153080081635641e-05,
"loss": 0.0553,
"step": 1409
},
{
"epoch": 2.68,
"grad_norm": 1.2086094617843628,
"learning_rate": 5.151793309064728e-05,
"loss": 0.069,
"step": 1410
},
{
"epoch": 2.68,
"grad_norm": 1.0489845275878906,
"learning_rate": 5.1505057206267984e-05,
"loss": 0.0451,
"step": 1411
},
{
"epoch": 2.68,
"grad_norm": 1.1169242858886719,
"learning_rate": 5.149217316810052e-05,
"loss": 0.0515,
"step": 1412
},
{
"epoch": 2.69,
"grad_norm": 1.2907397747039795,
"learning_rate": 5.147928098102993e-05,
"loss": 0.0585,
"step": 1413
},
{
"epoch": 2.69,
"grad_norm": 1.691044569015503,
"learning_rate": 5.146638064994439e-05,
"loss": 0.0491,
"step": 1414
},
{
"epoch": 2.69,
"grad_norm": 1.2878209352493286,
"learning_rate": 5.145347217973515e-05,
"loss": 0.0457,
"step": 1415
},
{
"epoch": 2.69,
"grad_norm": 1.3205225467681885,
"learning_rate": 5.144055557529654e-05,
"loss": 0.0651,
"step": 1416
},
{
"epoch": 2.69,
"grad_norm": 1.3645832538604736,
"learning_rate": 5.142763084152597e-05,
"loss": 0.076,
"step": 1417
},
{
"epoch": 2.7,
"grad_norm": 1.48106050491333,
"learning_rate": 5.141469798332395e-05,
"loss": 0.0674,
"step": 1418
},
{
"epoch": 2.7,
"grad_norm": 1.2958459854125977,
"learning_rate": 5.140175700559406e-05,
"loss": 0.0491,
"step": 1419
},
{
"epoch": 2.7,
"grad_norm": 1.3527158498764038,
"learning_rate": 5.138880791324295e-05,
"loss": 0.0449,
"step": 1420
},
{
"epoch": 2.7,
"grad_norm": 1.1957794427871704,
"learning_rate": 5.137585071118037e-05,
"loss": 0.0371,
"step": 1421
},
{
"epoch": 2.7,
"grad_norm": 1.2718093395233154,
"learning_rate": 5.136288540431913e-05,
"loss": 0.0668,
"step": 1422
},
{
"epoch": 2.71,
"grad_norm": 1.1913398504257202,
"learning_rate": 5.13499119975751e-05,
"loss": 0.033,
"step": 1423
},
{
"epoch": 2.71,
"grad_norm": 1.0553014278411865,
"learning_rate": 5.1336930495867245e-05,
"loss": 0.0323,
"step": 1424
},
{
"epoch": 2.71,
"grad_norm": 1.2635596990585327,
"learning_rate": 5.1323940904117595e-05,
"loss": 0.0577,
"step": 1425
},
{
"epoch": 2.71,
"grad_norm": 1.6414707899093628,
"learning_rate": 5.1310943227251236e-05,
"loss": 0.0654,
"step": 1426
},
{
"epoch": 2.71,
"grad_norm": 1.2460765838623047,
"learning_rate": 5.129793747019632e-05,
"loss": 0.0413,
"step": 1427
},
{
"epoch": 2.71,
"grad_norm": 1.9726555347442627,
"learning_rate": 5.128492363788407e-05,
"loss": 0.0692,
"step": 1428
},
{
"epoch": 2.72,
"grad_norm": 2.187358856201172,
"learning_rate": 5.127190173524877e-05,
"loss": 0.062,
"step": 1429
},
{
"epoch": 2.72,
"grad_norm": 1.2415469884872437,
"learning_rate": 5.125887176722777e-05,
"loss": 0.0515,
"step": 1430
},
{
"epoch": 2.72,
"grad_norm": 1.8226250410079956,
"learning_rate": 5.1245833738761456e-05,
"loss": 0.0656,
"step": 1431
},
{
"epoch": 2.72,
"grad_norm": 1.5468683242797852,
"learning_rate": 5.1232787654793304e-05,
"loss": 0.0729,
"step": 1432
},
{
"epoch": 2.72,
"grad_norm": 1.1349869966506958,
"learning_rate": 5.1219733520269805e-05,
"loss": 0.0428,
"step": 1433
},
{
"epoch": 2.73,
"grad_norm": 1.0630717277526855,
"learning_rate": 5.120667134014053e-05,
"loss": 0.0376,
"step": 1434
},
{
"epoch": 2.73,
"grad_norm": 1.0929280519485474,
"learning_rate": 5.119360111935811e-05,
"loss": 0.0403,
"step": 1435
},
{
"epoch": 2.73,
"grad_norm": 1.5913951396942139,
"learning_rate": 5.118052286287818e-05,
"loss": 0.0665,
"step": 1436
},
{
"epoch": 2.73,
"grad_norm": 1.3882554769515991,
"learning_rate": 5.1167436575659474e-05,
"loss": 0.0505,
"step": 1437
},
{
"epoch": 2.73,
"grad_norm": 0.9949247241020203,
"learning_rate": 5.115434226266372e-05,
"loss": 0.0313,
"step": 1438
},
{
"epoch": 2.74,
"grad_norm": 1.439896821975708,
"learning_rate": 5.114123992885574e-05,
"loss": 0.0654,
"step": 1439
},
{
"epoch": 2.74,
"grad_norm": 1.2304211854934692,
"learning_rate": 5.112812957920336e-05,
"loss": 0.0555,
"step": 1440
},
{
"epoch": 2.74,
"grad_norm": 1.1082305908203125,
"learning_rate": 5.111501121867746e-05,
"loss": 0.0565,
"step": 1441
},
{
"epoch": 2.74,
"grad_norm": 1.5437490940093994,
"learning_rate": 5.110188485225195e-05,
"loss": 0.0793,
"step": 1442
},
{
"epoch": 2.74,
"grad_norm": 1.0498133897781372,
"learning_rate": 5.1088750484903805e-05,
"loss": 0.037,
"step": 1443
},
{
"epoch": 2.75,
"grad_norm": 1.113671898841858,
"learning_rate": 5.107560812161298e-05,
"loss": 0.05,
"step": 1444
},
{
"epoch": 2.75,
"grad_norm": 1.5177197456359863,
"learning_rate": 5.10624577673625e-05,
"loss": 0.0532,
"step": 1445
},
{
"epoch": 2.75,
"grad_norm": 1.4293206930160522,
"learning_rate": 5.104929942713842e-05,
"loss": 0.0673,
"step": 1446
},
{
"epoch": 2.75,
"grad_norm": 1.8016672134399414,
"learning_rate": 5.10361331059298e-05,
"loss": 0.0754,
"step": 1447
},
{
"epoch": 2.75,
"grad_norm": 1.3068186044692993,
"learning_rate": 5.1022958808728756e-05,
"loss": 0.0725,
"step": 1448
},
{
"epoch": 2.75,
"grad_norm": 1.2950875759124756,
"learning_rate": 5.10097765405304e-05,
"loss": 0.0714,
"step": 1449
},
{
"epoch": 2.76,
"grad_norm": 1.2699477672576904,
"learning_rate": 5.0996586306332874e-05,
"loss": 0.0561,
"step": 1450
},
{
"epoch": 2.76,
"grad_norm": 1.286299228668213,
"learning_rate": 5.098338811113736e-05,
"loss": 0.0568,
"step": 1451
},
{
"epoch": 2.76,
"grad_norm": 1.4500685930252075,
"learning_rate": 5.097018195994804e-05,
"loss": 0.0903,
"step": 1452
},
{
"epoch": 2.76,
"grad_norm": 1.4785531759262085,
"learning_rate": 5.095696785777211e-05,
"loss": 0.051,
"step": 1453
},
{
"epoch": 2.76,
"grad_norm": 1.1381351947784424,
"learning_rate": 5.0943745809619774e-05,
"loss": 0.0594,
"step": 1454
},
{
"epoch": 2.77,
"grad_norm": 0.9356142282485962,
"learning_rate": 5.093051582050429e-05,
"loss": 0.0295,
"step": 1455
},
{
"epoch": 2.77,
"grad_norm": 1.3292661905288696,
"learning_rate": 5.091727789544188e-05,
"loss": 0.0532,
"step": 1456
},
{
"epoch": 2.77,
"grad_norm": 1.4142946004867554,
"learning_rate": 5.0904032039451786e-05,
"loss": 0.0508,
"step": 1457
},
{
"epoch": 2.77,
"grad_norm": 1.006205677986145,
"learning_rate": 5.089077825755628e-05,
"loss": 0.0484,
"step": 1458
},
{
"epoch": 2.77,
"grad_norm": 1.1057794094085693,
"learning_rate": 5.0877516554780626e-05,
"loss": 0.0442,
"step": 1459
},
{
"epoch": 2.78,
"grad_norm": 0.9902070164680481,
"learning_rate": 5.0864246936153064e-05,
"loss": 0.0475,
"step": 1460
},
{
"epoch": 2.78,
"grad_norm": 1.318869709968567,
"learning_rate": 5.0850969406704896e-05,
"loss": 0.0484,
"step": 1461
},
{
"epoch": 2.78,
"grad_norm": 1.1171830892562866,
"learning_rate": 5.0837683971470344e-05,
"loss": 0.0495,
"step": 1462
},
{
"epoch": 2.78,
"grad_norm": 0.9523420929908752,
"learning_rate": 5.0824390635486706e-05,
"loss": 0.0346,
"step": 1463
},
{
"epoch": 2.78,
"grad_norm": 1.015285611152649,
"learning_rate": 5.081108940379423e-05,
"loss": 0.0578,
"step": 1464
},
{
"epoch": 2.79,
"grad_norm": 1.2422425746917725,
"learning_rate": 5.079778028143615e-05,
"loss": 0.0322,
"step": 1465
},
{
"epoch": 2.79,
"grad_norm": 1.1792361736297607,
"learning_rate": 5.0784463273458746e-05,
"loss": 0.0553,
"step": 1466
},
{
"epoch": 2.79,
"grad_norm": 1.1954118013381958,
"learning_rate": 5.077113838491122e-05,
"loss": 0.0341,
"step": 1467
},
{
"epoch": 2.79,
"grad_norm": 1.3955225944519043,
"learning_rate": 5.0757805620845807e-05,
"loss": 0.0642,
"step": 1468
},
{
"epoch": 2.79,
"grad_norm": 1.4002469778060913,
"learning_rate": 5.074446498631773e-05,
"loss": 0.0456,
"step": 1469
},
{
"epoch": 2.79,
"grad_norm": 0.9521331191062927,
"learning_rate": 5.073111648638514e-05,
"loss": 0.0418,
"step": 1470
},
{
"epoch": 2.8,
"grad_norm": 1.162889838218689,
"learning_rate": 5.071776012610925e-05,
"loss": 0.051,
"step": 1471
},
{
"epoch": 2.8,
"grad_norm": 1.115676760673523,
"learning_rate": 5.070439591055419e-05,
"loss": 0.055,
"step": 1472
},
{
"epoch": 2.8,
"grad_norm": 0.9132686257362366,
"learning_rate": 5.069102384478711e-05,
"loss": 0.0341,
"step": 1473
},
{
"epoch": 2.8,
"grad_norm": 1.2627500295639038,
"learning_rate": 5.067764393387811e-05,
"loss": 0.0527,
"step": 1474
},
{
"epoch": 2.8,
"grad_norm": 0.8691284656524658,
"learning_rate": 5.066425618290029e-05,
"loss": 0.0409,
"step": 1475
},
{
"epoch": 2.81,
"grad_norm": 1.5841206312179565,
"learning_rate": 5.065086059692967e-05,
"loss": 0.0733,
"step": 1476
},
{
"epoch": 2.81,
"grad_norm": 0.9372629523277283,
"learning_rate": 5.063745718104532e-05,
"loss": 0.0378,
"step": 1477
},
{
"epoch": 2.81,
"grad_norm": 1.27884840965271,
"learning_rate": 5.06240459403292e-05,
"loss": 0.0493,
"step": 1478
},
{
"epoch": 2.81,
"grad_norm": 1.505638599395752,
"learning_rate": 5.0610626879866307e-05,
"loss": 0.0601,
"step": 1479
},
{
"epoch": 2.81,
"grad_norm": 1.518418788909912,
"learning_rate": 5.0597200004744534e-05,
"loss": 0.0535,
"step": 1480
},
{
"epoch": 2.82,
"grad_norm": 1.3699830770492554,
"learning_rate": 5.058376532005479e-05,
"loss": 0.051,
"step": 1481
},
{
"epoch": 2.82,
"grad_norm": 0.997786819934845,
"learning_rate": 5.057032283089094e-05,
"loss": 0.0342,
"step": 1482
},
{
"epoch": 2.82,
"grad_norm": 1.0185227394104004,
"learning_rate": 5.055687254234976e-05,
"loss": 0.0432,
"step": 1483
},
{
"epoch": 2.82,
"grad_norm": 1.4160194396972656,
"learning_rate": 5.0543414459531054e-05,
"loss": 0.0548,
"step": 1484
},
{
"epoch": 2.82,
"grad_norm": 1.0580137968063354,
"learning_rate": 5.052994858753752e-05,
"loss": 0.0377,
"step": 1485
},
{
"epoch": 2.83,
"grad_norm": 1.2387824058532715,
"learning_rate": 5.0516474931474846e-05,
"loss": 0.06,
"step": 1486
},
{
"epoch": 2.83,
"grad_norm": 1.4400213956832886,
"learning_rate": 5.0502993496451663e-05,
"loss": 0.0757,
"step": 1487
},
{
"epoch": 2.83,
"grad_norm": 1.4797674417495728,
"learning_rate": 5.048950428757954e-05,
"loss": 0.0644,
"step": 1488
},
{
"epoch": 2.83,
"grad_norm": 1.3595073223114014,
"learning_rate": 5.0476007309973e-05,
"loss": 0.0435,
"step": 1489
},
{
"epoch": 2.83,
"grad_norm": 1.1305921077728271,
"learning_rate": 5.046250256874953e-05,
"loss": 0.0458,
"step": 1490
},
{
"epoch": 2.83,
"grad_norm": 1.4059298038482666,
"learning_rate": 5.044899006902952e-05,
"loss": 0.0429,
"step": 1491
},
{
"epoch": 2.84,
"grad_norm": 1.059950590133667,
"learning_rate": 5.043546981593635e-05,
"loss": 0.0401,
"step": 1492
},
{
"epoch": 2.84,
"grad_norm": 1.7377105951309204,
"learning_rate": 5.04219418145963e-05,
"loss": 0.0633,
"step": 1493
},
{
"epoch": 2.84,
"grad_norm": 1.316605567932129,
"learning_rate": 5.040840607013861e-05,
"loss": 0.0672,
"step": 1494
},
{
"epoch": 2.84,
"grad_norm": 1.0921210050582886,
"learning_rate": 5.0394862587695444e-05,
"loss": 0.0579,
"step": 1495
},
{
"epoch": 2.84,
"grad_norm": 1.499089241027832,
"learning_rate": 5.0381311372401906e-05,
"loss": 0.0397,
"step": 1496
},
{
"epoch": 2.85,
"grad_norm": 0.9490413069725037,
"learning_rate": 5.036775242939604e-05,
"loss": 0.035,
"step": 1497
},
{
"epoch": 2.85,
"grad_norm": 1.2133374214172363,
"learning_rate": 5.03541857638188e-05,
"loss": 0.0603,
"step": 1498
},
{
"epoch": 2.85,
"grad_norm": 1.5666515827178955,
"learning_rate": 5.034061138081408e-05,
"loss": 0.0464,
"step": 1499
},
{
"epoch": 2.85,
"grad_norm": 1.1105576753616333,
"learning_rate": 5.0327029285528715e-05,
"loss": 0.0561,
"step": 1500
},
{
"epoch": 2.85,
"eval_blimp_filtered_avg": 0.7035820895522388,
"eval_blimp_filtered_std": 0.005067038783636171,
"step": 1500
},
{
"epoch": 2.85,
"eval_blimp_supplement_avg": 0.7844827586206896,
"eval_blimp_supplement_std": 0.017506099557336195,
"step": 1500
},
{
"epoch": 2.85,
"eval_vqa_filtered_avg": 0.44,
"eval_vqa_filtered_std": 0.04988876515698589,
"step": 1500
},
{
"epoch": 2.85,
"eval_winoground_filtered_avg": 0.5,
"eval_winoground_filtered_std": 0.050251890762960605,
"step": 1500
},
{
"epoch": 2.85,
"grad_norm": 1.33514404296875,
"learning_rate": 5.031343948311243e-05,
"loss": 0.0627,
"step": 1501
},
{
"epoch": 2.86,
"grad_norm": 1.1609441041946411,
"learning_rate": 5.0299841978717895e-05,
"loss": 0.0602,
"step": 1502
},
{
"epoch": 2.86,
"grad_norm": 1.1101309061050415,
"learning_rate": 5.02862367775007e-05,
"loss": 0.047,
"step": 1503
},
{
"epoch": 2.86,
"grad_norm": 1.5659412145614624,
"learning_rate": 5.027262388461934e-05,
"loss": 0.0483,
"step": 1504
},
{
"epoch": 2.86,
"grad_norm": 1.0206491947174072,
"learning_rate": 5.025900330523526e-05,
"loss": 0.0387,
"step": 1505
},
{
"epoch": 2.86,
"grad_norm": 1.7750976085662842,
"learning_rate": 5.024537504451277e-05,
"loss": 0.0778,
"step": 1506
},
{
"epoch": 2.87,
"grad_norm": 1.1326957941055298,
"learning_rate": 5.023173910761912e-05,
"loss": 0.0642,
"step": 1507
},
{
"epoch": 2.87,
"grad_norm": 1.3838077783584595,
"learning_rate": 5.021809549972448e-05,
"loss": 0.0328,
"step": 1508
},
{
"epoch": 2.87,
"grad_norm": 1.2861162424087524,
"learning_rate": 5.0204444226001906e-05,
"loss": 0.0353,
"step": 1509
},
{
"epoch": 2.87,
"grad_norm": 1.1258809566497803,
"learning_rate": 5.019078529162737e-05,
"loss": 0.0453,
"step": 1510
},
{
"epoch": 2.87,
"grad_norm": 1.3726227283477783,
"learning_rate": 5.017711870177976e-05,
"loss": 0.0583,
"step": 1511
},
{
"epoch": 2.87,
"grad_norm": 1.1814793348312378,
"learning_rate": 5.016344446164085e-05,
"loss": 0.0573,
"step": 1512
},
{
"epoch": 2.88,
"grad_norm": 1.6128199100494385,
"learning_rate": 5.0149762576395324e-05,
"loss": 0.0682,
"step": 1513
},
{
"epoch": 2.88,
"grad_norm": 1.444899559020996,
"learning_rate": 5.013607305123076e-05,
"loss": 0.043,
"step": 1514
},
{
"epoch": 2.88,
"grad_norm": 1.0560177564620972,
"learning_rate": 5.012237589133763e-05,
"loss": 0.0444,
"step": 1515
},
{
"epoch": 2.88,
"grad_norm": 1.1536376476287842,
"learning_rate": 5.01086711019093e-05,
"loss": 0.0576,
"step": 1516
},
{
"epoch": 2.88,
"grad_norm": 1.484714388847351,
"learning_rate": 5.0094958688142054e-05,
"loss": 0.063,
"step": 1517
},
{
"epoch": 2.89,
"grad_norm": 1.1258857250213623,
"learning_rate": 5.008123865523503e-05,
"loss": 0.0395,
"step": 1518
},
{
"epoch": 2.89,
"grad_norm": 1.2557101249694824,
"learning_rate": 5.0067511008390266e-05,
"loss": 0.0648,
"step": 1519
},
{
"epoch": 2.89,
"grad_norm": 1.3863277435302734,
"learning_rate": 5.005377575281272e-05,
"loss": 0.054,
"step": 1520
},
{
"epoch": 2.89,
"grad_norm": 1.1406670808792114,
"learning_rate": 5.004003289371018e-05,
"loss": 0.0401,
"step": 1521
},
{
"epoch": 2.89,
"grad_norm": 0.9542426466941833,
"learning_rate": 5.0026282436293345e-05,
"loss": 0.0413,
"step": 1522
},
{
"epoch": 2.9,
"grad_norm": 1.1478551626205444,
"learning_rate": 5.001252438577581e-05,
"loss": 0.0453,
"step": 1523
},
{
"epoch": 2.9,
"grad_norm": 0.9466152787208557,
"learning_rate": 4.999875874737403e-05,
"loss": 0.0354,
"step": 1524
},
{
"epoch": 2.9,
"grad_norm": 0.92267245054245,
"learning_rate": 4.998498552630733e-05,
"loss": 0.043,
"step": 1525
},
{
"epoch": 2.9,
"grad_norm": 1.1041741371154785,
"learning_rate": 4.997120472779793e-05,
"loss": 0.0396,
"step": 1526
},
{
"epoch": 2.9,
"grad_norm": 1.1936851739883423,
"learning_rate": 4.995741635707091e-05,
"loss": 0.0527,
"step": 1527
},
{
"epoch": 2.9,
"grad_norm": 1.2095351219177246,
"learning_rate": 4.9943620419354224e-05,
"loss": 0.0437,
"step": 1528
},
{
"epoch": 2.91,
"grad_norm": 1.4864997863769531,
"learning_rate": 4.992981691987869e-05,
"loss": 0.1188,
"step": 1529
},
{
"epoch": 2.91,
"grad_norm": 1.3914469480514526,
"learning_rate": 4.9916005863878e-05,
"loss": 0.0797,
"step": 1530
},
{
"epoch": 2.91,
"grad_norm": 0.993284285068512,
"learning_rate": 4.9902187256588725e-05,
"loss": 0.0395,
"step": 1531
},
{
"epoch": 2.91,
"grad_norm": 1.1299805641174316,
"learning_rate": 4.9888361103250255e-05,
"loss": 0.0475,
"step": 1532
},
{
"epoch": 2.91,
"grad_norm": 1.427839756011963,
"learning_rate": 4.987452740910489e-05,
"loss": 0.0678,
"step": 1533
},
{
"epoch": 2.92,
"grad_norm": 1.363171100616455,
"learning_rate": 4.986068617939777e-05,
"loss": 0.0398,
"step": 1534
},
{
"epoch": 2.92,
"grad_norm": 0.9385173916816711,
"learning_rate": 4.984683741937688e-05,
"loss": 0.0433,
"step": 1535
},
{
"epoch": 2.92,
"grad_norm": 1.0321054458618164,
"learning_rate": 4.9832981134293086e-05,
"loss": 0.0489,
"step": 1536
},
{
"epoch": 2.92,
"grad_norm": 1.4978641271591187,
"learning_rate": 4.981911732940008e-05,
"loss": 0.0647,
"step": 1537
},
{
"epoch": 2.92,
"grad_norm": 1.2878303527832031,
"learning_rate": 4.980524600995442e-05,
"loss": 0.0464,
"step": 1538
},
{
"epoch": 2.93,
"grad_norm": 1.2294539213180542,
"learning_rate": 4.979136718121551e-05,
"loss": 0.0455,
"step": 1539
},
{
"epoch": 2.93,
"grad_norm": 1.0920627117156982,
"learning_rate": 4.97774808484456e-05,
"loss": 0.0451,
"step": 1540
},
{
"epoch": 2.93,
"grad_norm": 1.2794826030731201,
"learning_rate": 4.97635870169098e-05,
"loss": 0.0656,
"step": 1541
},
{
"epoch": 2.93,
"grad_norm": 1.5503891706466675,
"learning_rate": 4.974968569187603e-05,
"loss": 0.0492,
"step": 1542
},
{
"epoch": 2.93,
"grad_norm": 1.243180274963379,
"learning_rate": 4.973577687861509e-05,
"loss": 0.0493,
"step": 1543
},
{
"epoch": 2.94,
"grad_norm": 1.523897647857666,
"learning_rate": 4.9721860582400596e-05,
"loss": 0.0714,
"step": 1544
},
{
"epoch": 2.94,
"grad_norm": 1.1595388650894165,
"learning_rate": 4.9707936808509005e-05,
"loss": 0.047,
"step": 1545
},
{
"epoch": 2.94,
"grad_norm": 1.6461683511734009,
"learning_rate": 4.96940055622196e-05,
"loss": 0.0506,
"step": 1546
},
{
"epoch": 2.94,
"grad_norm": 1.0963214635849,
"learning_rate": 4.9680066848814514e-05,
"loss": 0.04,
"step": 1547
},
{
"epoch": 2.94,
"grad_norm": 1.2672911882400513,
"learning_rate": 4.9666120673578716e-05,
"loss": 0.0504,
"step": 1548
},
{
"epoch": 2.94,
"grad_norm": 1.6729899644851685,
"learning_rate": 4.965216704179999e-05,
"loss": 0.08,
"step": 1549
},
{
"epoch": 2.95,
"grad_norm": 1.7353764772415161,
"learning_rate": 4.963820595876893e-05,
"loss": 0.0608,
"step": 1550
},
{
"epoch": 2.95,
"grad_norm": 1.408637285232544,
"learning_rate": 4.9624237429778994e-05,
"loss": 0.0512,
"step": 1551
},
{
"epoch": 2.95,
"grad_norm": 1.8854478597640991,
"learning_rate": 4.9610261460126445e-05,
"loss": 0.0756,
"step": 1552
},
{
"epoch": 2.95,
"grad_norm": 1.2744637727737427,
"learning_rate": 4.9596278055110353e-05,
"loss": 0.0509,
"step": 1553
},
{
"epoch": 2.95,
"grad_norm": 0.9074198603630066,
"learning_rate": 4.958228722003263e-05,
"loss": 0.0394,
"step": 1554
},
{
"epoch": 2.96,
"grad_norm": 1.1266288757324219,
"learning_rate": 4.956828896019799e-05,
"loss": 0.0545,
"step": 1555
},
{
"epoch": 2.96,
"grad_norm": 1.4009594917297363,
"learning_rate": 4.9554283280913985e-05,
"loss": 0.0669,
"step": 1556
},
{
"epoch": 2.96,
"grad_norm": 1.1936756372451782,
"learning_rate": 4.954027018749095e-05,
"loss": 0.0524,
"step": 1557
},
{
"epoch": 2.96,
"grad_norm": 3.585347890853882,
"learning_rate": 4.952624968524204e-05,
"loss": 0.0623,
"step": 1558
},
{
"epoch": 2.96,
"grad_norm": 0.9836923480033875,
"learning_rate": 4.951222177948323e-05,
"loss": 0.0293,
"step": 1559
},
{
"epoch": 2.97,
"grad_norm": 1.2324680089950562,
"learning_rate": 4.949818647553329e-05,
"loss": 0.0445,
"step": 1560
},
{
"epoch": 2.97,
"grad_norm": 1.052343487739563,
"learning_rate": 4.948414377871382e-05,
"loss": 0.0546,
"step": 1561
},
{
"epoch": 2.97,
"grad_norm": 1.2935190200805664,
"learning_rate": 4.9470093694349174e-05,
"loss": 0.0369,
"step": 1562
},
{
"epoch": 2.97,
"grad_norm": 1.357293963432312,
"learning_rate": 4.9456036227766564e-05,
"loss": 0.0662,
"step": 1563
},
{
"epoch": 2.97,
"grad_norm": 1.2057098150253296,
"learning_rate": 4.9441971384295954e-05,
"loss": 0.0616,
"step": 1564
},
{
"epoch": 2.98,
"grad_norm": 1.3598567247390747,
"learning_rate": 4.942789916927013e-05,
"loss": 0.0458,
"step": 1565
},
{
"epoch": 2.98,
"grad_norm": 1.2794535160064697,
"learning_rate": 4.941381958802467e-05,
"loss": 0.0441,
"step": 1566
},
{
"epoch": 2.98,
"grad_norm": 1.3057520389556885,
"learning_rate": 4.939973264589795e-05,
"loss": 0.055,
"step": 1567
},
{
"epoch": 2.98,
"grad_norm": 1.6847010850906372,
"learning_rate": 4.938563834823111e-05,
"loss": 0.0652,
"step": 1568
},
{
"epoch": 2.98,
"grad_norm": 1.5347710847854614,
"learning_rate": 4.937153670036811e-05,
"loss": 0.0599,
"step": 1569
},
{
"epoch": 2.98,
"grad_norm": 1.0964845418930054,
"learning_rate": 4.93574277076557e-05,
"loss": 0.0388,
"step": 1570
},
{
"epoch": 2.99,
"grad_norm": 1.3546236753463745,
"learning_rate": 4.934331137544337e-05,
"loss": 0.0698,
"step": 1571
},
{
"epoch": 2.99,
"grad_norm": 1.2396783828735352,
"learning_rate": 4.932918770908345e-05,
"loss": 0.0435,
"step": 1572
},
{
"epoch": 2.99,
"grad_norm": 1.5094010829925537,
"learning_rate": 4.9315056713931014e-05,
"loss": 0.0628,
"step": 1573
},
{
"epoch": 2.99,
"grad_norm": 1.2919801473617554,
"learning_rate": 4.930091839534392e-05,
"loss": 0.0516,
"step": 1574
},
{
"epoch": 2.99,
"grad_norm": 1.116053581237793,
"learning_rate": 4.9286772758682804e-05,
"loss": 0.0442,
"step": 1575
},
{
"epoch": 3.0,
"grad_norm": 1.0668492317199707,
"learning_rate": 4.92726198093111e-05,
"loss": 0.0343,
"step": 1576
},
{
"epoch": 3.0,
"grad_norm": 1.250438928604126,
"learning_rate": 4.925845955259497e-05,
"loss": 0.0379,
"step": 1577
},
{
"epoch": 3.0,
"grad_norm": 1.3699959516525269,
"learning_rate": 4.9244291993903386e-05,
"loss": 0.0466,
"step": 1578
},
{
"epoch": 3.0,
"grad_norm": 0.7261015176773071,
"learning_rate": 4.9230117138608066e-05,
"loss": 0.0305,
"step": 1579
},
{
"epoch": 3.0,
"grad_norm": 0.9778686761856079,
"learning_rate": 4.9215934992083515e-05,
"loss": 0.0356,
"step": 1580
},
{
"epoch": 3.01,
"grad_norm": 1.525650978088379,
"learning_rate": 4.920174555970698e-05,
"loss": 0.0402,
"step": 1581
},
{
"epoch": 3.01,
"grad_norm": 1.1927603483200073,
"learning_rate": 4.918754884685848e-05,
"loss": 0.0453,
"step": 1582
},
{
"epoch": 3.01,
"grad_norm": 1.0774887800216675,
"learning_rate": 4.91733448589208e-05,
"loss": 0.0437,
"step": 1583
},
{
"epoch": 3.01,
"grad_norm": 1.146826982498169,
"learning_rate": 4.915913360127947e-05,
"loss": 0.0271,
"step": 1584
},
{
"epoch": 3.01,
"grad_norm": 0.9338875412940979,
"learning_rate": 4.91449150793228e-05,
"loss": 0.0258,
"step": 1585
},
{
"epoch": 3.02,
"grad_norm": 0.6554346084594727,
"learning_rate": 4.9130689298441824e-05,
"loss": 0.0281,
"step": 1586
},
{
"epoch": 3.02,
"grad_norm": 0.9850267171859741,
"learning_rate": 4.911645626403035e-05,
"loss": 0.0291,
"step": 1587
},
{
"epoch": 3.02,
"grad_norm": 0.9125579595565796,
"learning_rate": 4.9102215981484926e-05,
"loss": 0.0416,
"step": 1588
},
{
"epoch": 3.02,
"grad_norm": 1.1733673810958862,
"learning_rate": 4.908796845620485e-05,
"loss": 0.0322,
"step": 1589
},
{
"epoch": 3.02,
"grad_norm": 1.3528740406036377,
"learning_rate": 4.9073713693592176e-05,
"loss": 0.036,
"step": 1590
},
{
"epoch": 3.02,
"grad_norm": 1.1316100358963013,
"learning_rate": 4.905945169905169e-05,
"loss": 0.05,
"step": 1591
},
{
"epoch": 3.03,
"grad_norm": 0.9560241103172302,
"learning_rate": 4.904518247799092e-05,
"loss": 0.0225,
"step": 1592
},
{
"epoch": 3.03,
"grad_norm": 0.9647613167762756,
"learning_rate": 4.903090603582014e-05,
"loss": 0.0275,
"step": 1593
},
{
"epoch": 3.03,
"grad_norm": 1.0331250429153442,
"learning_rate": 4.9016622377952363e-05,
"loss": 0.035,
"step": 1594
},
{
"epoch": 3.03,
"grad_norm": 0.9462150931358337,
"learning_rate": 4.900233150980333e-05,
"loss": 0.0188,
"step": 1595
},
{
"epoch": 3.03,
"grad_norm": 1.1495990753173828,
"learning_rate": 4.898803343679152e-05,
"loss": 0.0497,
"step": 1596
},
{
"epoch": 3.04,
"grad_norm": 1.5427407026290894,
"learning_rate": 4.897372816433816e-05,
"loss": 0.0454,
"step": 1597
},
{
"epoch": 3.04,
"grad_norm": 1.6270979642868042,
"learning_rate": 4.8959415697867164e-05,
"loss": 0.0329,
"step": 1598
},
{
"epoch": 3.04,
"grad_norm": 1.1162407398223877,
"learning_rate": 4.8945096042805226e-05,
"loss": 0.0358,
"step": 1599
},
{
"epoch": 3.04,
"grad_norm": 1.1508558988571167,
"learning_rate": 4.893076920458173e-05,
"loss": 0.0346,
"step": 1600
},
{
"epoch": 3.04,
"eval_blimp_filtered_avg": 0.7285074626865672,
"eval_blimp_filtered_std": 0.004915998735763533,
"step": 1600
},
{
"epoch": 3.04,
"eval_blimp_supplement_avg": 0.7758620689655172,
"eval_blimp_supplement_std": 0.018135803405118965,
"step": 1600
},
{
"epoch": 3.04,
"eval_vqa_filtered_avg": 0.41,
"eval_vqa_filtered_std": 0.049431107042371025,
"step": 1600
},
{
"epoch": 3.04,
"eval_winoground_filtered_avg": 0.52,
"eval_winoground_filtered_std": 0.05021167315686779,
"step": 1600
},
{
"epoch": 3.04,
"grad_norm": 1.0557070970535278,
"learning_rate": 4.891643518862878e-05,
"loss": 0.0282,
"step": 1601
},
{
"epoch": 3.05,
"grad_norm": 1.0456703901290894,
"learning_rate": 4.890209400038124e-05,
"loss": 0.04,
"step": 1602
},
{
"epoch": 3.05,
"grad_norm": 1.1515144109725952,
"learning_rate": 4.888774564527665e-05,
"loss": 0.0501,
"step": 1603
},
{
"epoch": 3.05,
"grad_norm": 0.854369044303894,
"learning_rate": 4.8873390128755294e-05,
"loss": 0.04,
"step": 1604
},
{
"epoch": 3.05,
"grad_norm": 1.3028532266616821,
"learning_rate": 4.885902745626017e-05,
"loss": 0.0516,
"step": 1605
},
{
"epoch": 3.05,
"grad_norm": 1.0903302431106567,
"learning_rate": 4.8844657633236965e-05,
"loss": 0.0469,
"step": 1606
},
{
"epoch": 3.06,
"grad_norm": 1.0096029043197632,
"learning_rate": 4.8830280665134103e-05,
"loss": 0.0231,
"step": 1607
},
{
"epoch": 3.06,
"grad_norm": 1.4321801662445068,
"learning_rate": 4.881589655740272e-05,
"loss": 0.0308,
"step": 1608
},
{
"epoch": 3.06,
"grad_norm": 1.1996229887008667,
"learning_rate": 4.8801505315496636e-05,
"loss": 0.0315,
"step": 1609
},
{
"epoch": 3.06,
"grad_norm": 1.2638792991638184,
"learning_rate": 4.878710694487239e-05,
"loss": 0.0554,
"step": 1610
},
{
"epoch": 3.06,
"grad_norm": 0.8024887442588806,
"learning_rate": 4.877270145098922e-05,
"loss": 0.0205,
"step": 1611
},
{
"epoch": 3.06,
"grad_norm": 1.1552773714065552,
"learning_rate": 4.875828883930907e-05,
"loss": 0.0787,
"step": 1612
},
{
"epoch": 3.07,
"grad_norm": 1.411600112915039,
"learning_rate": 4.874386911529658e-05,
"loss": 0.0404,
"step": 1613
},
{
"epoch": 3.07,
"grad_norm": 0.8389589190483093,
"learning_rate": 4.8729442284419084e-05,
"loss": 0.0323,
"step": 1614
},
{
"epoch": 3.07,
"grad_norm": 0.8246679902076721,
"learning_rate": 4.8715008352146615e-05,
"loss": 0.0281,
"step": 1615
},
{
"epoch": 3.07,
"grad_norm": 1.219714641571045,
"learning_rate": 4.870056732395189e-05,
"loss": 0.0448,
"step": 1616
},
{
"epoch": 3.07,
"grad_norm": 0.6665266156196594,
"learning_rate": 4.8686119205310324e-05,
"loss": 0.0133,
"step": 1617
},
{
"epoch": 3.08,
"grad_norm": 1.150477409362793,
"learning_rate": 4.867166400170003e-05,
"loss": 0.0359,
"step": 1618
},
{
"epoch": 3.08,
"grad_norm": 1.0337445735931396,
"learning_rate": 4.865720171860179e-05,
"loss": 0.0391,
"step": 1619
},
{
"epoch": 3.08,
"grad_norm": 1.1228854656219482,
"learning_rate": 4.8642732361499074e-05,
"loss": 0.044,
"step": 1620
},
{
"epoch": 3.08,
"grad_norm": 1.2126796245574951,
"learning_rate": 4.862825593587802e-05,
"loss": 0.0276,
"step": 1621
},
{
"epoch": 3.08,
"grad_norm": 1.208730936050415,
"learning_rate": 4.86137724472275e-05,
"loss": 0.0419,
"step": 1622
},
{
"epoch": 3.09,
"grad_norm": 1.0781577825546265,
"learning_rate": 4.8599281901039004e-05,
"loss": 0.0375,
"step": 1623
},
{
"epoch": 3.09,
"grad_norm": 0.9868907928466797,
"learning_rate": 4.858478430280673e-05,
"loss": 0.0259,
"step": 1624
},
{
"epoch": 3.09,
"grad_norm": 0.6946584582328796,
"learning_rate": 4.857027965802753e-05,
"loss": 0.0195,
"step": 1625
},
{
"epoch": 3.09,
"grad_norm": 2.4690189361572266,
"learning_rate": 4.855576797220094e-05,
"loss": 0.065,
"step": 1626
},
{
"epoch": 3.09,
"grad_norm": 1.1611250638961792,
"learning_rate": 4.8541249250829166e-05,
"loss": 0.0302,
"step": 1627
},
{
"epoch": 3.1,
"grad_norm": 1.561187982559204,
"learning_rate": 4.85267234994171e-05,
"loss": 0.0374,
"step": 1628
},
{
"epoch": 3.1,
"grad_norm": 1.1392189264297485,
"learning_rate": 4.8512190723472245e-05,
"loss": 0.0576,
"step": 1629
},
{
"epoch": 3.1,
"grad_norm": 1.045131802558899,
"learning_rate": 4.8497650928504816e-05,
"loss": 0.0279,
"step": 1630
},
{
"epoch": 3.1,
"grad_norm": 0.8496038317680359,
"learning_rate": 4.848310412002768e-05,
"loss": 0.0283,
"step": 1631
},
{
"epoch": 3.1,
"grad_norm": 1.0137355327606201,
"learning_rate": 4.846855030355634e-05,
"loss": 0.0329,
"step": 1632
},
{
"epoch": 3.1,
"grad_norm": 1.435897946357727,
"learning_rate": 4.8453989484609e-05,
"loss": 0.0355,
"step": 1633
},
{
"epoch": 3.11,
"grad_norm": 1.1362143754959106,
"learning_rate": 4.8439421668706484e-05,
"loss": 0.0467,
"step": 1634
},
{
"epoch": 3.11,
"grad_norm": 0.9589322805404663,
"learning_rate": 4.842484686137227e-05,
"loss": 0.0439,
"step": 1635
},
{
"epoch": 3.11,
"grad_norm": 1.0431162118911743,
"learning_rate": 4.84102650681325e-05,
"loss": 0.0291,
"step": 1636
},
{
"epoch": 3.11,
"grad_norm": 1.0131702423095703,
"learning_rate": 4.839567629451596e-05,
"loss": 0.0261,
"step": 1637
},
{
"epoch": 3.11,
"grad_norm": 0.9963091015815735,
"learning_rate": 4.838108054605409e-05,
"loss": 0.0248,
"step": 1638
},
{
"epoch": 3.12,
"grad_norm": 0.8630114793777466,
"learning_rate": 4.836647782828095e-05,
"loss": 0.0187,
"step": 1639
},
{
"epoch": 3.12,
"grad_norm": 0.9646517038345337,
"learning_rate": 4.8351868146733265e-05,
"loss": 0.0223,
"step": 1640
},
{
"epoch": 3.12,
"grad_norm": 0.9125486612319946,
"learning_rate": 4.833725150695041e-05,
"loss": 0.0329,
"step": 1641
},
{
"epoch": 3.12,
"grad_norm": 0.9489634037017822,
"learning_rate": 4.8322627914474365e-05,
"loss": 0.0224,
"step": 1642
},
{
"epoch": 3.12,
"grad_norm": 1.033660650253296,
"learning_rate": 4.8307997374849766e-05,
"loss": 0.033,
"step": 1643
},
{
"epoch": 3.13,
"grad_norm": 1.0925120115280151,
"learning_rate": 4.8293359893623895e-05,
"loss": 0.0294,
"step": 1644
},
{
"epoch": 3.13,
"grad_norm": 1.0587942600250244,
"learning_rate": 4.827871547634664e-05,
"loss": 0.0268,
"step": 1645
},
{
"epoch": 3.13,
"grad_norm": 1.4853657484054565,
"learning_rate": 4.826406412857053e-05,
"loss": 0.0606,
"step": 1646
},
{
"epoch": 3.13,
"grad_norm": 2.1428487300872803,
"learning_rate": 4.824940585585073e-05,
"loss": 0.0275,
"step": 1647
},
{
"epoch": 3.13,
"grad_norm": 1.174332857131958,
"learning_rate": 4.823474066374502e-05,
"loss": 0.039,
"step": 1648
},
{
"epoch": 3.13,
"grad_norm": 1.4633725881576538,
"learning_rate": 4.822006855781381e-05,
"loss": 0.0498,
"step": 1649
},
{
"epoch": 3.14,
"grad_norm": 0.982125461101532,
"learning_rate": 4.820538954362012e-05,
"loss": 0.027,
"step": 1650
},
{
"epoch": 3.14,
"grad_norm": 1.6673634052276611,
"learning_rate": 4.819070362672961e-05,
"loss": 0.0463,
"step": 1651
},
{
"epoch": 3.14,
"grad_norm": 1.2644197940826416,
"learning_rate": 4.817601081271054e-05,
"loss": 0.0395,
"step": 1652
},
{
"epoch": 3.14,
"grad_norm": 1.5220816135406494,
"learning_rate": 4.8161311107133784e-05,
"loss": 0.0266,
"step": 1653
},
{
"epoch": 3.14,
"grad_norm": 1.1871048212051392,
"learning_rate": 4.8146604515572836e-05,
"loss": 0.0461,
"step": 1654
},
{
"epoch": 3.15,
"grad_norm": 1.0855940580368042,
"learning_rate": 4.813189104360381e-05,
"loss": 0.0267,
"step": 1655
},
{
"epoch": 3.15,
"grad_norm": 1.1501593589782715,
"learning_rate": 4.81171706968054e-05,
"loss": 0.0549,
"step": 1656
},
{
"epoch": 3.15,
"grad_norm": 0.8133015632629395,
"learning_rate": 4.810244348075894e-05,
"loss": 0.0124,
"step": 1657
},
{
"epoch": 3.15,
"grad_norm": 0.57613205909729,
"learning_rate": 4.808770940104836e-05,
"loss": 0.0204,
"step": 1658
},
{
"epoch": 3.15,
"grad_norm": 0.9837971329689026,
"learning_rate": 4.807296846326017e-05,
"loss": 0.0421,
"step": 1659
},
{
"epoch": 3.16,
"grad_norm": 4.073154449462891,
"learning_rate": 4.805822067298351e-05,
"loss": 0.0476,
"step": 1660
},
{
"epoch": 3.16,
"grad_norm": 1.3084887266159058,
"learning_rate": 4.8043466035810095e-05,
"loss": 0.0455,
"step": 1661
},
{
"epoch": 3.16,
"grad_norm": 1.5498074293136597,
"learning_rate": 4.802870455733425e-05,
"loss": 0.0341,
"step": 1662
},
{
"epoch": 3.16,
"grad_norm": 1.3016843795776367,
"learning_rate": 4.801393624315288e-05,
"loss": 0.0268,
"step": 1663
},
{
"epoch": 3.16,
"grad_norm": 0.8876961469650269,
"learning_rate": 4.79991610988655e-05,
"loss": 0.0372,
"step": 1664
},
{
"epoch": 3.17,
"grad_norm": 1.1805915832519531,
"learning_rate": 4.798437913007422e-05,
"loss": 0.0376,
"step": 1665
},
{
"epoch": 3.17,
"grad_norm": 1.4229131937026978,
"learning_rate": 4.7969590342383695e-05,
"loss": 0.0452,
"step": 1666
},
{
"epoch": 3.17,
"grad_norm": 1.4386954307556152,
"learning_rate": 4.795479474140122e-05,
"loss": 0.0339,
"step": 1667
},
{
"epoch": 3.17,
"grad_norm": 1.1300697326660156,
"learning_rate": 4.793999233273663e-05,
"loss": 0.0376,
"step": 1668
},
{
"epoch": 3.17,
"grad_norm": 0.6204638481140137,
"learning_rate": 4.7925183122002363e-05,
"loss": 0.0353,
"step": 1669
},
{
"epoch": 3.17,
"grad_norm": 0.7379445433616638,
"learning_rate": 4.791036711481343e-05,
"loss": 0.0222,
"step": 1670
},
{
"epoch": 3.18,
"grad_norm": 1.2310633659362793,
"learning_rate": 4.789554431678744e-05,
"loss": 0.0298,
"step": 1671
},
{
"epoch": 3.18,
"grad_norm": 0.9455915093421936,
"learning_rate": 4.788071473354453e-05,
"loss": 0.0271,
"step": 1672
},
{
"epoch": 3.18,
"grad_norm": 0.6749678254127502,
"learning_rate": 4.7865878370707444e-05,
"loss": 0.0244,
"step": 1673
},
{
"epoch": 3.18,
"grad_norm": 1.4911385774612427,
"learning_rate": 4.7851035233901496e-05,
"loss": 0.0411,
"step": 1674
},
{
"epoch": 3.18,
"grad_norm": 1.0540554523468018,
"learning_rate": 4.783618532875457e-05,
"loss": 0.0278,
"step": 1675
},
{
"epoch": 3.19,
"grad_norm": 1.7726151943206787,
"learning_rate": 4.782132866089708e-05,
"loss": 0.0497,
"step": 1676
},
{
"epoch": 3.19,
"grad_norm": 1.4795124530792236,
"learning_rate": 4.780646523596206e-05,
"loss": 0.0458,
"step": 1677
},
{
"epoch": 3.19,
"grad_norm": 1.0853097438812256,
"learning_rate": 4.7791595059585066e-05,
"loss": 0.0464,
"step": 1678
},
{
"epoch": 3.19,
"grad_norm": 0.8567585349082947,
"learning_rate": 4.777671813740422e-05,
"loss": 0.0248,
"step": 1679
},
{
"epoch": 3.19,
"grad_norm": 0.8010883331298828,
"learning_rate": 4.776183447506024e-05,
"loss": 0.0323,
"step": 1680
},
{
"epoch": 3.2,
"grad_norm": 1.9582961797714233,
"learning_rate": 4.774694407819632e-05,
"loss": 0.0604,
"step": 1681
},
{
"epoch": 3.2,
"grad_norm": 0.8874621391296387,
"learning_rate": 4.773204695245829e-05,
"loss": 0.0293,
"step": 1682
},
{
"epoch": 3.2,
"grad_norm": 0.8441827297210693,
"learning_rate": 4.7717143103494485e-05,
"loss": 0.0341,
"step": 1683
},
{
"epoch": 3.2,
"grad_norm": 1.0981353521347046,
"learning_rate": 4.7702232536955794e-05,
"loss": 0.0394,
"step": 1684
},
{
"epoch": 3.2,
"grad_norm": 0.9108256101608276,
"learning_rate": 4.768731525849567e-05,
"loss": 0.0254,
"step": 1685
},
{
"epoch": 3.21,
"grad_norm": 1.0528583526611328,
"learning_rate": 4.767239127377009e-05,
"loss": 0.0337,
"step": 1686
},
{
"epoch": 3.21,
"grad_norm": 1.2546643018722534,
"learning_rate": 4.7657460588437596e-05,
"loss": 0.051,
"step": 1687
},
{
"epoch": 3.21,
"grad_norm": 1.630698800086975,
"learning_rate": 4.7642523208159254e-05,
"loss": 0.0276,
"step": 1688
},
{
"epoch": 3.21,
"grad_norm": 1.235308289527893,
"learning_rate": 4.762757913859867e-05,
"loss": 0.0321,
"step": 1689
},
{
"epoch": 3.21,
"grad_norm": 1.7361657619476318,
"learning_rate": 4.761262838542198e-05,
"loss": 0.038,
"step": 1690
},
{
"epoch": 3.21,
"grad_norm": 1.1773277521133423,
"learning_rate": 4.759767095429789e-05,
"loss": 0.0366,
"step": 1691
},
{
"epoch": 3.22,
"grad_norm": 1.512444019317627,
"learning_rate": 4.758270685089758e-05,
"loss": 0.0598,
"step": 1692
},
{
"epoch": 3.22,
"grad_norm": 1.1710867881774902,
"learning_rate": 4.756773608089481e-05,
"loss": 0.0388,
"step": 1693
},
{
"epoch": 3.22,
"grad_norm": 1.6597633361816406,
"learning_rate": 4.755275864996586e-05,
"loss": 0.048,
"step": 1694
},
{
"epoch": 3.22,
"grad_norm": 1.2361621856689453,
"learning_rate": 4.753777456378949e-05,
"loss": 0.0315,
"step": 1695
},
{
"epoch": 3.22,
"grad_norm": 1.1621630191802979,
"learning_rate": 4.7522783828047055e-05,
"loss": 0.0407,
"step": 1696
},
{
"epoch": 3.23,
"grad_norm": 0.6419005393981934,
"learning_rate": 4.750778644842237e-05,
"loss": 0.0178,
"step": 1697
},
{
"epoch": 3.23,
"grad_norm": 1.3872300386428833,
"learning_rate": 4.7492782430601794e-05,
"loss": 0.0407,
"step": 1698
},
{
"epoch": 3.23,
"grad_norm": 0.9001232385635376,
"learning_rate": 4.747777178027421e-05,
"loss": 0.019,
"step": 1699
},
{
"epoch": 3.23,
"grad_norm": 1.4317269325256348,
"learning_rate": 4.746275450313101e-05,
"loss": 0.0523,
"step": 1700
},
{
"epoch": 3.23,
"eval_blimp_filtered_avg": 0.6932835820895522,
"eval_blimp_filtered_std": 0.00496369106882512,
"step": 1700
},
{
"epoch": 3.23,
"eval_blimp_supplement_avg": 0.7650862068965517,
"eval_blimp_supplement_std": 0.018646179777208408,
"step": 1700
},
{
"epoch": 3.23,
"eval_vqa_filtered_avg": 0.34,
"eval_vqa_filtered_std": 0.04760952285695235,
"step": 1700
},
{
"epoch": 3.23,
"eval_winoground_filtered_avg": 0.49,
"eval_winoground_filtered_std": 0.05024183937956912,
"step": 1700
},
{
"epoch": 3.23,
"grad_norm": 0.9339033961296082,
"learning_rate": 4.744773060486609e-05,
"loss": 0.0264,
"step": 1701
},
{
"epoch": 3.24,
"grad_norm": 1.6355280876159668,
"learning_rate": 4.743270009117586e-05,
"loss": 0.0561,
"step": 1702
},
{
"epoch": 3.24,
"grad_norm": 0.9683668613433838,
"learning_rate": 4.7417662967759247e-05,
"loss": 0.0303,
"step": 1703
},
{
"epoch": 3.24,
"grad_norm": 1.2073055505752563,
"learning_rate": 4.740261924031768e-05,
"loss": 0.0248,
"step": 1704
},
{
"epoch": 3.24,
"grad_norm": 1.995705485343933,
"learning_rate": 4.738756891455509e-05,
"loss": 0.0285,
"step": 1705
},
{
"epoch": 3.24,
"grad_norm": 1.2198842763900757,
"learning_rate": 4.73725119961779e-05,
"loss": 0.0422,
"step": 1706
},
{
"epoch": 3.25,
"grad_norm": 0.5957209467887878,
"learning_rate": 4.735744849089506e-05,
"loss": 0.0175,
"step": 1707
},
{
"epoch": 3.25,
"grad_norm": 1.545203447341919,
"learning_rate": 4.734237840441798e-05,
"loss": 0.0666,
"step": 1708
},
{
"epoch": 3.25,
"grad_norm": 4.228770732879639,
"learning_rate": 4.73273017424606e-05,
"loss": 0.0446,
"step": 1709
},
{
"epoch": 3.25,
"grad_norm": 1.2417970895767212,
"learning_rate": 4.731221851073934e-05,
"loss": 0.0403,
"step": 1710
},
{
"epoch": 3.25,
"grad_norm": 1.0373398065567017,
"learning_rate": 4.729712871497309e-05,
"loss": 0.0422,
"step": 1711
},
{
"epoch": 3.25,
"grad_norm": 1.1415021419525146,
"learning_rate": 4.728203236088327e-05,
"loss": 0.0439,
"step": 1712
},
{
"epoch": 3.26,
"grad_norm": 1.1209381818771362,
"learning_rate": 4.726692945419376e-05,
"loss": 0.0313,
"step": 1713
},
{
"epoch": 3.26,
"grad_norm": 1.0905299186706543,
"learning_rate": 4.725182000063093e-05,
"loss": 0.0254,
"step": 1714
},
{
"epoch": 3.26,
"grad_norm": 0.9422305226325989,
"learning_rate": 4.723670400592361e-05,
"loss": 0.0223,
"step": 1715
},
{
"epoch": 3.26,
"grad_norm": 1.2118544578552246,
"learning_rate": 4.7221581475803164e-05,
"loss": 0.0558,
"step": 1716
},
{
"epoch": 3.26,
"grad_norm": 1.4376803636550903,
"learning_rate": 4.720645241600339e-05,
"loss": 0.0562,
"step": 1717
},
{
"epoch": 3.27,
"grad_norm": 1.1702667474746704,
"learning_rate": 4.7191316832260564e-05,
"loss": 0.027,
"step": 1718
},
{
"epoch": 3.27,
"grad_norm": 1.0566704273223877,
"learning_rate": 4.717617473031346e-05,
"loss": 0.0556,
"step": 1719
},
{
"epoch": 3.27,
"grad_norm": 0.8927348256111145,
"learning_rate": 4.716102611590329e-05,
"loss": 0.0365,
"step": 1720
},
{
"epoch": 3.27,
"grad_norm": 1.032292366027832,
"learning_rate": 4.714587099477377e-05,
"loss": 0.0291,
"step": 1721
},
{
"epoch": 3.27,
"grad_norm": 1.4841428995132446,
"learning_rate": 4.713070937267106e-05,
"loss": 0.0413,
"step": 1722
},
{
"epoch": 3.28,
"grad_norm": 1.1563156843185425,
"learning_rate": 4.71155412553438e-05,
"loss": 0.0375,
"step": 1723
},
{
"epoch": 3.28,
"grad_norm": 1.297371745109558,
"learning_rate": 4.710036664854308e-05,
"loss": 0.0344,
"step": 1724
},
{
"epoch": 3.28,
"grad_norm": 0.9905734062194824,
"learning_rate": 4.7085185558022446e-05,
"loss": 0.0245,
"step": 1725
},
{
"epoch": 3.28,
"grad_norm": 1.4286359548568726,
"learning_rate": 4.706999798953792e-05,
"loss": 0.038,
"step": 1726
},
{
"epoch": 3.28,
"grad_norm": 0.9452222585678101,
"learning_rate": 4.7054803948847965e-05,
"loss": 0.0251,
"step": 1727
},
{
"epoch": 3.29,
"grad_norm": 0.9268828630447388,
"learning_rate": 4.703960344171352e-05,
"loss": 0.033,
"step": 1728
},
{
"epoch": 3.29,
"grad_norm": 0.7375654578208923,
"learning_rate": 4.702439647389794e-05,
"loss": 0.0267,
"step": 1729
},
{
"epoch": 3.29,
"grad_norm": 1.0199426412582397,
"learning_rate": 4.700918305116706e-05,
"loss": 0.0397,
"step": 1730
},
{
"epoch": 3.29,
"grad_norm": 1.587390661239624,
"learning_rate": 4.6993963179289156e-05,
"loss": 0.0567,
"step": 1731
},
{
"epoch": 3.29,
"grad_norm": 0.91463702917099,
"learning_rate": 4.697873686403494e-05,
"loss": 0.0169,
"step": 1732
},
{
"epoch": 3.29,
"grad_norm": 1.0621304512023926,
"learning_rate": 4.696350411117757e-05,
"loss": 0.0317,
"step": 1733
},
{
"epoch": 3.3,
"grad_norm": 1.0592411756515503,
"learning_rate": 4.694826492649266e-05,
"loss": 0.0389,
"step": 1734
},
{
"epoch": 3.3,
"grad_norm": 1.0435583591461182,
"learning_rate": 4.693301931575824e-05,
"loss": 0.0325,
"step": 1735
},
{
"epoch": 3.3,
"grad_norm": 1.4165093898773193,
"learning_rate": 4.691776728475478e-05,
"loss": 0.0461,
"step": 1736
},
{
"epoch": 3.3,
"grad_norm": 0.9727306962013245,
"learning_rate": 4.690250883926521e-05,
"loss": 0.0309,
"step": 1737
},
{
"epoch": 3.3,
"grad_norm": 1.1539250612258911,
"learning_rate": 4.6887243985074854e-05,
"loss": 0.0286,
"step": 1738
},
{
"epoch": 3.31,
"grad_norm": 1.2665977478027344,
"learning_rate": 4.68719727279715e-05,
"loss": 0.0358,
"step": 1739
},
{
"epoch": 3.31,
"grad_norm": 0.9670816659927368,
"learning_rate": 4.6856695073745346e-05,
"loss": 0.033,
"step": 1740
},
{
"epoch": 3.31,
"grad_norm": 1.0603492259979248,
"learning_rate": 4.6841411028189e-05,
"loss": 0.0312,
"step": 1741
},
{
"epoch": 3.31,
"grad_norm": 0.6751345992088318,
"learning_rate": 4.6826120597097536e-05,
"loss": 0.0231,
"step": 1742
},
{
"epoch": 3.31,
"grad_norm": 0.8814055919647217,
"learning_rate": 4.681082378626841e-05,
"loss": 0.0216,
"step": 1743
},
{
"epoch": 3.32,
"grad_norm": 1.3309929370880127,
"learning_rate": 4.679552060150151e-05,
"loss": 0.0389,
"step": 1744
},
{
"epoch": 3.32,
"grad_norm": 1.1946076154708862,
"learning_rate": 4.678021104859916e-05,
"loss": 0.034,
"step": 1745
},
{
"epoch": 3.32,
"grad_norm": 1.378662109375,
"learning_rate": 4.6764895133366066e-05,
"loss": 0.0692,
"step": 1746
},
{
"epoch": 3.32,
"grad_norm": 0.9366088509559631,
"learning_rate": 4.674957286160935e-05,
"loss": 0.027,
"step": 1747
},
{
"epoch": 3.32,
"grad_norm": 0.8396125435829163,
"learning_rate": 4.6734244239138574e-05,
"loss": 0.0126,
"step": 1748
},
{
"epoch": 3.33,
"grad_norm": 1.0137218236923218,
"learning_rate": 4.671890927176569e-05,
"loss": 0.0356,
"step": 1749
},
{
"epoch": 3.33,
"grad_norm": 1.667273998260498,
"learning_rate": 4.670356796530505e-05,
"loss": 0.0305,
"step": 1750
},
{
"epoch": 3.33,
"grad_norm": 1.441938877105713,
"learning_rate": 4.6688220325573416e-05,
"loss": 0.027,
"step": 1751
},
{
"epoch": 3.33,
"grad_norm": 1.963052749633789,
"learning_rate": 4.667286635838994e-05,
"loss": 0.0371,
"step": 1752
},
{
"epoch": 3.33,
"grad_norm": 1.5050019025802612,
"learning_rate": 4.665750606957619e-05,
"loss": 0.0397,
"step": 1753
},
{
"epoch": 3.33,
"grad_norm": 1.2591651678085327,
"learning_rate": 4.664213946495611e-05,
"loss": 0.06,
"step": 1754
},
{
"epoch": 3.34,
"grad_norm": 1.4289031028747559,
"learning_rate": 4.662676655035608e-05,
"loss": 0.0341,
"step": 1755
},
{
"epoch": 3.34,
"grad_norm": 0.8640409111976624,
"learning_rate": 4.661138733160483e-05,
"loss": 0.0296,
"step": 1756
},
{
"epoch": 3.34,
"grad_norm": 0.930654764175415,
"learning_rate": 4.6596001814533494e-05,
"loss": 0.0241,
"step": 1757
},
{
"epoch": 3.34,
"grad_norm": 1.390769600868225,
"learning_rate": 4.65806100049756e-05,
"loss": 0.0429,
"step": 1758
},
{
"epoch": 3.34,
"grad_norm": 0.8745943307876587,
"learning_rate": 4.656521190876704e-05,
"loss": 0.0277,
"step": 1759
},
{
"epoch": 3.35,
"grad_norm": 1.508611798286438,
"learning_rate": 4.654980753174613e-05,
"loss": 0.0432,
"step": 1760
},
{
"epoch": 3.35,
"grad_norm": 1.0874576568603516,
"learning_rate": 4.653439687975353e-05,
"loss": 0.0336,
"step": 1761
},
{
"epoch": 3.35,
"grad_norm": 0.9956321716308594,
"learning_rate": 4.65189799586323e-05,
"loss": 0.0279,
"step": 1762
},
{
"epoch": 3.35,
"grad_norm": 1.5806199312210083,
"learning_rate": 4.6503556774227856e-05,
"loss": 0.0406,
"step": 1763
},
{
"epoch": 3.35,
"grad_norm": 1.134081482887268,
"learning_rate": 4.6488127332388014e-05,
"loss": 0.0364,
"step": 1764
},
{
"epoch": 3.36,
"grad_norm": 1.0967165231704712,
"learning_rate": 4.6472691638962965e-05,
"loss": 0.03,
"step": 1765
},
{
"epoch": 3.36,
"grad_norm": 1.407957673072815,
"learning_rate": 4.645724969980524e-05,
"loss": 0.0514,
"step": 1766
},
{
"epoch": 3.36,
"grad_norm": 1.0269697904586792,
"learning_rate": 4.644180152076974e-05,
"loss": 0.032,
"step": 1767
},
{
"epoch": 3.36,
"grad_norm": 1.4709781408309937,
"learning_rate": 4.6426347107713787e-05,
"loss": 0.0441,
"step": 1768
},
{
"epoch": 3.36,
"grad_norm": 1.1574596166610718,
"learning_rate": 4.6410886466497e-05,
"loss": 0.0402,
"step": 1769
},
{
"epoch": 3.37,
"grad_norm": 1.2102223634719849,
"learning_rate": 4.6395419602981374e-05,
"loss": 0.0314,
"step": 1770
},
{
"epoch": 3.37,
"grad_norm": 1.1190632581710815,
"learning_rate": 4.6379946523031304e-05,
"loss": 0.0278,
"step": 1771
},
{
"epoch": 3.37,
"grad_norm": 1.5308424234390259,
"learning_rate": 4.6364467232513494e-05,
"loss": 0.0406,
"step": 1772
},
{
"epoch": 3.37,
"grad_norm": 1.606535792350769,
"learning_rate": 4.634898173729704e-05,
"loss": 0.0458,
"step": 1773
},
{
"epoch": 3.37,
"grad_norm": 1.0908801555633545,
"learning_rate": 4.633349004325334e-05,
"loss": 0.0423,
"step": 1774
},
{
"epoch": 3.37,
"grad_norm": 1.0532349348068237,
"learning_rate": 4.631799215625622e-05,
"loss": 0.0448,
"step": 1775
},
{
"epoch": 3.38,
"grad_norm": 1.1966639757156372,
"learning_rate": 4.6302488082181766e-05,
"loss": 0.0464,
"step": 1776
},
{
"epoch": 3.38,
"grad_norm": 1.0183231830596924,
"learning_rate": 4.628697782690847e-05,
"loss": 0.0234,
"step": 1777
},
{
"epoch": 3.38,
"grad_norm": 1.2215787172317505,
"learning_rate": 4.6271461396317154e-05,
"loss": 0.0455,
"step": 1778
},
{
"epoch": 3.38,
"grad_norm": 0.8732157349586487,
"learning_rate": 4.625593879629098e-05,
"loss": 0.0318,
"step": 1779
},
{
"epoch": 3.38,
"grad_norm": 1.2745345830917358,
"learning_rate": 4.624041003271544e-05,
"loss": 0.0538,
"step": 1780
},
{
"epoch": 3.39,
"grad_norm": 0.86319899559021,
"learning_rate": 4.6224875111478376e-05,
"loss": 0.0382,
"step": 1781
},
{
"epoch": 3.39,
"grad_norm": 1.1608169078826904,
"learning_rate": 4.6209334038469944e-05,
"loss": 0.0457,
"step": 1782
},
{
"epoch": 3.39,
"grad_norm": 0.7134714126586914,
"learning_rate": 4.6193786819582656e-05,
"loss": 0.023,
"step": 1783
},
{
"epoch": 3.39,
"grad_norm": 1.1232423782348633,
"learning_rate": 4.617823346071136e-05,
"loss": 0.0353,
"step": 1784
},
{
"epoch": 3.39,
"grad_norm": 0.7866341471672058,
"learning_rate": 4.61626739677532e-05,
"loss": 0.0245,
"step": 1785
},
{
"epoch": 3.4,
"grad_norm": 1.5635771751403809,
"learning_rate": 4.614710834660768e-05,
"loss": 0.0424,
"step": 1786
},
{
"epoch": 3.4,
"grad_norm": 1.1548285484313965,
"learning_rate": 4.6131536603176575e-05,
"loss": 0.0278,
"step": 1787
},
{
"epoch": 3.4,
"grad_norm": 0.9978801012039185,
"learning_rate": 4.611595874336405e-05,
"loss": 0.0598,
"step": 1788
},
{
"epoch": 3.4,
"grad_norm": 1.2230035066604614,
"learning_rate": 4.610037477307656e-05,
"loss": 0.0434,
"step": 1789
},
{
"epoch": 3.4,
"grad_norm": 0.955231249332428,
"learning_rate": 4.6084784698222844e-05,
"loss": 0.0268,
"step": 1790
},
{
"epoch": 3.4,
"grad_norm": 0.9208203554153442,
"learning_rate": 4.6069188524714e-05,
"loss": 0.0238,
"step": 1791
},
{
"epoch": 3.41,
"grad_norm": 0.9514766931533813,
"learning_rate": 4.605358625846343e-05,
"loss": 0.0203,
"step": 1792
},
{
"epoch": 3.41,
"grad_norm": 0.8037521243095398,
"learning_rate": 4.603797790538683e-05,
"loss": 0.0284,
"step": 1793
},
{
"epoch": 3.41,
"grad_norm": 1.0160737037658691,
"learning_rate": 4.602236347140221e-05,
"loss": 0.0254,
"step": 1794
},
{
"epoch": 3.41,
"grad_norm": 1.0170652866363525,
"learning_rate": 4.600674296242989e-05,
"loss": 0.0225,
"step": 1795
},
{
"epoch": 3.41,
"grad_norm": 1.1635174751281738,
"learning_rate": 4.5991116384392514e-05,
"loss": 0.0396,
"step": 1796
},
{
"epoch": 3.42,
"grad_norm": 0.9921762943267822,
"learning_rate": 4.597548374321497e-05,
"loss": 0.0333,
"step": 1797
},
{
"epoch": 3.42,
"grad_norm": 1.8394516706466675,
"learning_rate": 4.5959845044824506e-05,
"loss": 0.0679,
"step": 1798
},
{
"epoch": 3.42,
"grad_norm": 0.7406845688819885,
"learning_rate": 4.594420029515063e-05,
"loss": 0.0172,
"step": 1799
},
{
"epoch": 3.42,
"grad_norm": 1.1034274101257324,
"learning_rate": 4.592854950012515e-05,
"loss": 0.0286,
"step": 1800
},
{
"epoch": 3.42,
"eval_blimp_filtered_avg": 0.7002985074626865,
"eval_blimp_filtered_std": 0.0050718769431210954,
"step": 1800
},
{
"epoch": 3.42,
"eval_blimp_supplement_avg": 0.7521551724137931,
"eval_blimp_supplement_std": 0.01897713204472892,
"step": 1800
},
{
"epoch": 3.42,
"eval_vqa_filtered_avg": 0.37,
"eval_vqa_filtered_std": 0.048523658709391,
"step": 1800
},
{
"epoch": 3.42,
"eval_winoground_filtered_avg": 0.48,
"eval_winoground_filtered_std": 0.05021167315686779,
"step": 1800
},
{
"epoch": 3.42,
"grad_norm": 0.6230857372283936,
"learning_rate": 4.591289266568218e-05,
"loss": 0.0205,
"step": 1801
},
{
"epoch": 3.43,
"grad_norm": 1.364378571510315,
"learning_rate": 4.5897229797758104e-05,
"loss": 0.0495,
"step": 1802
},
{
"epoch": 3.43,
"grad_norm": 1.1185011863708496,
"learning_rate": 4.588156090229162e-05,
"loss": 0.0485,
"step": 1803
},
{
"epoch": 3.43,
"grad_norm": 1.344459891319275,
"learning_rate": 4.5865885985223686e-05,
"loss": 0.0522,
"step": 1804
},
{
"epoch": 3.43,
"grad_norm": 1.0751395225524902,
"learning_rate": 4.5850205052497554e-05,
"loss": 0.0407,
"step": 1805
},
{
"epoch": 3.43,
"grad_norm": 5.028260707855225,
"learning_rate": 4.583451811005875e-05,
"loss": 0.0691,
"step": 1806
},
{
"epoch": 3.44,
"grad_norm": 1.2126531600952148,
"learning_rate": 4.5818825163855106e-05,
"loss": 0.0246,
"step": 1807
},
{
"epoch": 3.44,
"grad_norm": 1.2034071683883667,
"learning_rate": 4.5803126219836674e-05,
"loss": 0.0438,
"step": 1808
},
{
"epoch": 3.44,
"grad_norm": 0.5612877607345581,
"learning_rate": 4.5787421283955855e-05,
"loss": 0.019,
"step": 1809
},
{
"epoch": 3.44,
"grad_norm": 1.3463836908340454,
"learning_rate": 4.5771710362167254e-05,
"loss": 0.0471,
"step": 1810
},
{
"epoch": 3.44,
"grad_norm": 1.6143767833709717,
"learning_rate": 4.575599346042778e-05,
"loss": 0.0513,
"step": 1811
},
{
"epoch": 3.44,
"grad_norm": 1.1454827785491943,
"learning_rate": 4.574027058469661e-05,
"loss": 0.0394,
"step": 1812
},
{
"epoch": 3.45,
"grad_norm": 0.7395484447479248,
"learning_rate": 4.5724541740935163e-05,
"loss": 0.0312,
"step": 1813
},
{
"epoch": 3.45,
"grad_norm": 0.807180643081665,
"learning_rate": 4.570880693510716e-05,
"loss": 0.0233,
"step": 1814
},
{
"epoch": 3.45,
"grad_norm": 1.4437193870544434,
"learning_rate": 4.5693066173178534e-05,
"loss": 0.0338,
"step": 1815
},
{
"epoch": 3.45,
"grad_norm": 1.082722783088684,
"learning_rate": 4.567731946111752e-05,
"loss": 0.0375,
"step": 1816
},
{
"epoch": 3.45,
"grad_norm": 1.340394139289856,
"learning_rate": 4.566156680489458e-05,
"loss": 0.0332,
"step": 1817
},
{
"epoch": 3.46,
"grad_norm": 1.051161289215088,
"learning_rate": 4.564580821048245e-05,
"loss": 0.0369,
"step": 1818
},
{
"epoch": 3.46,
"grad_norm": 1.1002857685089111,
"learning_rate": 4.5630043683856095e-05,
"loss": 0.037,
"step": 1819
},
{
"epoch": 3.46,
"grad_norm": 1.6606191396713257,
"learning_rate": 4.561427323099276e-05,
"loss": 0.0526,
"step": 1820
},
{
"epoch": 3.46,
"grad_norm": 2.6770827770233154,
"learning_rate": 4.55984968578719e-05,
"loss": 0.0485,
"step": 1821
},
{
"epoch": 3.46,
"grad_norm": 0.9643968343734741,
"learning_rate": 4.5582714570475245e-05,
"loss": 0.0243,
"step": 1822
},
{
"epoch": 3.47,
"grad_norm": 1.061124324798584,
"learning_rate": 4.556692637478677e-05,
"loss": 0.0222,
"step": 1823
},
{
"epoch": 3.47,
"grad_norm": 1.0488009452819824,
"learning_rate": 4.555113227679264e-05,
"loss": 0.0259,
"step": 1824
},
{
"epoch": 3.47,
"grad_norm": 1.1751123666763306,
"learning_rate": 4.553533228248132e-05,
"loss": 0.0435,
"step": 1825
},
{
"epoch": 3.47,
"grad_norm": 1.6343899965286255,
"learning_rate": 4.5519526397843475e-05,
"loss": 0.0467,
"step": 1826
},
{
"epoch": 3.47,
"grad_norm": 1.1615169048309326,
"learning_rate": 4.550371462887203e-05,
"loss": 0.0357,
"step": 1827
},
{
"epoch": 3.48,
"grad_norm": 1.3036237955093384,
"learning_rate": 4.54878969815621e-05,
"loss": 0.0406,
"step": 1828
},
{
"epoch": 3.48,
"grad_norm": 1.0059996843338013,
"learning_rate": 4.547207346191106e-05,
"loss": 0.0188,
"step": 1829
},
{
"epoch": 3.48,
"grad_norm": 1.6136471033096313,
"learning_rate": 4.545624407591853e-05,
"loss": 0.0469,
"step": 1830
},
{
"epoch": 3.48,
"grad_norm": 0.8997081518173218,
"learning_rate": 4.54404088295863e-05,
"loss": 0.026,
"step": 1831
},
{
"epoch": 3.48,
"grad_norm": 0.8576598167419434,
"learning_rate": 4.542456772891841e-05,
"loss": 0.0293,
"step": 1832
},
{
"epoch": 3.48,
"grad_norm": 0.8680685758590698,
"learning_rate": 4.5408720779921144e-05,
"loss": 0.0412,
"step": 1833
},
{
"epoch": 3.49,
"grad_norm": 1.0791398286819458,
"learning_rate": 4.5392867988602965e-05,
"loss": 0.0456,
"step": 1834
},
{
"epoch": 3.49,
"grad_norm": 1.201923131942749,
"learning_rate": 4.537700936097456e-05,
"loss": 0.0429,
"step": 1835
},
{
"epoch": 3.49,
"grad_norm": 1.1412664651870728,
"learning_rate": 4.536114490304885e-05,
"loss": 0.0454,
"step": 1836
},
{
"epoch": 3.49,
"grad_norm": 1.3357428312301636,
"learning_rate": 4.534527462084095e-05,
"loss": 0.0474,
"step": 1837
},
{
"epoch": 3.49,
"grad_norm": 2.066304922103882,
"learning_rate": 4.5329398520368174e-05,
"loss": 0.0285,
"step": 1838
},
{
"epoch": 3.5,
"grad_norm": 1.0713777542114258,
"learning_rate": 4.531351660765005e-05,
"loss": 0.0404,
"step": 1839
},
{
"epoch": 3.5,
"grad_norm": 0.9669035077095032,
"learning_rate": 4.5297628888708325e-05,
"loss": 0.0327,
"step": 1840
},
{
"epoch": 3.5,
"grad_norm": 0.9242938160896301,
"learning_rate": 4.528173536956694e-05,
"loss": 0.0292,
"step": 1841
},
{
"epoch": 3.5,
"grad_norm": 0.9377745985984802,
"learning_rate": 4.526583605625201e-05,
"loss": 0.0284,
"step": 1842
},
{
"epoch": 3.5,
"grad_norm": 1.1098551750183105,
"learning_rate": 4.5249930954791887e-05,
"loss": 0.045,
"step": 1843
},
{
"epoch": 3.51,
"grad_norm": 0.7606746554374695,
"learning_rate": 4.5234020071217096e-05,
"loss": 0.0318,
"step": 1844
},
{
"epoch": 3.51,
"grad_norm": 1.3565802574157715,
"learning_rate": 4.5218103411560354e-05,
"loss": 0.0258,
"step": 1845
},
{
"epoch": 3.51,
"grad_norm": 0.9434876441955566,
"learning_rate": 4.520218098185656e-05,
"loss": 0.0155,
"step": 1846
},
{
"epoch": 3.51,
"grad_norm": 0.8103033304214478,
"learning_rate": 4.5186252788142826e-05,
"loss": 0.0347,
"step": 1847
},
{
"epoch": 3.51,
"grad_norm": 0.8523113131523132,
"learning_rate": 4.517031883645842e-05,
"loss": 0.0267,
"step": 1848
},
{
"epoch": 3.52,
"grad_norm": 1.1769499778747559,
"learning_rate": 4.515437913284481e-05,
"loss": 0.0413,
"step": 1849
},
{
"epoch": 3.52,
"grad_norm": 0.9778563380241394,
"learning_rate": 4.513843368334566e-05,
"loss": 0.0344,
"step": 1850
},
{
"epoch": 3.52,
"grad_norm": 1.1736937761306763,
"learning_rate": 4.512248249400678e-05,
"loss": 0.0414,
"step": 1851
},
{
"epoch": 3.52,
"grad_norm": 0.852597177028656,
"learning_rate": 4.510652557087617e-05,
"loss": 0.0414,
"step": 1852
},
{
"epoch": 3.52,
"grad_norm": 1.5087406635284424,
"learning_rate": 4.509056292000402e-05,
"loss": 0.0426,
"step": 1853
},
{
"epoch": 3.52,
"grad_norm": 0.8300724625587463,
"learning_rate": 4.507459454744266e-05,
"loss": 0.0204,
"step": 1854
},
{
"epoch": 3.53,
"grad_norm": 1.3391302824020386,
"learning_rate": 4.5058620459246604e-05,
"loss": 0.0496,
"step": 1855
},
{
"epoch": 3.53,
"grad_norm": 1.3395427465438843,
"learning_rate": 4.504264066147256e-05,
"loss": 0.0446,
"step": 1856
},
{
"epoch": 3.53,
"grad_norm": 0.8964810371398926,
"learning_rate": 4.502665516017936e-05,
"loss": 0.0333,
"step": 1857
},
{
"epoch": 3.53,
"grad_norm": 1.3518861532211304,
"learning_rate": 4.5010663961428024e-05,
"loss": 0.0383,
"step": 1858
},
{
"epoch": 3.53,
"grad_norm": 1.2300162315368652,
"learning_rate": 4.4994667071281706e-05,
"loss": 0.0455,
"step": 1859
},
{
"epoch": 3.54,
"grad_norm": 1.0372908115386963,
"learning_rate": 4.4978664495805753e-05,
"loss": 0.025,
"step": 1860
},
{
"epoch": 3.54,
"grad_norm": 0.9992857575416565,
"learning_rate": 4.496265624106764e-05,
"loss": 0.0252,
"step": 1861
},
{
"epoch": 3.54,
"grad_norm": 1.083809733390808,
"learning_rate": 4.4946642313137004e-05,
"loss": 0.0352,
"step": 1862
},
{
"epoch": 3.54,
"grad_norm": 1.0915288925170898,
"learning_rate": 4.4930622718085645e-05,
"loss": 0.0497,
"step": 1863
},
{
"epoch": 3.54,
"grad_norm": 1.4360071420669556,
"learning_rate": 4.4914597461987494e-05,
"loss": 0.0575,
"step": 1864
},
{
"epoch": 3.55,
"grad_norm": 0.7923672795295715,
"learning_rate": 4.489856655091863e-05,
"loss": 0.0421,
"step": 1865
},
{
"epoch": 3.55,
"grad_norm": 1.0525002479553223,
"learning_rate": 4.4882529990957305e-05,
"loss": 0.0394,
"step": 1866
},
{
"epoch": 3.55,
"grad_norm": 0.7268140316009521,
"learning_rate": 4.486648778818386e-05,
"loss": 0.037,
"step": 1867
},
{
"epoch": 3.55,
"grad_norm": 1.0527366399765015,
"learning_rate": 4.4850439948680824e-05,
"loss": 0.0301,
"step": 1868
},
{
"epoch": 3.55,
"grad_norm": 1.2026972770690918,
"learning_rate": 4.483438647853283e-05,
"loss": 0.0377,
"step": 1869
},
{
"epoch": 3.56,
"grad_norm": 0.9189443588256836,
"learning_rate": 4.481832738382667e-05,
"loss": 0.025,
"step": 1870
},
{
"epoch": 3.56,
"grad_norm": 1.2599409818649292,
"learning_rate": 4.480226267065126e-05,
"loss": 0.0383,
"step": 1871
},
{
"epoch": 3.56,
"grad_norm": 0.880893349647522,
"learning_rate": 4.4786192345097635e-05,
"loss": 0.028,
"step": 1872
},
{
"epoch": 3.56,
"grad_norm": 1.1711302995681763,
"learning_rate": 4.4770116413258976e-05,
"loss": 0.0419,
"step": 1873
},
{
"epoch": 3.56,
"grad_norm": 1.0173710584640503,
"learning_rate": 4.475403488123058e-05,
"loss": 0.0342,
"step": 1874
},
{
"epoch": 3.56,
"grad_norm": 1.1565979719161987,
"learning_rate": 4.473794775510986e-05,
"loss": 0.0341,
"step": 1875
},
{
"epoch": 3.57,
"grad_norm": 1.0757721662521362,
"learning_rate": 4.472185504099638e-05,
"loss": 0.0322,
"step": 1876
},
{
"epoch": 3.57,
"grad_norm": 1.2329587936401367,
"learning_rate": 4.4705756744991784e-05,
"loss": 0.0354,
"step": 1877
},
{
"epoch": 3.57,
"grad_norm": 0.9543439745903015,
"learning_rate": 4.468965287319985e-05,
"loss": 0.0295,
"step": 1878
},
{
"epoch": 3.57,
"grad_norm": 1.02653968334198,
"learning_rate": 4.4673543431726475e-05,
"loss": 0.0272,
"step": 1879
},
{
"epoch": 3.57,
"grad_norm": 0.5178062319755554,
"learning_rate": 4.4657428426679674e-05,
"loss": 0.0175,
"step": 1880
},
{
"epoch": 3.58,
"grad_norm": 1.5902003049850464,
"learning_rate": 4.4641307864169537e-05,
"loss": 0.0479,
"step": 1881
},
{
"epoch": 3.58,
"grad_norm": 0.8545593619346619,
"learning_rate": 4.4625181750308306e-05,
"loss": 0.0427,
"step": 1882
},
{
"epoch": 3.58,
"grad_norm": 0.6384052634239197,
"learning_rate": 4.460905009121029e-05,
"loss": 0.0302,
"step": 1883
},
{
"epoch": 3.58,
"grad_norm": 1.1815310716629028,
"learning_rate": 4.459291289299193e-05,
"loss": 0.0263,
"step": 1884
},
{
"epoch": 3.58,
"grad_norm": 1.224098563194275,
"learning_rate": 4.457677016177174e-05,
"loss": 0.0307,
"step": 1885
},
{
"epoch": 3.59,
"grad_norm": 0.9876607656478882,
"learning_rate": 4.456062190367035e-05,
"loss": 0.0383,
"step": 1886
},
{
"epoch": 3.59,
"grad_norm": 1.0243010520935059,
"learning_rate": 4.45444681248105e-05,
"loss": 0.0433,
"step": 1887
},
{
"epoch": 3.59,
"grad_norm": 1.389990210533142,
"learning_rate": 4.452830883131697e-05,
"loss": 0.0541,
"step": 1888
},
{
"epoch": 3.59,
"grad_norm": 0.8667845726013184,
"learning_rate": 4.45121440293167e-05,
"loss": 0.0247,
"step": 1889
},
{
"epoch": 3.59,
"grad_norm": 1.0832589864730835,
"learning_rate": 4.449597372493868e-05,
"loss": 0.0313,
"step": 1890
},
{
"epoch": 3.6,
"grad_norm": 0.6944572329521179,
"learning_rate": 4.447979792431396e-05,
"loss": 0.0144,
"step": 1891
},
{
"epoch": 3.6,
"grad_norm": 1.1078029870986938,
"learning_rate": 4.4463616633575735e-05,
"loss": 0.0341,
"step": 1892
},
{
"epoch": 3.6,
"grad_norm": 1.065233826637268,
"learning_rate": 4.4447429858859245e-05,
"loss": 0.0323,
"step": 1893
},
{
"epoch": 3.6,
"grad_norm": 1.1714134216308594,
"learning_rate": 4.44312376063018e-05,
"loss": 0.0431,
"step": 1894
},
{
"epoch": 3.6,
"grad_norm": 0.8641266822814941,
"learning_rate": 4.441503988204282e-05,
"loss": 0.0253,
"step": 1895
},
{
"epoch": 3.6,
"grad_norm": 1.0842008590698242,
"learning_rate": 4.439883669222377e-05,
"loss": 0.0265,
"step": 1896
},
{
"epoch": 3.61,
"grad_norm": 1.0234638452529907,
"learning_rate": 4.438262804298822e-05,
"loss": 0.0294,
"step": 1897
},
{
"epoch": 3.61,
"grad_norm": 1.265733242034912,
"learning_rate": 4.436641394048175e-05,
"loss": 0.0213,
"step": 1898
},
{
"epoch": 3.61,
"grad_norm": 0.8639823794364929,
"learning_rate": 4.435019439085209e-05,
"loss": 0.0328,
"step": 1899
},
{
"epoch": 3.61,
"grad_norm": 0.8810359239578247,
"learning_rate": 4.433396940024897e-05,
"loss": 0.0262,
"step": 1900
},
{
"epoch": 3.61,
"eval_blimp_filtered_avg": 0.7122388059701492,
"eval_blimp_filtered_std": 0.004978696139854068,
"step": 1900
},
{
"epoch": 3.61,
"eval_blimp_supplement_avg": 0.7543103448275862,
"eval_blimp_supplement_std": 0.019028402902200935,
"step": 1900
},
{
"epoch": 3.61,
"eval_vqa_filtered_avg": 0.39,
"eval_vqa_filtered_std": 0.04902071300001975,
"step": 1900
},
{
"epoch": 3.61,
"eval_winoground_filtered_avg": 0.47,
"eval_winoground_filtered_std": 0.05016135580465919,
"step": 1900
},
{
"epoch": 3.61,
"grad_norm": 1.0713419914245605,
"learning_rate": 4.431773897482419e-05,
"loss": 0.0235,
"step": 1901
},
{
"epoch": 3.62,
"grad_norm": 1.0905838012695312,
"learning_rate": 4.430150312073167e-05,
"loss": 0.0492,
"step": 1902
},
{
"epoch": 3.62,
"grad_norm": 1.3141876459121704,
"learning_rate": 4.42852618441273e-05,
"loss": 0.0307,
"step": 1903
},
{
"epoch": 3.62,
"grad_norm": 0.738760232925415,
"learning_rate": 4.42690151511691e-05,
"loss": 0.0243,
"step": 1904
},
{
"epoch": 3.62,
"grad_norm": 0.9884717464447021,
"learning_rate": 4.4252763048017095e-05,
"loss": 0.0356,
"step": 1905
},
{
"epoch": 3.62,
"grad_norm": 0.7481637001037598,
"learning_rate": 4.4236505540833386e-05,
"loss": 0.0247,
"step": 1906
},
{
"epoch": 3.63,
"grad_norm": 1.0265824794769287,
"learning_rate": 4.4220242635782126e-05,
"loss": 0.0395,
"step": 1907
},
{
"epoch": 3.63,
"grad_norm": 0.9594699144363403,
"learning_rate": 4.4203974339029485e-05,
"loss": 0.0349,
"step": 1908
},
{
"epoch": 3.63,
"grad_norm": 1.2261050939559937,
"learning_rate": 4.418770065674372e-05,
"loss": 0.0377,
"step": 1909
},
{
"epoch": 3.63,
"grad_norm": 1.5818653106689453,
"learning_rate": 4.417142159509509e-05,
"loss": 0.0564,
"step": 1910
},
{
"epoch": 3.63,
"grad_norm": 0.8432475924491882,
"learning_rate": 4.4155137160255924e-05,
"loss": 0.0315,
"step": 1911
},
{
"epoch": 3.63,
"grad_norm": 1.2255982160568237,
"learning_rate": 4.413884735840058e-05,
"loss": 0.0405,
"step": 1912
},
{
"epoch": 3.64,
"grad_norm": 0.8597308993339539,
"learning_rate": 4.412255219570543e-05,
"loss": 0.0224,
"step": 1913
},
{
"epoch": 3.64,
"grad_norm": 1.1213595867156982,
"learning_rate": 4.4106251678348905e-05,
"loss": 0.0356,
"step": 1914
},
{
"epoch": 3.64,
"grad_norm": 0.923098623752594,
"learning_rate": 4.408994581251146e-05,
"loss": 0.0346,
"step": 1915
},
{
"epoch": 3.64,
"grad_norm": 0.9506223797798157,
"learning_rate": 4.407363460437557e-05,
"loss": 0.0245,
"step": 1916
},
{
"epoch": 3.64,
"grad_norm": 1.2314410209655762,
"learning_rate": 4.405731806012574e-05,
"loss": 0.0348,
"step": 1917
},
{
"epoch": 3.65,
"grad_norm": 1.251181721687317,
"learning_rate": 4.4040996185948495e-05,
"loss": 0.0424,
"step": 1918
},
{
"epoch": 3.65,
"grad_norm": 1.3881059885025024,
"learning_rate": 4.402466898803239e-05,
"loss": 0.0243,
"step": 1919
},
{
"epoch": 3.65,
"grad_norm": 0.9905797243118286,
"learning_rate": 4.4008336472567984e-05,
"loss": 0.0263,
"step": 1920
},
{
"epoch": 3.65,
"grad_norm": 1.0251002311706543,
"learning_rate": 4.399199864574787e-05,
"loss": 0.0428,
"step": 1921
},
{
"epoch": 3.65,
"grad_norm": 0.6155865788459778,
"learning_rate": 4.3975655513766645e-05,
"loss": 0.0229,
"step": 1922
},
{
"epoch": 3.66,
"grad_norm": 1.2263048887252808,
"learning_rate": 4.395930708282091e-05,
"loss": 0.0303,
"step": 1923
},
{
"epoch": 3.66,
"grad_norm": 0.7947932481765747,
"learning_rate": 4.3942953359109295e-05,
"loss": 0.0327,
"step": 1924
},
{
"epoch": 3.66,
"grad_norm": 0.6304160952568054,
"learning_rate": 4.392659434883243e-05,
"loss": 0.0161,
"step": 1925
},
{
"epoch": 3.66,
"grad_norm": 0.8869005441665649,
"learning_rate": 4.3910230058192926e-05,
"loss": 0.0395,
"step": 1926
},
{
"epoch": 3.66,
"grad_norm": 1.1942654848098755,
"learning_rate": 4.389386049339542e-05,
"loss": 0.028,
"step": 1927
},
{
"epoch": 3.67,
"grad_norm": 1.2440085411071777,
"learning_rate": 4.387748566064656e-05,
"loss": 0.0642,
"step": 1928
},
{
"epoch": 3.67,
"grad_norm": 0.7649359703063965,
"learning_rate": 4.386110556615495e-05,
"loss": 0.024,
"step": 1929
},
{
"epoch": 3.67,
"grad_norm": 0.9489123821258545,
"learning_rate": 4.384472021613122e-05,
"loss": 0.0274,
"step": 1930
},
{
"epoch": 3.67,
"grad_norm": 0.9573070406913757,
"learning_rate": 4.3828329616788e-05,
"loss": 0.0229,
"step": 1931
},
{
"epoch": 3.67,
"grad_norm": 0.8571633696556091,
"learning_rate": 4.381193377433989e-05,
"loss": 0.0293,
"step": 1932
},
{
"epoch": 3.67,
"grad_norm": 0.7860523462295532,
"learning_rate": 4.379553269500349e-05,
"loss": 0.0263,
"step": 1933
},
{
"epoch": 3.68,
"grad_norm": 1.6772990226745605,
"learning_rate": 4.377912638499736e-05,
"loss": 0.0479,
"step": 1934
},
{
"epoch": 3.68,
"grad_norm": 0.634107768535614,
"learning_rate": 4.376271485054209e-05,
"loss": 0.0181,
"step": 1935
},
{
"epoch": 3.68,
"grad_norm": 1.1136505603790283,
"learning_rate": 4.3746298097860196e-05,
"loss": 0.0288,
"step": 1936
},
{
"epoch": 3.68,
"grad_norm": 1.1266599893569946,
"learning_rate": 4.372987613317623e-05,
"loss": 0.044,
"step": 1937
},
{
"epoch": 3.68,
"grad_norm": 0.9847716093063354,
"learning_rate": 4.3713448962716674e-05,
"loss": 0.0311,
"step": 1938
},
{
"epoch": 3.69,
"grad_norm": 0.9045705199241638,
"learning_rate": 4.369701659271001e-05,
"loss": 0.0323,
"step": 1939
},
{
"epoch": 3.69,
"grad_norm": 0.8589094877243042,
"learning_rate": 4.368057902938666e-05,
"loss": 0.0283,
"step": 1940
},
{
"epoch": 3.69,
"grad_norm": 1.344851016998291,
"learning_rate": 4.3664136278979076e-05,
"loss": 0.0332,
"step": 1941
},
{
"epoch": 3.69,
"grad_norm": 0.7911847233772278,
"learning_rate": 4.3647688347721614e-05,
"loss": 0.0285,
"step": 1942
},
{
"epoch": 3.69,
"grad_norm": 0.8800073266029358,
"learning_rate": 4.3631235241850626e-05,
"loss": 0.0379,
"step": 1943
},
{
"epoch": 3.7,
"grad_norm": 0.8295862078666687,
"learning_rate": 4.3614776967604397e-05,
"loss": 0.0272,
"step": 1944
},
{
"epoch": 3.7,
"grad_norm": 1.1330403089523315,
"learning_rate": 4.3598313531223226e-05,
"loss": 0.0327,
"step": 1945
},
{
"epoch": 3.7,
"grad_norm": 2.3198530673980713,
"learning_rate": 4.358184493894932e-05,
"loss": 0.03,
"step": 1946
},
{
"epoch": 3.7,
"grad_norm": 0.9906660318374634,
"learning_rate": 4.356537119702685e-05,
"loss": 0.0259,
"step": 1947
},
{
"epoch": 3.7,
"grad_norm": 0.8446246385574341,
"learning_rate": 4.354889231170196e-05,
"loss": 0.0265,
"step": 1948
},
{
"epoch": 3.71,
"grad_norm": 0.7495526075363159,
"learning_rate": 4.3532408289222726e-05,
"loss": 0.0291,
"step": 1949
},
{
"epoch": 3.71,
"grad_norm": 0.9084073901176453,
"learning_rate": 4.351591913583918e-05,
"loss": 0.0242,
"step": 1950
},
{
"epoch": 3.71,
"grad_norm": 1.2089170217514038,
"learning_rate": 4.34994248578033e-05,
"loss": 0.0329,
"step": 1951
},
{
"epoch": 3.71,
"grad_norm": 1.047277569770813,
"learning_rate": 4.348292546136899e-05,
"loss": 0.0276,
"step": 1952
},
{
"epoch": 3.71,
"grad_norm": 1.0374094247817993,
"learning_rate": 4.3466420952792115e-05,
"loss": 0.0336,
"step": 1953
},
{
"epoch": 3.71,
"grad_norm": 0.7848939895629883,
"learning_rate": 4.3449911338330474e-05,
"loss": 0.0205,
"step": 1954
},
{
"epoch": 3.72,
"grad_norm": 1.2703661918640137,
"learning_rate": 4.343339662424379e-05,
"loss": 0.0385,
"step": 1955
},
{
"epoch": 3.72,
"grad_norm": 0.9336416125297546,
"learning_rate": 4.3416876816793744e-05,
"loss": 0.025,
"step": 1956
},
{
"epoch": 3.72,
"grad_norm": 1.1855322122573853,
"learning_rate": 4.340035192224392e-05,
"loss": 0.0549,
"step": 1957
},
{
"epoch": 3.72,
"grad_norm": 1.1683907508850098,
"learning_rate": 4.338382194685985e-05,
"loss": 0.0337,
"step": 1958
},
{
"epoch": 3.72,
"grad_norm": 0.7613326907157898,
"learning_rate": 4.3367286896908994e-05,
"loss": 0.0175,
"step": 1959
},
{
"epoch": 3.73,
"grad_norm": 0.957425057888031,
"learning_rate": 4.33507467786607e-05,
"loss": 0.0246,
"step": 1960
},
{
"epoch": 3.73,
"grad_norm": 0.9485387206077576,
"learning_rate": 4.3334201598386295e-05,
"loss": 0.0262,
"step": 1961
},
{
"epoch": 3.73,
"grad_norm": 0.9271544814109802,
"learning_rate": 4.331765136235899e-05,
"loss": 0.0427,
"step": 1962
},
{
"epoch": 3.73,
"grad_norm": 1.0190984010696411,
"learning_rate": 4.33010960768539e-05,
"loss": 0.0335,
"step": 1963
},
{
"epoch": 3.73,
"grad_norm": 0.8287107348442078,
"learning_rate": 4.32845357481481e-05,
"loss": 0.0153,
"step": 1964
},
{
"epoch": 3.74,
"grad_norm": 1.134548306465149,
"learning_rate": 4.326797038252054e-05,
"loss": 0.0393,
"step": 1965
},
{
"epoch": 3.74,
"grad_norm": 1.0449700355529785,
"learning_rate": 4.325139998625208e-05,
"loss": 0.0353,
"step": 1966
},
{
"epoch": 3.74,
"grad_norm": 0.8753441572189331,
"learning_rate": 4.3234824565625516e-05,
"loss": 0.0275,
"step": 1967
},
{
"epoch": 3.74,
"grad_norm": 0.9925007224082947,
"learning_rate": 4.321824412692552e-05,
"loss": 0.0479,
"step": 1968
},
{
"epoch": 3.74,
"grad_norm": 0.951697051525116,
"learning_rate": 4.3201658676438684e-05,
"loss": 0.0335,
"step": 1969
},
{
"epoch": 3.75,
"grad_norm": 1.1725651025772095,
"learning_rate": 4.3185068220453484e-05,
"loss": 0.0294,
"step": 1970
},
{
"epoch": 3.75,
"grad_norm": 0.7476635575294495,
"learning_rate": 4.3168472765260314e-05,
"loss": 0.0241,
"step": 1971
},
{
"epoch": 3.75,
"grad_norm": 0.7880444526672363,
"learning_rate": 4.3151872317151456e-05,
"loss": 0.0139,
"step": 1972
},
{
"epoch": 3.75,
"grad_norm": 0.9709190726280212,
"learning_rate": 4.3135266882421076e-05,
"loss": 0.0283,
"step": 1973
},
{
"epoch": 3.75,
"grad_norm": 0.6388635635375977,
"learning_rate": 4.311865646736524e-05,
"loss": 0.0139,
"step": 1974
},
{
"epoch": 3.75,
"grad_norm": 1.018311619758606,
"learning_rate": 4.310204107828189e-05,
"loss": 0.0416,
"step": 1975
},
{
"epoch": 3.76,
"grad_norm": 0.9112716913223267,
"learning_rate": 4.3085420721470864e-05,
"loss": 0.0331,
"step": 1976
},
{
"epoch": 3.76,
"grad_norm": 0.9825039505958557,
"learning_rate": 4.30687954032339e-05,
"loss": 0.0323,
"step": 1977
},
{
"epoch": 3.76,
"grad_norm": 0.8889324069023132,
"learning_rate": 4.305216512987459e-05,
"loss": 0.028,
"step": 1978
},
{
"epoch": 3.76,
"grad_norm": 1.0754867792129517,
"learning_rate": 4.30355299076984e-05,
"loss": 0.0284,
"step": 1979
},
{
"epoch": 3.76,
"grad_norm": 1.0186346769332886,
"learning_rate": 4.301888974301271e-05,
"loss": 0.0293,
"step": 1980
},
{
"epoch": 3.77,
"grad_norm": 1.0310583114624023,
"learning_rate": 4.300224464212675e-05,
"loss": 0.0244,
"step": 1981
},
{
"epoch": 3.77,
"grad_norm": 0.9594578146934509,
"learning_rate": 4.2985594611351614e-05,
"loss": 0.0365,
"step": 1982
},
{
"epoch": 3.77,
"grad_norm": 0.7327626943588257,
"learning_rate": 4.2968939657000267e-05,
"loss": 0.0237,
"step": 1983
},
{
"epoch": 3.77,
"grad_norm": 0.9437559843063354,
"learning_rate": 4.2952279785387564e-05,
"loss": 0.0231,
"step": 1984
},
{
"epoch": 3.77,
"grad_norm": 0.6189577579498291,
"learning_rate": 4.29356150028302e-05,
"loss": 0.0149,
"step": 1985
},
{
"epoch": 3.78,
"grad_norm": 0.8695778250694275,
"learning_rate": 4.291894531564673e-05,
"loss": 0.0275,
"step": 1986
},
{
"epoch": 3.78,
"grad_norm": 0.9207600355148315,
"learning_rate": 4.2902270730157595e-05,
"loss": 0.038,
"step": 1987
},
{
"epoch": 3.78,
"grad_norm": 0.9751211404800415,
"learning_rate": 4.288559125268508e-05,
"loss": 0.0199,
"step": 1988
},
{
"epoch": 3.78,
"grad_norm": 1.0845248699188232,
"learning_rate": 4.2868906889553305e-05,
"loss": 0.046,
"step": 1989
},
{
"epoch": 3.78,
"grad_norm": 1.0575621128082275,
"learning_rate": 4.285221764708827e-05,
"loss": 0.0205,
"step": 1990
},
{
"epoch": 3.79,
"grad_norm": 1.0269300937652588,
"learning_rate": 4.283552353161781e-05,
"loss": 0.0373,
"step": 1991
},
{
"epoch": 3.79,
"grad_norm": 0.731108546257019,
"learning_rate": 4.2818824549471624e-05,
"loss": 0.0239,
"step": 1992
},
{
"epoch": 3.79,
"grad_norm": 0.875087559223175,
"learning_rate": 4.280212070698122e-05,
"loss": 0.0273,
"step": 1993
},
{
"epoch": 3.79,
"grad_norm": 1.243884801864624,
"learning_rate": 4.278541201048e-05,
"loss": 0.0423,
"step": 1994
},
{
"epoch": 3.79,
"grad_norm": 0.82115638256073,
"learning_rate": 4.276869846630316e-05,
"loss": 0.0196,
"step": 1995
},
{
"epoch": 3.79,
"grad_norm": 0.6430832147598267,
"learning_rate": 4.275198008078777e-05,
"loss": 0.0172,
"step": 1996
},
{
"epoch": 3.8,
"grad_norm": 0.9115181565284729,
"learning_rate": 4.27352568602727e-05,
"loss": 0.0312,
"step": 1997
},
{
"epoch": 3.8,
"grad_norm": 0.9637769460678101,
"learning_rate": 4.271852881109869e-05,
"loss": 0.0331,
"step": 1998
},
{
"epoch": 3.8,
"grad_norm": 0.9175263047218323,
"learning_rate": 4.270179593960828e-05,
"loss": 0.029,
"step": 1999
},
{
"epoch": 3.8,
"grad_norm": 0.7917808294296265,
"learning_rate": 4.268505825214586e-05,
"loss": 0.0247,
"step": 2000
},
{
"epoch": 3.8,
"eval_blimp_filtered_avg": 0.7138805970149253,
"eval_blimp_filtered_std": 0.004994241876351901,
"step": 2000
},
{
"epoch": 3.8,
"eval_blimp_supplement_avg": 0.7672413793103449,
"eval_blimp_supplement_std": 0.018502406408271665,
"step": 2000
},
{
"epoch": 3.8,
"eval_vqa_filtered_avg": 0.31,
"eval_vqa_filtered_std": 0.04648231987117316,
"step": 2000
},
{
"epoch": 3.8,
"eval_winoground_filtered_avg": 0.49,
"eval_winoground_filtered_std": 0.05024183937956912,
"step": 2000
},
{
"epoch": 3.8,
"grad_norm": 0.8691048622131348,
"learning_rate": 4.2668315755057624e-05,
"loss": 0.0191,
"step": 2001
},
{
"epoch": 3.81,
"grad_norm": 1.3015999794006348,
"learning_rate": 4.2651568454691625e-05,
"loss": 0.0521,
"step": 2002
},
{
"epoch": 3.81,
"grad_norm": 0.7557706832885742,
"learning_rate": 4.26348163573977e-05,
"loss": 0.0244,
"step": 2003
},
{
"epoch": 3.81,
"grad_norm": 0.8561527132987976,
"learning_rate": 4.261805946952753e-05,
"loss": 0.0302,
"step": 2004
},
{
"epoch": 3.81,
"grad_norm": 1.2092703580856323,
"learning_rate": 4.2601297797434594e-05,
"loss": 0.0377,
"step": 2005
},
{
"epoch": 3.81,
"grad_norm": 0.6772445440292358,
"learning_rate": 4.258453134747421e-05,
"loss": 0.0144,
"step": 2006
},
{
"epoch": 3.82,
"grad_norm": 0.9493342041969299,
"learning_rate": 4.2567760126003456e-05,
"loss": 0.0156,
"step": 2007
},
{
"epoch": 3.82,
"grad_norm": 1.2367883920669556,
"learning_rate": 4.2550984139381293e-05,
"loss": 0.0343,
"step": 2008
},
{
"epoch": 3.82,
"grad_norm": 0.6001228094100952,
"learning_rate": 4.253420339396842e-05,
"loss": 0.0125,
"step": 2009
},
{
"epoch": 3.82,
"grad_norm": 1.1965322494506836,
"learning_rate": 4.251741789612739e-05,
"loss": 0.0248,
"step": 2010
},
{
"epoch": 3.82,
"grad_norm": 1.1930829286575317,
"learning_rate": 4.250062765222254e-05,
"loss": 0.0448,
"step": 2011
},
{
"epoch": 3.83,
"grad_norm": 1.2990081310272217,
"learning_rate": 4.248383266861998e-05,
"loss": 0.0294,
"step": 2012
},
{
"epoch": 3.83,
"grad_norm": 0.5081372261047363,
"learning_rate": 4.246703295168767e-05,
"loss": 0.0151,
"step": 2013
},
{
"epoch": 3.83,
"grad_norm": 1.3554402589797974,
"learning_rate": 4.245022850779532e-05,
"loss": 0.0217,
"step": 2014
},
{
"epoch": 3.83,
"grad_norm": 1.4262875318527222,
"learning_rate": 4.243341934331445e-05,
"loss": 0.0334,
"step": 2015
},
{
"epoch": 3.83,
"grad_norm": 0.8028334975242615,
"learning_rate": 4.241660546461838e-05,
"loss": 0.0201,
"step": 2016
},
{
"epoch": 3.83,
"grad_norm": 0.7291403412818909,
"learning_rate": 4.23997868780822e-05,
"loss": 0.0164,
"step": 2017
},
{
"epoch": 3.84,
"grad_norm": 1.8509825468063354,
"learning_rate": 4.238296359008279e-05,
"loss": 0.0293,
"step": 2018
},
{
"epoch": 3.84,
"grad_norm": 0.9662816524505615,
"learning_rate": 4.2366135606998816e-05,
"loss": 0.0298,
"step": 2019
},
{
"epoch": 3.84,
"grad_norm": 1.011863350868225,
"learning_rate": 4.234930293521071e-05,
"loss": 0.0299,
"step": 2020
},
{
"epoch": 3.84,
"grad_norm": 1.1940956115722656,
"learning_rate": 4.2332465581100705e-05,
"loss": 0.035,
"step": 2021
},
{
"epoch": 3.84,
"grad_norm": 1.1523157358169556,
"learning_rate": 4.2315623551052796e-05,
"loss": 0.0386,
"step": 2022
},
{
"epoch": 3.85,
"grad_norm": 0.5764814615249634,
"learning_rate": 4.2298776851452755e-05,
"loss": 0.0194,
"step": 2023
},
{
"epoch": 3.85,
"grad_norm": 1.1772462129592896,
"learning_rate": 4.228192548868811e-05,
"loss": 0.044,
"step": 2024
},
{
"epoch": 3.85,
"grad_norm": 1.1659222841262817,
"learning_rate": 4.226506946914819e-05,
"loss": 0.049,
"step": 2025
},
{
"epoch": 3.85,
"grad_norm": 1.7661395072937012,
"learning_rate": 4.224820879922406e-05,
"loss": 0.0678,
"step": 2026
},
{
"epoch": 3.85,
"grad_norm": 0.8586690425872803,
"learning_rate": 4.223134348530855e-05,
"loss": 0.0207,
"step": 2027
},
{
"epoch": 3.86,
"grad_norm": 1.2062411308288574,
"learning_rate": 4.221447353379626e-05,
"loss": 0.0444,
"step": 2028
},
{
"epoch": 3.86,
"grad_norm": 0.9996699094772339,
"learning_rate": 4.219759895108356e-05,
"loss": 0.0303,
"step": 2029
},
{
"epoch": 3.86,
"grad_norm": 0.7963805198669434,
"learning_rate": 4.218071974356855e-05,
"loss": 0.0276,
"step": 2030
},
{
"epoch": 3.86,
"grad_norm": 1.2967747449874878,
"learning_rate": 4.2163835917651094e-05,
"loss": 0.0245,
"step": 2031
},
{
"epoch": 3.86,
"grad_norm": 0.782240092754364,
"learning_rate": 4.214694747973281e-05,
"loss": 0.026,
"step": 2032
},
{
"epoch": 3.87,
"grad_norm": 1.246574878692627,
"learning_rate": 4.2130054436217075e-05,
"loss": 0.0371,
"step": 2033
},
{
"epoch": 3.87,
"grad_norm": 1.0264090299606323,
"learning_rate": 4.211315679350899e-05,
"loss": 0.0297,
"step": 2034
},
{
"epoch": 3.87,
"grad_norm": 0.7853670716285706,
"learning_rate": 4.2096254558015405e-05,
"loss": 0.0197,
"step": 2035
},
{
"epoch": 3.87,
"grad_norm": 0.852821409702301,
"learning_rate": 4.2079347736144934e-05,
"loss": 0.0458,
"step": 2036
},
{
"epoch": 3.87,
"grad_norm": 1.168858528137207,
"learning_rate": 4.206243633430791e-05,
"loss": 0.057,
"step": 2037
},
{
"epoch": 3.87,
"grad_norm": 1.0644727945327759,
"learning_rate": 4.204552035891639e-05,
"loss": 0.0312,
"step": 2038
},
{
"epoch": 3.88,
"grad_norm": 0.9905727505683899,
"learning_rate": 4.2028599816384195e-05,
"loss": 0.0437,
"step": 2039
},
{
"epoch": 3.88,
"grad_norm": 1.132205605506897,
"learning_rate": 4.201167471312687e-05,
"loss": 0.0243,
"step": 2040
},
{
"epoch": 3.88,
"grad_norm": 0.6083950996398926,
"learning_rate": 4.199474505556167e-05,
"loss": 0.0115,
"step": 2041
},
{
"epoch": 3.88,
"grad_norm": 0.9922460317611694,
"learning_rate": 4.197781085010758e-05,
"loss": 0.0445,
"step": 2042
},
{
"epoch": 3.88,
"grad_norm": 0.49889275431632996,
"learning_rate": 4.196087210318534e-05,
"loss": 0.0138,
"step": 2043
},
{
"epoch": 3.89,
"grad_norm": 1.2560492753982544,
"learning_rate": 4.1943928821217375e-05,
"loss": 0.0392,
"step": 2044
},
{
"epoch": 3.89,
"grad_norm": 0.7352301478385925,
"learning_rate": 4.1926981010627846e-05,
"loss": 0.0225,
"step": 2045
},
{
"epoch": 3.89,
"grad_norm": 1.2378711700439453,
"learning_rate": 4.191002867784265e-05,
"loss": 0.0345,
"step": 2046
},
{
"epoch": 3.89,
"grad_norm": 0.5803519487380981,
"learning_rate": 4.189307182928936e-05,
"loss": 0.0163,
"step": 2047
},
{
"epoch": 3.89,
"grad_norm": 0.9804807305335999,
"learning_rate": 4.187611047139727e-05,
"loss": 0.0285,
"step": 2048
},
{
"epoch": 3.9,
"grad_norm": 0.5991160869598389,
"learning_rate": 4.185914461059742e-05,
"loss": 0.0204,
"step": 2049
},
{
"epoch": 3.9,
"grad_norm": 1.6420141458511353,
"learning_rate": 4.184217425332253e-05,
"loss": 0.0334,
"step": 2050
},
{
"epoch": 3.9,
"grad_norm": 0.847574770450592,
"learning_rate": 4.1825199406007006e-05,
"loss": 0.0224,
"step": 2051
},
{
"epoch": 3.9,
"grad_norm": 1.0283008813858032,
"learning_rate": 4.180822007508698e-05,
"loss": 0.02,
"step": 2052
},
{
"epoch": 3.9,
"grad_norm": 0.5995702743530273,
"learning_rate": 4.1791236267000295e-05,
"loss": 0.027,
"step": 2053
},
{
"epoch": 3.9,
"grad_norm": 1.0898993015289307,
"learning_rate": 4.177424798818646e-05,
"loss": 0.0222,
"step": 2054
},
{
"epoch": 3.91,
"grad_norm": 1.4892098903656006,
"learning_rate": 4.175725524508671e-05,
"loss": 0.0364,
"step": 2055
},
{
"epoch": 3.91,
"grad_norm": 0.8933863043785095,
"learning_rate": 4.1740258044143945e-05,
"loss": 0.0339,
"step": 2056
},
{
"epoch": 3.91,
"grad_norm": 0.946526825428009,
"learning_rate": 4.172325639180279e-05,
"loss": 0.0215,
"step": 2057
},
{
"epoch": 3.91,
"grad_norm": 1.1317492723464966,
"learning_rate": 4.170625029450951e-05,
"loss": 0.0443,
"step": 2058
},
{
"epoch": 3.91,
"grad_norm": 1.1115379333496094,
"learning_rate": 4.1689239758712104e-05,
"loss": 0.0338,
"step": 2059
},
{
"epoch": 3.92,
"grad_norm": 1.4766833782196045,
"learning_rate": 4.167222479086022e-05,
"loss": 0.0406,
"step": 2060
},
{
"epoch": 3.92,
"grad_norm": 2.2056808471679688,
"learning_rate": 4.16552053974052e-05,
"loss": 0.0433,
"step": 2061
},
{
"epoch": 3.92,
"grad_norm": 1.5896015167236328,
"learning_rate": 4.163818158480007e-05,
"loss": 0.0355,
"step": 2062
},
{
"epoch": 3.92,
"grad_norm": 0.8700101971626282,
"learning_rate": 4.162115335949951e-05,
"loss": 0.0241,
"step": 2063
},
{
"epoch": 3.92,
"grad_norm": 1.0466557741165161,
"learning_rate": 4.160412072795991e-05,
"loss": 0.0401,
"step": 2064
},
{
"epoch": 3.93,
"grad_norm": 1.175431251525879,
"learning_rate": 4.158708369663928e-05,
"loss": 0.0196,
"step": 2065
},
{
"epoch": 3.93,
"grad_norm": 1.515027403831482,
"learning_rate": 4.157004227199735e-05,
"loss": 0.0428,
"step": 2066
},
{
"epoch": 3.93,
"grad_norm": 1.2282580137252808,
"learning_rate": 4.1552996460495475e-05,
"loss": 0.0236,
"step": 2067
},
{
"epoch": 3.93,
"grad_norm": 0.8119418621063232,
"learning_rate": 4.1535946268596684e-05,
"loss": 0.0206,
"step": 2068
},
{
"epoch": 3.93,
"grad_norm": 1.0901038646697998,
"learning_rate": 4.151889170276569e-05,
"loss": 0.0231,
"step": 2069
},
{
"epoch": 3.94,
"grad_norm": 1.1174488067626953,
"learning_rate": 4.150183276946884e-05,
"loss": 0.0396,
"step": 2070
},
{
"epoch": 3.94,
"grad_norm": 1.1922675371170044,
"learning_rate": 4.1484769475174155e-05,
"loss": 0.0257,
"step": 2071
},
{
"epoch": 3.94,
"grad_norm": 0.7354407906532288,
"learning_rate": 4.146770182635128e-05,
"loss": 0.0196,
"step": 2072
},
{
"epoch": 3.94,
"grad_norm": 1.237693428993225,
"learning_rate": 4.145062982947154e-05,
"loss": 0.0314,
"step": 2073
},
{
"epoch": 3.94,
"grad_norm": 1.545982837677002,
"learning_rate": 4.143355349100791e-05,
"loss": 0.0379,
"step": 2074
},
{
"epoch": 3.94,
"grad_norm": 1.116953730583191,
"learning_rate": 4.1416472817434976e-05,
"loss": 0.0436,
"step": 2075
},
{
"epoch": 3.95,
"grad_norm": 0.8453783392906189,
"learning_rate": 4.1399387815229e-05,
"loss": 0.0216,
"step": 2076
},
{
"epoch": 3.95,
"grad_norm": 1.111954689025879,
"learning_rate": 4.138229849086788e-05,
"loss": 0.0326,
"step": 2077
},
{
"epoch": 3.95,
"grad_norm": 1.1272625923156738,
"learning_rate": 4.1365204850831144e-05,
"loss": 0.0281,
"step": 2078
},
{
"epoch": 3.95,
"grad_norm": 0.9043669700622559,
"learning_rate": 4.134810690159997e-05,
"loss": 0.0391,
"step": 2079
},
{
"epoch": 3.95,
"grad_norm": 0.9364864826202393,
"learning_rate": 4.133100464965716e-05,
"loss": 0.0337,
"step": 2080
},
{
"epoch": 3.96,
"grad_norm": 1.29058039188385,
"learning_rate": 4.1313898101487136e-05,
"loss": 0.042,
"step": 2081
},
{
"epoch": 3.96,
"grad_norm": 1.1474173069000244,
"learning_rate": 4.129678726357597e-05,
"loss": 0.0359,
"step": 2082
},
{
"epoch": 3.96,
"grad_norm": 0.7792196869850159,
"learning_rate": 4.127967214241136e-05,
"loss": 0.0265,
"step": 2083
},
{
"epoch": 3.96,
"grad_norm": 0.8722406625747681,
"learning_rate": 4.12625527444826e-05,
"loss": 0.0124,
"step": 2084
},
{
"epoch": 3.96,
"grad_norm": 1.1275458335876465,
"learning_rate": 4.124542907628065e-05,
"loss": 0.0261,
"step": 2085
},
{
"epoch": 3.97,
"grad_norm": 1.2096914052963257,
"learning_rate": 4.122830114429806e-05,
"loss": 0.0313,
"step": 2086
},
{
"epoch": 3.97,
"grad_norm": 1.4907256364822388,
"learning_rate": 4.1211168955028986e-05,
"loss": 0.0474,
"step": 2087
},
{
"epoch": 3.97,
"grad_norm": 1.1119025945663452,
"learning_rate": 4.1194032514969225e-05,
"loss": 0.0255,
"step": 2088
},
{
"epoch": 3.97,
"grad_norm": 1.4212636947631836,
"learning_rate": 4.117689183061617e-05,
"loss": 0.0422,
"step": 2089
},
{
"epoch": 3.97,
"grad_norm": 1.0075161457061768,
"learning_rate": 4.1159746908468834e-05,
"loss": 0.0405,
"step": 2090
},
{
"epoch": 3.98,
"grad_norm": 1.3709818124771118,
"learning_rate": 4.1142597755027815e-05,
"loss": 0.0374,
"step": 2091
},
{
"epoch": 3.98,
"grad_norm": 1.0767892599105835,
"learning_rate": 4.112544437679535e-05,
"loss": 0.0321,
"step": 2092
},
{
"epoch": 3.98,
"grad_norm": 1.0866590738296509,
"learning_rate": 4.110828678027525e-05,
"loss": 0.0302,
"step": 2093
},
{
"epoch": 3.98,
"grad_norm": 0.6481738090515137,
"learning_rate": 4.109112497197293e-05,
"loss": 0.013,
"step": 2094
},
{
"epoch": 3.98,
"grad_norm": 1.6165426969528198,
"learning_rate": 4.107395895839541e-05,
"loss": 0.051,
"step": 2095
},
{
"epoch": 3.98,
"grad_norm": 0.7774046659469604,
"learning_rate": 4.1056788746051316e-05,
"loss": 0.0364,
"step": 2096
},
{
"epoch": 3.99,
"grad_norm": 0.8804233074188232,
"learning_rate": 4.1039614341450823e-05,
"loss": 0.0299,
"step": 2097
},
{
"epoch": 3.99,
"grad_norm": 0.8414275050163269,
"learning_rate": 4.102243575110574e-05,
"loss": 0.0241,
"step": 2098
},
{
"epoch": 3.99,
"grad_norm": 1.0845783948898315,
"learning_rate": 4.1005252981529434e-05,
"loss": 0.0277,
"step": 2099
},
{
"epoch": 3.99,
"grad_norm": 1.1695177555084229,
"learning_rate": 4.098806603923688e-05,
"loss": 0.0402,
"step": 2100
},
{
"epoch": 3.99,
"eval_blimp_filtered_avg": 0.718955223880597,
"eval_blimp_filtered_std": 0.004979803317671439,
"step": 2100
},
{
"epoch": 3.99,
"eval_blimp_supplement_avg": 0.7564655172413793,
"eval_blimp_supplement_std": 0.018843050356252562,
"step": 2100
},
{
"epoch": 3.99,
"eval_vqa_filtered_avg": 0.37,
"eval_vqa_filtered_std": 0.048523658709391,
"step": 2100
},
{
"epoch": 3.99,
"eval_winoground_filtered_avg": 0.5,
"eval_winoground_filtered_std": 0.050251890762960605,
"step": 2100
},
{
"epoch": 3.99,
"grad_norm": 0.8187968730926514,
"learning_rate": 4.097087493074461e-05,
"loss": 0.0338,
"step": 2101
},
{
"epoch": 4.0,
"grad_norm": 0.838556170463562,
"learning_rate": 4.095367966257077e-05,
"loss": 0.0309,
"step": 2102
},
{
"epoch": 4.0,
"grad_norm": 0.9308669567108154,
"learning_rate": 4.0936480241235047e-05,
"loss": 0.0267,
"step": 2103
},
{
"epoch": 4.0,
"grad_norm": 1.143006682395935,
"learning_rate": 4.091927667325872e-05,
"loss": 0.0434,
"step": 2104
},
{
"epoch": 4.0,
"grad_norm": 1.4613186120986938,
"learning_rate": 4.090206896516464e-05,
"loss": 0.021,
"step": 2105
},
{
"epoch": 4.0,
"grad_norm": 0.6796694397926331,
"learning_rate": 4.088485712347723e-05,
"loss": 0.0219,
"step": 2106
},
{
"epoch": 4.01,
"grad_norm": 0.5982329845428467,
"learning_rate": 4.086764115472247e-05,
"loss": 0.024,
"step": 2107
},
{
"epoch": 4.01,
"grad_norm": 1.0702226161956787,
"learning_rate": 4.085042106542792e-05,
"loss": 0.0447,
"step": 2108
},
{
"epoch": 4.01,
"grad_norm": 0.7936846613883972,
"learning_rate": 4.0833196862122686e-05,
"loss": 0.0201,
"step": 2109
},
{
"epoch": 4.01,
"grad_norm": 0.6613113880157471,
"learning_rate": 4.081596855133744e-05,
"loss": 0.0142,
"step": 2110
},
{
"epoch": 4.01,
"grad_norm": 0.6447283029556274,
"learning_rate": 4.079873613960442e-05,
"loss": 0.0222,
"step": 2111
},
{
"epoch": 4.02,
"grad_norm": 0.9131982326507568,
"learning_rate": 4.0781499633457404e-05,
"loss": 0.0411,
"step": 2112
},
{
"epoch": 4.02,
"grad_norm": 0.8375611305236816,
"learning_rate": 4.0764259039431735e-05,
"loss": 0.0235,
"step": 2113
},
{
"epoch": 4.02,
"grad_norm": 0.9506580829620361,
"learning_rate": 4.07470143640643e-05,
"loss": 0.0256,
"step": 2114
},
{
"epoch": 4.02,
"grad_norm": 0.6485140323638916,
"learning_rate": 4.0729765613893534e-05,
"loss": 0.0184,
"step": 2115
},
{
"epoch": 4.02,
"grad_norm": 0.862721860408783,
"learning_rate": 4.071251279545942e-05,
"loss": 0.0367,
"step": 2116
},
{
"epoch": 4.02,
"grad_norm": 0.656412661075592,
"learning_rate": 4.069525591530348e-05,
"loss": 0.0254,
"step": 2117
},
{
"epoch": 4.03,
"grad_norm": 0.45743611454963684,
"learning_rate": 4.067799497996877e-05,
"loss": 0.0113,
"step": 2118
},
{
"epoch": 4.03,
"grad_norm": 0.5529853701591492,
"learning_rate": 4.06607299959999e-05,
"loss": 0.0136,
"step": 2119
},
{
"epoch": 4.03,
"grad_norm": 0.6535254120826721,
"learning_rate": 4.064346096994301e-05,
"loss": 0.0149,
"step": 2120
},
{
"epoch": 4.03,
"grad_norm": 0.6137189269065857,
"learning_rate": 4.062618790834575e-05,
"loss": 0.022,
"step": 2121
},
{
"epoch": 4.03,
"grad_norm": 0.7624489665031433,
"learning_rate": 4.060891081775733e-05,
"loss": 0.016,
"step": 2122
},
{
"epoch": 4.04,
"grad_norm": 0.9278644919395447,
"learning_rate": 4.059162970472848e-05,
"loss": 0.0344,
"step": 2123
},
{
"epoch": 4.04,
"grad_norm": 0.6781790256500244,
"learning_rate": 4.057434457581144e-05,
"loss": 0.0183,
"step": 2124
},
{
"epoch": 4.04,
"grad_norm": 0.8344194889068604,
"learning_rate": 4.055705543756e-05,
"loss": 0.0256,
"step": 2125
},
{
"epoch": 4.04,
"grad_norm": 0.5823060870170593,
"learning_rate": 4.053976229652943e-05,
"loss": 0.0138,
"step": 2126
},
{
"epoch": 4.04,
"grad_norm": 0.6960210204124451,
"learning_rate": 4.0522465159276565e-05,
"loss": 0.0145,
"step": 2127
},
{
"epoch": 4.05,
"grad_norm": 0.5206294655799866,
"learning_rate": 4.0505164032359716e-05,
"loss": 0.0211,
"step": 2128
},
{
"epoch": 4.05,
"grad_norm": 1.2688753604888916,
"learning_rate": 4.048785892233874e-05,
"loss": 0.0441,
"step": 2129
},
{
"epoch": 4.05,
"grad_norm": 1.675310730934143,
"learning_rate": 4.047054983577496e-05,
"loss": 0.0309,
"step": 2130
},
{
"epoch": 4.05,
"grad_norm": 1.2165312767028809,
"learning_rate": 4.045323677923126e-05,
"loss": 0.0392,
"step": 2131
},
{
"epoch": 4.05,
"grad_norm": 0.8578410744667053,
"learning_rate": 4.043591975927199e-05,
"loss": 0.0232,
"step": 2132
},
{
"epoch": 4.06,
"grad_norm": 0.675644040107727,
"learning_rate": 4.041859878246301e-05,
"loss": 0.0148,
"step": 2133
},
{
"epoch": 4.06,
"grad_norm": 0.6937960386276245,
"learning_rate": 4.040127385537171e-05,
"loss": 0.0181,
"step": 2134
},
{
"epoch": 4.06,
"grad_norm": 0.9674006700515747,
"learning_rate": 4.038394498456692e-05,
"loss": 0.0331,
"step": 2135
},
{
"epoch": 4.06,
"grad_norm": 0.6659774780273438,
"learning_rate": 4.0366612176619014e-05,
"loss": 0.0224,
"step": 2136
},
{
"epoch": 4.06,
"grad_norm": 0.8424309492111206,
"learning_rate": 4.0349275438099855e-05,
"loss": 0.0252,
"step": 2137
},
{
"epoch": 4.06,
"grad_norm": 0.7981933951377869,
"learning_rate": 4.0331934775582776e-05,
"loss": 0.016,
"step": 2138
},
{
"epoch": 4.07,
"grad_norm": 1.4502568244934082,
"learning_rate": 4.03145901956426e-05,
"loss": 0.0267,
"step": 2139
},
{
"epoch": 4.07,
"grad_norm": 1.272951364517212,
"learning_rate": 4.0297241704855645e-05,
"loss": 0.0269,
"step": 2140
},
{
"epoch": 4.07,
"grad_norm": 0.7161095142364502,
"learning_rate": 4.0279889309799716e-05,
"loss": 0.0114,
"step": 2141
},
{
"epoch": 4.07,
"grad_norm": 0.843532383441925,
"learning_rate": 4.0262533017054087e-05,
"loss": 0.0233,
"step": 2142
},
{
"epoch": 4.07,
"grad_norm": 0.8392140865325928,
"learning_rate": 4.024517283319951e-05,
"loss": 0.02,
"step": 2143
},
{
"epoch": 4.08,
"grad_norm": 0.6384946703910828,
"learning_rate": 4.0227808764818225e-05,
"loss": 0.0161,
"step": 2144
},
{
"epoch": 4.08,
"grad_norm": 1.0312801599502563,
"learning_rate": 4.021044081849393e-05,
"loss": 0.0172,
"step": 2145
},
{
"epoch": 4.08,
"grad_norm": 0.8512808680534363,
"learning_rate": 4.01930690008118e-05,
"loss": 0.0268,
"step": 2146
},
{
"epoch": 4.08,
"grad_norm": 1.0077192783355713,
"learning_rate": 4.017569331835847e-05,
"loss": 0.0266,
"step": 2147
},
{
"epoch": 4.08,
"grad_norm": 0.5882465243339539,
"learning_rate": 4.0158313777722064e-05,
"loss": 0.0094,
"step": 2148
},
{
"epoch": 4.09,
"grad_norm": 0.7828996181488037,
"learning_rate": 4.014093038549214e-05,
"loss": 0.0226,
"step": 2149
},
{
"epoch": 4.09,
"grad_norm": 0.8872832655906677,
"learning_rate": 4.012354314825974e-05,
"loss": 0.0301,
"step": 2150
},
{
"epoch": 4.09,
"grad_norm": 0.9625177383422852,
"learning_rate": 4.010615207261735e-05,
"loss": 0.0306,
"step": 2151
},
{
"epoch": 4.09,
"grad_norm": 0.4337451756000519,
"learning_rate": 4.00887571651589e-05,
"loss": 0.0071,
"step": 2152
},
{
"epoch": 4.09,
"grad_norm": 1.3269392251968384,
"learning_rate": 4.007135843247979e-05,
"loss": 0.0167,
"step": 2153
},
{
"epoch": 4.1,
"grad_norm": 0.6793783903121948,
"learning_rate": 4.005395588117689e-05,
"loss": 0.0156,
"step": 2154
},
{
"epoch": 4.1,
"grad_norm": 0.7077756524085999,
"learning_rate": 4.003654951784848e-05,
"loss": 0.0137,
"step": 2155
},
{
"epoch": 4.1,
"grad_norm": 0.9005546569824219,
"learning_rate": 4.00191393490943e-05,
"loss": 0.0259,
"step": 2156
},
{
"epoch": 4.1,
"grad_norm": 0.780119776725769,
"learning_rate": 4.0001725381515534e-05,
"loss": 0.0259,
"step": 2157
},
{
"epoch": 4.1,
"grad_norm": 0.6831116080284119,
"learning_rate": 3.99843076217148e-05,
"loss": 0.0122,
"step": 2158
},
{
"epoch": 4.1,
"grad_norm": 0.8950064778327942,
"learning_rate": 3.996688607629617e-05,
"loss": 0.0293,
"step": 2159
},
{
"epoch": 4.11,
"grad_norm": 0.9443300366401672,
"learning_rate": 3.9949460751865135e-05,
"loss": 0.0172,
"step": 2160
},
{
"epoch": 4.11,
"grad_norm": 0.9208161234855652,
"learning_rate": 3.9932031655028625e-05,
"loss": 0.0098,
"step": 2161
},
{
"epoch": 4.11,
"grad_norm": 1.2588132619857788,
"learning_rate": 3.9914598792395006e-05,
"loss": 0.0457,
"step": 2162
},
{
"epoch": 4.11,
"grad_norm": 0.6830695867538452,
"learning_rate": 3.989716217057406e-05,
"loss": 0.0122,
"step": 2163
},
{
"epoch": 4.11,
"grad_norm": 0.8135195374488831,
"learning_rate": 3.9879721796177e-05,
"loss": 0.0182,
"step": 2164
},
{
"epoch": 4.12,
"grad_norm": 0.6787047386169434,
"learning_rate": 3.986227767581647e-05,
"loss": 0.0158,
"step": 2165
},
{
"epoch": 4.12,
"grad_norm": 0.5418242812156677,
"learning_rate": 3.9844829816106515e-05,
"loss": 0.012,
"step": 2166
},
{
"epoch": 4.12,
"grad_norm": 0.6519037485122681,
"learning_rate": 3.982737822366262e-05,
"loss": 0.0154,
"step": 2167
},
{
"epoch": 4.12,
"grad_norm": 0.5462363958358765,
"learning_rate": 3.980992290510168e-05,
"loss": 0.0083,
"step": 2168
},
{
"epoch": 4.12,
"grad_norm": 0.8263380527496338,
"learning_rate": 3.979246386704196e-05,
"loss": 0.0274,
"step": 2169
},
{
"epoch": 4.13,
"grad_norm": 1.5204890966415405,
"learning_rate": 3.977500111610322e-05,
"loss": 0.0252,
"step": 2170
},
{
"epoch": 4.13,
"grad_norm": 0.6026468873023987,
"learning_rate": 3.975753465890656e-05,
"loss": 0.0257,
"step": 2171
},
{
"epoch": 4.13,
"grad_norm": 1.0865201950073242,
"learning_rate": 3.974006450207449e-05,
"loss": 0.0372,
"step": 2172
},
{
"epoch": 4.13,
"grad_norm": 1.1244971752166748,
"learning_rate": 3.972259065223096e-05,
"loss": 0.0312,
"step": 2173
},
{
"epoch": 4.13,
"grad_norm": 1.3925502300262451,
"learning_rate": 3.97051131160013e-05,
"loss": 0.0199,
"step": 2174
},
{
"epoch": 4.13,
"grad_norm": 0.5363264083862305,
"learning_rate": 3.9687631900012216e-05,
"loss": 0.0079,
"step": 2175
},
{
"epoch": 4.14,
"grad_norm": 0.6999717354774475,
"learning_rate": 3.9670147010891854e-05,
"loss": 0.0212,
"step": 2176
},
{
"epoch": 4.14,
"grad_norm": 0.7961714267730713,
"learning_rate": 3.965265845526971e-05,
"loss": 0.0179,
"step": 2177
},
{
"epoch": 4.14,
"grad_norm": 0.6794847846031189,
"learning_rate": 3.96351662397767e-05,
"loss": 0.022,
"step": 2178
},
{
"epoch": 4.14,
"grad_norm": 0.9816306233406067,
"learning_rate": 3.961767037104512e-05,
"loss": 0.0169,
"step": 2179
},
{
"epoch": 4.14,
"grad_norm": 0.9381592869758606,
"learning_rate": 3.960017085570862e-05,
"loss": 0.0244,
"step": 2180
},
{
"epoch": 4.15,
"grad_norm": 1.271181583404541,
"learning_rate": 3.9582667700402294e-05,
"loss": 0.0313,
"step": 2181
},
{
"epoch": 4.15,
"grad_norm": 0.44248637557029724,
"learning_rate": 3.956516091176255e-05,
"loss": 0.0096,
"step": 2182
},
{
"epoch": 4.15,
"grad_norm": 0.7948805689811707,
"learning_rate": 3.954765049642723e-05,
"loss": 0.0138,
"step": 2183
},
{
"epoch": 4.15,
"grad_norm": 1.1203492879867554,
"learning_rate": 3.953013646103552e-05,
"loss": 0.0235,
"step": 2184
},
{
"epoch": 4.15,
"grad_norm": 1.0503700971603394,
"learning_rate": 3.951261881222798e-05,
"loss": 0.0336,
"step": 2185
},
{
"epoch": 4.16,
"grad_norm": 0.7672589421272278,
"learning_rate": 3.949509755664655e-05,
"loss": 0.017,
"step": 2186
},
{
"epoch": 4.16,
"grad_norm": 0.8506165742874146,
"learning_rate": 3.947757270093453e-05,
"loss": 0.0284,
"step": 2187
},
{
"epoch": 4.16,
"grad_norm": 0.9379598498344421,
"learning_rate": 3.946004425173659e-05,
"loss": 0.0289,
"step": 2188
},
{
"epoch": 4.16,
"grad_norm": 0.6427805423736572,
"learning_rate": 3.944251221569876e-05,
"loss": 0.0133,
"step": 2189
},
{
"epoch": 4.16,
"grad_norm": 0.7319521307945251,
"learning_rate": 3.942497659946843e-05,
"loss": 0.0215,
"step": 2190
},
{
"epoch": 4.17,
"grad_norm": 0.7214787006378174,
"learning_rate": 3.940743740969435e-05,
"loss": 0.024,
"step": 2191
},
{
"epoch": 4.17,
"grad_norm": 0.5724555850028992,
"learning_rate": 3.9389894653026614e-05,
"loss": 0.0249,
"step": 2192
},
{
"epoch": 4.17,
"grad_norm": 0.8537221550941467,
"learning_rate": 3.937234833611669e-05,
"loss": 0.0301,
"step": 2193
},
{
"epoch": 4.17,
"grad_norm": 0.7909390926361084,
"learning_rate": 3.9354798465617365e-05,
"loss": 0.034,
"step": 2194
},
{
"epoch": 4.17,
"grad_norm": 0.6314064860343933,
"learning_rate": 3.933724504818281e-05,
"loss": 0.0121,
"step": 2195
},
{
"epoch": 4.17,
"grad_norm": 0.6617389917373657,
"learning_rate": 3.9319688090468516e-05,
"loss": 0.0143,
"step": 2196
},
{
"epoch": 4.18,
"grad_norm": 0.9619398713111877,
"learning_rate": 3.9302127599131325e-05,
"loss": 0.0312,
"step": 2197
},
{
"epoch": 4.18,
"grad_norm": 0.602358877658844,
"learning_rate": 3.92845635808294e-05,
"loss": 0.0255,
"step": 2198
},
{
"epoch": 4.18,
"grad_norm": 0.9547474980354309,
"learning_rate": 3.926699604222228e-05,
"loss": 0.0127,
"step": 2199
},
{
"epoch": 4.18,
"grad_norm": 0.7806571125984192,
"learning_rate": 3.924942498997081e-05,
"loss": 0.0186,
"step": 2200
},
{
"epoch": 4.18,
"eval_blimp_filtered_avg": 0.7102985074626865,
"eval_blimp_filtered_std": 0.005055190205109769,
"step": 2200
},
{
"epoch": 4.18,
"eval_blimp_supplement_avg": 0.7456896551724138,
"eval_blimp_supplement_std": 0.018874250250412932,
"step": 2200
},
{
"epoch": 4.18,
"eval_vqa_filtered_avg": 0.29,
"eval_vqa_filtered_std": 0.045604802157206845,
"step": 2200
},
{
"epoch": 4.18,
"eval_winoground_filtered_avg": 0.5,
"eval_winoground_filtered_std": 0.050251890762960605,
"step": 2200
},
{
"epoch": 4.18,
"grad_norm": 0.6022462844848633,
"learning_rate": 3.923185043073717e-05,
"loss": 0.0191,
"step": 2201
},
{
"epoch": 4.19,
"grad_norm": 1.2786915302276611,
"learning_rate": 3.9214272371184866e-05,
"loss": 0.0285,
"step": 2202
},
{
"epoch": 4.19,
"grad_norm": 0.9401484131813049,
"learning_rate": 3.919669081797875e-05,
"loss": 0.027,
"step": 2203
},
{
"epoch": 4.19,
"grad_norm": 0.5957093834877014,
"learning_rate": 3.917910577778498e-05,
"loss": 0.0225,
"step": 2204
},
{
"epoch": 4.19,
"grad_norm": 0.5970522165298462,
"learning_rate": 3.916151725727104e-05,
"loss": 0.0137,
"step": 2205
},
{
"epoch": 4.19,
"grad_norm": 1.0572301149368286,
"learning_rate": 3.914392526310575e-05,
"loss": 0.0156,
"step": 2206
},
{
"epoch": 4.2,
"grad_norm": 0.9380121231079102,
"learning_rate": 3.912632980195923e-05,
"loss": 0.0276,
"step": 2207
},
{
"epoch": 4.2,
"grad_norm": 0.6428413391113281,
"learning_rate": 3.91087308805029e-05,
"loss": 0.0157,
"step": 2208
},
{
"epoch": 4.2,
"grad_norm": 1.0269511938095093,
"learning_rate": 3.909112850540954e-05,
"loss": 0.0269,
"step": 2209
},
{
"epoch": 4.2,
"grad_norm": 0.9412804841995239,
"learning_rate": 3.907352268335319e-05,
"loss": 0.022,
"step": 2210
},
{
"epoch": 4.2,
"grad_norm": 0.7145057320594788,
"learning_rate": 3.905591342100923e-05,
"loss": 0.0209,
"step": 2211
},
{
"epoch": 4.21,
"grad_norm": 0.6359620094299316,
"learning_rate": 3.9038300725054316e-05,
"loss": 0.0206,
"step": 2212
},
{
"epoch": 4.21,
"grad_norm": 0.5400950908660889,
"learning_rate": 3.902068460216643e-05,
"loss": 0.0148,
"step": 2213
},
{
"epoch": 4.21,
"grad_norm": 0.9488170146942139,
"learning_rate": 3.9003065059024844e-05,
"loss": 0.0311,
"step": 2214
},
{
"epoch": 4.21,
"grad_norm": 0.9143253564834595,
"learning_rate": 3.898544210231011e-05,
"loss": 0.0206,
"step": 2215
},
{
"epoch": 4.21,
"grad_norm": 0.8775594234466553,
"learning_rate": 3.8967815738704126e-05,
"loss": 0.015,
"step": 2216
},
{
"epoch": 4.21,
"grad_norm": 0.7152052521705627,
"learning_rate": 3.895018597489002e-05,
"loss": 0.0181,
"step": 2217
},
{
"epoch": 4.22,
"grad_norm": 0.5914208889007568,
"learning_rate": 3.893255281755224e-05,
"loss": 0.0085,
"step": 2218
},
{
"epoch": 4.22,
"grad_norm": 1.0725632905960083,
"learning_rate": 3.891491627337653e-05,
"loss": 0.0208,
"step": 2219
},
{
"epoch": 4.22,
"grad_norm": 0.4632147550582886,
"learning_rate": 3.8897276349049896e-05,
"loss": 0.0092,
"step": 2220
},
{
"epoch": 4.22,
"grad_norm": 0.6670335531234741,
"learning_rate": 3.887963305126063e-05,
"loss": 0.0166,
"step": 2221
},
{
"epoch": 4.22,
"grad_norm": 0.7292976975440979,
"learning_rate": 3.886198638669831e-05,
"loss": 0.0107,
"step": 2222
},
{
"epoch": 4.23,
"grad_norm": 0.7694457769393921,
"learning_rate": 3.8844336362053807e-05,
"loss": 0.0224,
"step": 2223
},
{
"epoch": 4.23,
"grad_norm": 0.7484649419784546,
"learning_rate": 3.882668298401923e-05,
"loss": 0.0229,
"step": 2224
},
{
"epoch": 4.23,
"grad_norm": 0.6838115453720093,
"learning_rate": 3.880902625928797e-05,
"loss": 0.0182,
"step": 2225
},
{
"epoch": 4.23,
"grad_norm": 0.8893880844116211,
"learning_rate": 3.879136619455471e-05,
"loss": 0.0106,
"step": 2226
},
{
"epoch": 4.23,
"grad_norm": 0.9993874430656433,
"learning_rate": 3.877370279651538e-05,
"loss": 0.0307,
"step": 2227
},
{
"epoch": 4.24,
"grad_norm": 0.9051404595375061,
"learning_rate": 3.8756036071867166e-05,
"loss": 0.0175,
"step": 2228
},
{
"epoch": 4.24,
"grad_norm": 0.6256256103515625,
"learning_rate": 3.8738366027308534e-05,
"loss": 0.0207,
"step": 2229
},
{
"epoch": 4.24,
"grad_norm": 0.5313103795051575,
"learning_rate": 3.872069266953921e-05,
"loss": 0.0215,
"step": 2230
},
{
"epoch": 4.24,
"grad_norm": 0.728925883769989,
"learning_rate": 3.8703016005260145e-05,
"loss": 0.0139,
"step": 2231
},
{
"epoch": 4.24,
"grad_norm": 0.8880319595336914,
"learning_rate": 3.868533604117359e-05,
"loss": 0.0262,
"step": 2232
},
{
"epoch": 4.25,
"grad_norm": 0.6292276978492737,
"learning_rate": 3.866765278398301e-05,
"loss": 0.0238,
"step": 2233
},
{
"epoch": 4.25,
"grad_norm": 1.0122499465942383,
"learning_rate": 3.864996624039314e-05,
"loss": 0.0255,
"step": 2234
},
{
"epoch": 4.25,
"grad_norm": 0.6637560129165649,
"learning_rate": 3.8632276417109944e-05,
"loss": 0.0153,
"step": 2235
},
{
"epoch": 4.25,
"grad_norm": 0.41349828243255615,
"learning_rate": 3.861458332084064e-05,
"loss": 0.0075,
"step": 2236
},
{
"epoch": 4.25,
"grad_norm": 0.6367612481117249,
"learning_rate": 3.859688695829369e-05,
"loss": 0.0161,
"step": 2237
},
{
"epoch": 4.25,
"grad_norm": 0.5207117199897766,
"learning_rate": 3.857918733617879e-05,
"loss": 0.0078,
"step": 2238
},
{
"epoch": 4.26,
"grad_norm": 0.6043189764022827,
"learning_rate": 3.8561484461206854e-05,
"loss": 0.0087,
"step": 2239
},
{
"epoch": 4.26,
"grad_norm": 0.8773126602172852,
"learning_rate": 3.854377834009008e-05,
"loss": 0.0218,
"step": 2240
},
{
"epoch": 4.26,
"grad_norm": 0.28976312279701233,
"learning_rate": 3.8526068979541836e-05,
"loss": 0.0068,
"step": 2241
},
{
"epoch": 4.26,
"grad_norm": 0.8883360028266907,
"learning_rate": 3.850835638627675e-05,
"loss": 0.0152,
"step": 2242
},
{
"epoch": 4.26,
"grad_norm": 0.5347712635993958,
"learning_rate": 3.8490640567010695e-05,
"loss": 0.0175,
"step": 2243
},
{
"epoch": 4.27,
"grad_norm": 0.775051474571228,
"learning_rate": 3.8472921528460704e-05,
"loss": 0.0239,
"step": 2244
},
{
"epoch": 4.27,
"grad_norm": 0.5361863970756531,
"learning_rate": 3.845519927734512e-05,
"loss": 0.0061,
"step": 2245
},
{
"epoch": 4.27,
"grad_norm": 0.9159285426139832,
"learning_rate": 3.8437473820383406e-05,
"loss": 0.0145,
"step": 2246
},
{
"epoch": 4.27,
"grad_norm": 0.8836215734481812,
"learning_rate": 3.841974516429632e-05,
"loss": 0.0185,
"step": 2247
},
{
"epoch": 4.27,
"grad_norm": 1.2125509977340698,
"learning_rate": 3.840201331580579e-05,
"loss": 0.043,
"step": 2248
},
{
"epoch": 4.28,
"grad_norm": 1.3043692111968994,
"learning_rate": 3.838427828163497e-05,
"loss": 0.0294,
"step": 2249
},
{
"epoch": 4.28,
"grad_norm": 1.0944355726242065,
"learning_rate": 3.8366540068508223e-05,
"loss": 0.0192,
"step": 2250
},
{
"epoch": 4.28,
"grad_norm": 0.5080244541168213,
"learning_rate": 3.8348798683151095e-05,
"loss": 0.0088,
"step": 2251
},
{
"epoch": 4.28,
"grad_norm": 1.3902838230133057,
"learning_rate": 3.8331054132290375e-05,
"loss": 0.0248,
"step": 2252
},
{
"epoch": 4.28,
"grad_norm": 0.8109390139579773,
"learning_rate": 3.831330642265402e-05,
"loss": 0.0171,
"step": 2253
},
{
"epoch": 4.29,
"grad_norm": 0.8681051731109619,
"learning_rate": 3.829555556097119e-05,
"loss": 0.0186,
"step": 2254
},
{
"epoch": 4.29,
"grad_norm": 0.7938624620437622,
"learning_rate": 3.8277801553972244e-05,
"loss": 0.0183,
"step": 2255
},
{
"epoch": 4.29,
"grad_norm": 0.7191802263259888,
"learning_rate": 3.826004440838875e-05,
"loss": 0.0145,
"step": 2256
},
{
"epoch": 4.29,
"grad_norm": 0.6237900853157043,
"learning_rate": 3.824228413095343e-05,
"loss": 0.0108,
"step": 2257
},
{
"epoch": 4.29,
"grad_norm": 0.7636531591415405,
"learning_rate": 3.8224520728400224e-05,
"loss": 0.0174,
"step": 2258
},
{
"epoch": 4.29,
"grad_norm": 0.6067004799842834,
"learning_rate": 3.8206754207464256e-05,
"loss": 0.0106,
"step": 2259
},
{
"epoch": 4.3,
"grad_norm": 0.6134558916091919,
"learning_rate": 3.8188984574881796e-05,
"loss": 0.0163,
"step": 2260
},
{
"epoch": 4.3,
"grad_norm": 0.9270985126495361,
"learning_rate": 3.8171211837390346e-05,
"loss": 0.0159,
"step": 2261
},
{
"epoch": 4.3,
"grad_norm": 0.8644940257072449,
"learning_rate": 3.815343600172854e-05,
"loss": 0.0115,
"step": 2262
},
{
"epoch": 4.3,
"grad_norm": 0.6885161399841309,
"learning_rate": 3.813565707463623e-05,
"loss": 0.0059,
"step": 2263
},
{
"epoch": 4.3,
"grad_norm": 0.888508141040802,
"learning_rate": 3.811787506285439e-05,
"loss": 0.0138,
"step": 2264
},
{
"epoch": 4.31,
"grad_norm": 0.9297993779182434,
"learning_rate": 3.8100089973125214e-05,
"loss": 0.0289,
"step": 2265
},
{
"epoch": 4.31,
"grad_norm": 0.4766141176223755,
"learning_rate": 3.808230181219203e-05,
"loss": 0.013,
"step": 2266
},
{
"epoch": 4.31,
"grad_norm": 0.6746285557746887,
"learning_rate": 3.806451058679933e-05,
"loss": 0.0191,
"step": 2267
},
{
"epoch": 4.31,
"grad_norm": 0.9079482555389404,
"learning_rate": 3.804671630369279e-05,
"loss": 0.0224,
"step": 2268
},
{
"epoch": 4.31,
"grad_norm": 0.4020373225212097,
"learning_rate": 3.802891896961922e-05,
"loss": 0.0102,
"step": 2269
},
{
"epoch": 4.32,
"grad_norm": 0.5679680109024048,
"learning_rate": 3.801111859132663e-05,
"loss": 0.0068,
"step": 2270
},
{
"epoch": 4.32,
"grad_norm": 0.5791987776756287,
"learning_rate": 3.799331517556412e-05,
"loss": 0.015,
"step": 2271
},
{
"epoch": 4.32,
"grad_norm": 1.1974616050720215,
"learning_rate": 3.7975508729081994e-05,
"loss": 0.0217,
"step": 2272
},
{
"epoch": 4.32,
"grad_norm": 0.6375501155853271,
"learning_rate": 3.795769925863167e-05,
"loss": 0.0195,
"step": 2273
},
{
"epoch": 4.32,
"grad_norm": 0.6913089156150818,
"learning_rate": 3.793988677096576e-05,
"loss": 0.0162,
"step": 2274
},
{
"epoch": 4.33,
"grad_norm": 0.7762637138366699,
"learning_rate": 3.792207127283796e-05,
"loss": 0.0163,
"step": 2275
},
{
"epoch": 4.33,
"grad_norm": 0.812113881111145,
"learning_rate": 3.790425277100315e-05,
"loss": 0.0148,
"step": 2276
},
{
"epoch": 4.33,
"grad_norm": 0.9202824831008911,
"learning_rate": 3.788643127221732e-05,
"loss": 0.0339,
"step": 2277
},
{
"epoch": 4.33,
"grad_norm": 0.6353538036346436,
"learning_rate": 3.786860678323764e-05,
"loss": 0.0143,
"step": 2278
},
{
"epoch": 4.33,
"grad_norm": 0.9342000484466553,
"learning_rate": 3.7850779310822375e-05,
"loss": 0.0189,
"step": 2279
},
{
"epoch": 4.33,
"grad_norm": 0.32721957564353943,
"learning_rate": 3.783294886173093e-05,
"loss": 0.0044,
"step": 2280
},
{
"epoch": 4.34,
"grad_norm": 1.3283257484436035,
"learning_rate": 3.7815115442723824e-05,
"loss": 0.0176,
"step": 2281
},
{
"epoch": 4.34,
"grad_norm": 0.7589645981788635,
"learning_rate": 3.779727906056275e-05,
"loss": 0.0239,
"step": 2282
},
{
"epoch": 4.34,
"grad_norm": 0.9314877390861511,
"learning_rate": 3.7779439722010464e-05,
"loss": 0.0203,
"step": 2283
},
{
"epoch": 4.34,
"grad_norm": 0.8873271346092224,
"learning_rate": 3.776159743383088e-05,
"loss": 0.0125,
"step": 2284
},
{
"epoch": 4.34,
"grad_norm": 0.3702395260334015,
"learning_rate": 3.774375220278904e-05,
"loss": 0.0035,
"step": 2285
},
{
"epoch": 4.35,
"grad_norm": 0.6151421666145325,
"learning_rate": 3.772590403565107e-05,
"loss": 0.0124,
"step": 2286
},
{
"epoch": 4.35,
"grad_norm": 0.9653459787368774,
"learning_rate": 3.770805293918422e-05,
"loss": 0.0264,
"step": 2287
},
{
"epoch": 4.35,
"grad_norm": 1.0223948955535889,
"learning_rate": 3.769019892015686e-05,
"loss": 0.026,
"step": 2288
},
{
"epoch": 4.35,
"grad_norm": 0.6518673300743103,
"learning_rate": 3.7672341985338466e-05,
"loss": 0.0216,
"step": 2289
},
{
"epoch": 4.35,
"grad_norm": 1.2420117855072021,
"learning_rate": 3.765448214149959e-05,
"loss": 0.0255,
"step": 2290
},
{
"epoch": 4.36,
"grad_norm": 1.4792922735214233,
"learning_rate": 3.763661939541196e-05,
"loss": 0.0267,
"step": 2291
},
{
"epoch": 4.36,
"grad_norm": 0.6951161623001099,
"learning_rate": 3.761875375384832e-05,
"loss": 0.0131,
"step": 2292
},
{
"epoch": 4.36,
"grad_norm": 0.5609742999076843,
"learning_rate": 3.760088522358257e-05,
"loss": 0.0092,
"step": 2293
},
{
"epoch": 4.36,
"grad_norm": 0.5016159415245056,
"learning_rate": 3.758301381138967e-05,
"loss": 0.0121,
"step": 2294
},
{
"epoch": 4.36,
"grad_norm": 0.5049175024032593,
"learning_rate": 3.7565139524045695e-05,
"loss": 0.0091,
"step": 2295
},
{
"epoch": 4.37,
"grad_norm": 1.080764889717102,
"learning_rate": 3.754726236832779e-05,
"loss": 0.0284,
"step": 2296
},
{
"epoch": 4.37,
"grad_norm": 0.6445505023002625,
"learning_rate": 3.752938235101421e-05,
"loss": 0.0243,
"step": 2297
},
{
"epoch": 4.37,
"grad_norm": 0.6587095260620117,
"learning_rate": 3.75114994788843e-05,
"loss": 0.0165,
"step": 2298
},
{
"epoch": 4.37,
"grad_norm": 0.7057709693908691,
"learning_rate": 3.7493613758718445e-05,
"loss": 0.011,
"step": 2299
},
{
"epoch": 4.37,
"grad_norm": 0.8696507215499878,
"learning_rate": 3.747572519729814e-05,
"loss": 0.0205,
"step": 2300
},
{
"epoch": 4.37,
"eval_blimp_filtered_avg": 0.714776119402985,
"eval_blimp_filtered_std": 0.005026979869377555,
"step": 2300
},
{
"epoch": 4.37,
"eval_blimp_supplement_avg": 0.7672413793103449,
"eval_blimp_supplement_std": 0.018545497213800156,
"step": 2300
},
{
"epoch": 4.37,
"eval_vqa_filtered_avg": 0.39,
"eval_vqa_filtered_std": 0.04902071300001975,
"step": 2300
},
{
"epoch": 4.37,
"eval_winoground_filtered_avg": 0.52,
"eval_winoground_filtered_std": 0.05021167315686779,
"step": 2300
},
{
"epoch": 4.37,
"grad_norm": 0.6215664148330688,
"learning_rate": 3.745783380140597e-05,
"loss": 0.0126,
"step": 2301
},
{
"epoch": 4.38,
"grad_norm": 0.5077241659164429,
"learning_rate": 3.743993957782557e-05,
"loss": 0.0105,
"step": 2302
},
{
"epoch": 4.38,
"grad_norm": 0.7419450879096985,
"learning_rate": 3.742204253334166e-05,
"loss": 0.0182,
"step": 2303
},
{
"epoch": 4.38,
"grad_norm": 0.8969202041625977,
"learning_rate": 3.740414267474002e-05,
"loss": 0.0151,
"step": 2304
},
{
"epoch": 4.38,
"grad_norm": 0.673026978969574,
"learning_rate": 3.738624000880751e-05,
"loss": 0.0104,
"step": 2305
},
{
"epoch": 4.38,
"grad_norm": 0.8386924266815186,
"learning_rate": 3.7368334542332034e-05,
"loss": 0.022,
"step": 2306
},
{
"epoch": 4.39,
"grad_norm": 0.7275305986404419,
"learning_rate": 3.735042628210258e-05,
"loss": 0.0178,
"step": 2307
},
{
"epoch": 4.39,
"grad_norm": 0.4962146282196045,
"learning_rate": 3.7332515234909195e-05,
"loss": 0.0072,
"step": 2308
},
{
"epoch": 4.39,
"grad_norm": 0.9481974244117737,
"learning_rate": 3.7314601407542956e-05,
"loss": 0.0119,
"step": 2309
},
{
"epoch": 4.39,
"grad_norm": 0.5126856565475464,
"learning_rate": 3.7296684806796024e-05,
"loss": 0.0085,
"step": 2310
},
{
"epoch": 4.39,
"grad_norm": 0.5787120461463928,
"learning_rate": 3.72787654394616e-05,
"loss": 0.0165,
"step": 2311
},
{
"epoch": 4.4,
"grad_norm": 0.7442859411239624,
"learning_rate": 3.726084331233393e-05,
"loss": 0.0152,
"step": 2312
},
{
"epoch": 4.4,
"grad_norm": 0.9974928498268127,
"learning_rate": 3.724291843220831e-05,
"loss": 0.0139,
"step": 2313
},
{
"epoch": 4.4,
"grad_norm": 1.2986621856689453,
"learning_rate": 3.722499080588108e-05,
"loss": 0.0378,
"step": 2314
},
{
"epoch": 4.4,
"grad_norm": 0.5517431497573853,
"learning_rate": 3.720706044014962e-05,
"loss": 0.0102,
"step": 2315
},
{
"epoch": 4.4,
"grad_norm": 0.6805217266082764,
"learning_rate": 3.718912734181235e-05,
"loss": 0.0103,
"step": 2316
},
{
"epoch": 4.4,
"grad_norm": 0.738705575466156,
"learning_rate": 3.717119151766873e-05,
"loss": 0.018,
"step": 2317
},
{
"epoch": 4.41,
"grad_norm": 0.7503154873847961,
"learning_rate": 3.715325297451924e-05,
"loss": 0.0156,
"step": 2318
},
{
"epoch": 4.41,
"grad_norm": 1.6574206352233887,
"learning_rate": 3.71353117191654e-05,
"loss": 0.0425,
"step": 2319
},
{
"epoch": 4.41,
"grad_norm": 0.4708974063396454,
"learning_rate": 3.7117367758409775e-05,
"loss": 0.0151,
"step": 2320
},
{
"epoch": 4.41,
"grad_norm": 0.6024411916732788,
"learning_rate": 3.7099421099055926e-05,
"loss": 0.0152,
"step": 2321
},
{
"epoch": 4.41,
"grad_norm": 0.6361926794052124,
"learning_rate": 3.7081471747908444e-05,
"loss": 0.0196,
"step": 2322
},
{
"epoch": 4.42,
"grad_norm": 0.9504894614219666,
"learning_rate": 3.7063519711772974e-05,
"loss": 0.0149,
"step": 2323
},
{
"epoch": 4.42,
"grad_norm": 0.6873846650123596,
"learning_rate": 3.7045564997456124e-05,
"loss": 0.0168,
"step": 2324
},
{
"epoch": 4.42,
"grad_norm": 0.7538172602653503,
"learning_rate": 3.702760761176556e-05,
"loss": 0.0125,
"step": 2325
},
{
"epoch": 4.42,
"grad_norm": 0.5828614234924316,
"learning_rate": 3.7009647561509946e-05,
"loss": 0.0128,
"step": 2326
},
{
"epoch": 4.42,
"grad_norm": 0.48598381876945496,
"learning_rate": 3.699168485349895e-05,
"loss": 0.0082,
"step": 2327
},
{
"epoch": 4.43,
"grad_norm": 0.751524031162262,
"learning_rate": 3.697371949454325e-05,
"loss": 0.0169,
"step": 2328
},
{
"epoch": 4.43,
"grad_norm": 1.0344045162200928,
"learning_rate": 3.695575149145455e-05,
"loss": 0.0199,
"step": 2329
},
{
"epoch": 4.43,
"grad_norm": 0.6553796529769897,
"learning_rate": 3.693778085104553e-05,
"loss": 0.0106,
"step": 2330
},
{
"epoch": 4.43,
"grad_norm": 0.5290932059288025,
"learning_rate": 3.691980758012988e-05,
"loss": 0.01,
"step": 2331
},
{
"epoch": 4.43,
"grad_norm": 0.5171002745628357,
"learning_rate": 3.690183168552228e-05,
"loss": 0.0102,
"step": 2332
},
{
"epoch": 4.44,
"grad_norm": 0.42716488242149353,
"learning_rate": 3.688385317403844e-05,
"loss": 0.0069,
"step": 2333
},
{
"epoch": 4.44,
"grad_norm": 0.8648310303688049,
"learning_rate": 3.686587205249501e-05,
"loss": 0.0254,
"step": 2334
},
{
"epoch": 4.44,
"grad_norm": 0.7555816173553467,
"learning_rate": 3.684788832770966e-05,
"loss": 0.0167,
"step": 2335
},
{
"epoch": 4.44,
"grad_norm": 0.7447550296783447,
"learning_rate": 3.682990200650103e-05,
"loss": 0.0238,
"step": 2336
},
{
"epoch": 4.44,
"grad_norm": 0.7055520415306091,
"learning_rate": 3.681191309568878e-05,
"loss": 0.0206,
"step": 2337
},
{
"epoch": 4.44,
"grad_norm": 0.9553652405738831,
"learning_rate": 3.6793921602093514e-05,
"loss": 0.0232,
"step": 2338
},
{
"epoch": 4.45,
"grad_norm": 0.8372032046318054,
"learning_rate": 3.6775927532536826e-05,
"loss": 0.0217,
"step": 2339
},
{
"epoch": 4.45,
"grad_norm": 0.5850710868835449,
"learning_rate": 3.675793089384129e-05,
"loss": 0.0138,
"step": 2340
},
{
"epoch": 4.45,
"grad_norm": 0.9156349897384644,
"learning_rate": 3.673993169283046e-05,
"loss": 0.0168,
"step": 2341
},
{
"epoch": 4.45,
"grad_norm": 0.957348644733429,
"learning_rate": 3.672192993632884e-05,
"loss": 0.0251,
"step": 2342
},
{
"epoch": 4.45,
"grad_norm": 0.7744441032409668,
"learning_rate": 3.670392563116194e-05,
"loss": 0.0223,
"step": 2343
},
{
"epoch": 4.46,
"grad_norm": 0.8940926790237427,
"learning_rate": 3.66859187841562e-05,
"loss": 0.0195,
"step": 2344
},
{
"epoch": 4.46,
"grad_norm": 0.5279655456542969,
"learning_rate": 3.666790940213904e-05,
"loss": 0.0169,
"step": 2345
},
{
"epoch": 4.46,
"grad_norm": 0.7154120206832886,
"learning_rate": 3.664989749193884e-05,
"loss": 0.016,
"step": 2346
},
{
"epoch": 4.46,
"grad_norm": 0.32284796237945557,
"learning_rate": 3.663188306038494e-05,
"loss": 0.0106,
"step": 2347
},
{
"epoch": 4.46,
"grad_norm": 1.0852347612380981,
"learning_rate": 3.661386611430764e-05,
"loss": 0.0335,
"step": 2348
},
{
"epoch": 4.47,
"grad_norm": 0.6160424947738647,
"learning_rate": 3.659584666053817e-05,
"loss": 0.0072,
"step": 2349
},
{
"epoch": 4.47,
"grad_norm": 0.7414917945861816,
"learning_rate": 3.657782470590874e-05,
"loss": 0.0213,
"step": 2350
},
{
"epoch": 4.47,
"grad_norm": 0.8313021063804626,
"learning_rate": 3.6559800257252493e-05,
"loss": 0.01,
"step": 2351
},
{
"epoch": 4.47,
"grad_norm": 0.8849697113037109,
"learning_rate": 3.654177332140352e-05,
"loss": 0.0249,
"step": 2352
},
{
"epoch": 4.47,
"grad_norm": 0.7956735491752625,
"learning_rate": 3.6523743905196865e-05,
"loss": 0.0215,
"step": 2353
},
{
"epoch": 4.48,
"grad_norm": 1.0517834424972534,
"learning_rate": 3.6505712015468487e-05,
"loss": 0.0146,
"step": 2354
},
{
"epoch": 4.48,
"grad_norm": 0.5579755902290344,
"learning_rate": 3.648767765905531e-05,
"loss": 0.015,
"step": 2355
},
{
"epoch": 4.48,
"grad_norm": 0.843441367149353,
"learning_rate": 3.646964084279518e-05,
"loss": 0.0173,
"step": 2356
},
{
"epoch": 4.48,
"grad_norm": 0.8226516246795654,
"learning_rate": 3.645160157352688e-05,
"loss": 0.0213,
"step": 2357
},
{
"epoch": 4.48,
"grad_norm": 0.8201603889465332,
"learning_rate": 3.643355985809012e-05,
"loss": 0.0213,
"step": 2358
},
{
"epoch": 4.48,
"grad_norm": 1.334014892578125,
"learning_rate": 3.641551570332554e-05,
"loss": 0.0318,
"step": 2359
},
{
"epoch": 4.49,
"grad_norm": 0.6256591081619263,
"learning_rate": 3.6397469116074705e-05,
"loss": 0.0072,
"step": 2360
},
{
"epoch": 4.49,
"grad_norm": 0.7434335947036743,
"learning_rate": 3.6379420103180086e-05,
"loss": 0.0135,
"step": 2361
},
{
"epoch": 4.49,
"grad_norm": 0.5580800771713257,
"learning_rate": 3.636136867148511e-05,
"loss": 0.0166,
"step": 2362
},
{
"epoch": 4.49,
"grad_norm": 0.7984168529510498,
"learning_rate": 3.634331482783409e-05,
"loss": 0.0179,
"step": 2363
},
{
"epoch": 4.49,
"grad_norm": 0.7179736495018005,
"learning_rate": 3.6325258579072266e-05,
"loss": 0.0087,
"step": 2364
},
{
"epoch": 4.5,
"grad_norm": 0.7240386009216309,
"learning_rate": 3.630719993204577e-05,
"loss": 0.0224,
"step": 2365
},
{
"epoch": 4.5,
"grad_norm": 1.0392439365386963,
"learning_rate": 3.628913889360169e-05,
"loss": 0.027,
"step": 2366
},
{
"epoch": 4.5,
"grad_norm": 0.49123460054397583,
"learning_rate": 3.627107547058797e-05,
"loss": 0.012,
"step": 2367
},
{
"epoch": 4.5,
"grad_norm": 0.6300781965255737,
"learning_rate": 3.625300966985347e-05,
"loss": 0.009,
"step": 2368
},
{
"epoch": 4.5,
"grad_norm": 0.8128176927566528,
"learning_rate": 3.6234941498248e-05,
"loss": 0.01,
"step": 2369
},
{
"epoch": 4.51,
"grad_norm": 0.5441476702690125,
"learning_rate": 3.62168709626222e-05,
"loss": 0.0092,
"step": 2370
},
{
"epoch": 4.51,
"grad_norm": 0.9085813760757446,
"learning_rate": 3.619879806982764e-05,
"loss": 0.0226,
"step": 2371
},
{
"epoch": 4.51,
"grad_norm": 0.6555932760238647,
"learning_rate": 3.618072282671678e-05,
"loss": 0.0136,
"step": 2372
},
{
"epoch": 4.51,
"grad_norm": 1.1204301118850708,
"learning_rate": 3.6162645240142976e-05,
"loss": 0.0269,
"step": 2373
},
{
"epoch": 4.51,
"grad_norm": 0.7976241707801819,
"learning_rate": 3.614456531696046e-05,
"loss": 0.0138,
"step": 2374
},
{
"epoch": 4.52,
"grad_norm": 0.6329526901245117,
"learning_rate": 3.612648306402436e-05,
"loss": 0.0147,
"step": 2375
},
{
"epoch": 4.52,
"grad_norm": 0.7716643810272217,
"learning_rate": 3.610839848819069e-05,
"loss": 0.0192,
"step": 2376
},
{
"epoch": 4.52,
"grad_norm": 0.6423521041870117,
"learning_rate": 3.609031159631633e-05,
"loss": 0.0249,
"step": 2377
},
{
"epoch": 4.52,
"grad_norm": 0.6635744571685791,
"learning_rate": 3.607222239525905e-05,
"loss": 0.0127,
"step": 2378
},
{
"epoch": 4.52,
"grad_norm": 0.6144313216209412,
"learning_rate": 3.60541308918775e-05,
"loss": 0.0107,
"step": 2379
},
{
"epoch": 4.52,
"grad_norm": 0.7997187376022339,
"learning_rate": 3.60360370930312e-05,
"loss": 0.0214,
"step": 2380
},
{
"epoch": 4.53,
"grad_norm": 1.0279213190078735,
"learning_rate": 3.601794100558052e-05,
"loss": 0.0289,
"step": 2381
},
{
"epoch": 4.53,
"grad_norm": 0.47779718041419983,
"learning_rate": 3.5999842636386724e-05,
"loss": 0.0086,
"step": 2382
},
{
"epoch": 4.53,
"grad_norm": 0.7047209143638611,
"learning_rate": 3.598174199231194e-05,
"loss": 0.0175,
"step": 2383
},
{
"epoch": 4.53,
"grad_norm": 0.8702245354652405,
"learning_rate": 3.596363908021913e-05,
"loss": 0.0195,
"step": 2384
},
{
"epoch": 4.53,
"grad_norm": 0.8304027915000916,
"learning_rate": 3.5945533906972157e-05,
"loss": 0.015,
"step": 2385
},
{
"epoch": 4.54,
"grad_norm": 0.6494778394699097,
"learning_rate": 3.59274264794357e-05,
"loss": 0.0192,
"step": 2386
},
{
"epoch": 4.54,
"grad_norm": 0.35155418515205383,
"learning_rate": 3.590931680447533e-05,
"loss": 0.0046,
"step": 2387
},
{
"epoch": 4.54,
"grad_norm": 0.9364442825317383,
"learning_rate": 3.589120488895743e-05,
"loss": 0.0235,
"step": 2388
},
{
"epoch": 4.54,
"grad_norm": 0.9245356321334839,
"learning_rate": 3.587309073974928e-05,
"loss": 0.0184,
"step": 2389
},
{
"epoch": 4.54,
"grad_norm": 0.4553062915802002,
"learning_rate": 3.585497436371898e-05,
"loss": 0.0103,
"step": 2390
},
{
"epoch": 4.55,
"grad_norm": 1.0732531547546387,
"learning_rate": 3.583685576773545e-05,
"loss": 0.0405,
"step": 2391
},
{
"epoch": 4.55,
"grad_norm": 0.822386622428894,
"learning_rate": 3.581873495866852e-05,
"loss": 0.0331,
"step": 2392
},
{
"epoch": 4.55,
"grad_norm": 0.7858312129974365,
"learning_rate": 3.580061194338878e-05,
"loss": 0.0238,
"step": 2393
},
{
"epoch": 4.55,
"grad_norm": 0.9333310127258301,
"learning_rate": 3.578248672876772e-05,
"loss": 0.0266,
"step": 2394
},
{
"epoch": 4.55,
"grad_norm": 0.9160165786743164,
"learning_rate": 3.576435932167763e-05,
"loss": 0.0256,
"step": 2395
},
{
"epoch": 4.56,
"grad_norm": 0.7884373068809509,
"learning_rate": 3.574622972899163e-05,
"loss": 0.0236,
"step": 2396
},
{
"epoch": 4.56,
"grad_norm": 0.5851473808288574,
"learning_rate": 3.572809795758369e-05,
"loss": 0.0065,
"step": 2397
},
{
"epoch": 4.56,
"grad_norm": 0.5921894907951355,
"learning_rate": 3.570996401432858e-05,
"loss": 0.0087,
"step": 2398
},
{
"epoch": 4.56,
"grad_norm": 0.6487410068511963,
"learning_rate": 3.5691827906101925e-05,
"loss": 0.019,
"step": 2399
},
{
"epoch": 4.56,
"grad_norm": 0.4510621726512909,
"learning_rate": 3.567368963978015e-05,
"loss": 0.0085,
"step": 2400
},
{
"epoch": 4.56,
"eval_blimp_filtered_avg": 0.7111940298507463,
"eval_blimp_filtered_std": 0.005026263628709342,
"step": 2400
},
{
"epoch": 4.56,
"eval_blimp_supplement_avg": 0.7672413793103449,
"eval_blimp_supplement_std": 0.018423569598765968,
"step": 2400
},
{
"epoch": 4.56,
"eval_vqa_filtered_avg": 0.33,
"eval_vqa_filtered_std": 0.047258156262526045,
"step": 2400
},
{
"epoch": 4.56,
"eval_winoground_filtered_avg": 0.53,
"eval_winoground_filtered_std": 0.05016135580465919,
"step": 2400
},
{
"epoch": 4.56,
"grad_norm": 0.6274715065956116,
"learning_rate": 3.565554922224048e-05,
"loss": 0.0179,
"step": 2401
},
{
"epoch": 4.57,
"grad_norm": 0.9307816028594971,
"learning_rate": 3.5637406660360996e-05,
"loss": 0.0161,
"step": 2402
},
{
"epoch": 4.57,
"grad_norm": 0.8340302109718323,
"learning_rate": 3.561926196102058e-05,
"loss": 0.0206,
"step": 2403
},
{
"epoch": 4.57,
"grad_norm": 0.6703003644943237,
"learning_rate": 3.56011151310989e-05,
"loss": 0.0286,
"step": 2404
},
{
"epoch": 4.57,
"grad_norm": 0.959135115146637,
"learning_rate": 3.558296617747647e-05,
"loss": 0.0149,
"step": 2405
},
{
"epoch": 4.57,
"grad_norm": 0.5294992923736572,
"learning_rate": 3.556481510703457e-05,
"loss": 0.008,
"step": 2406
},
{
"epoch": 4.58,
"grad_norm": 0.5004578828811646,
"learning_rate": 3.554666192665531e-05,
"loss": 0.0192,
"step": 2407
},
{
"epoch": 4.58,
"grad_norm": 0.7217199802398682,
"learning_rate": 3.5528506643221594e-05,
"loss": 0.0135,
"step": 2408
},
{
"epoch": 4.58,
"grad_norm": 0.488598495721817,
"learning_rate": 3.5510349263617116e-05,
"loss": 0.0195,
"step": 2409
},
{
"epoch": 4.58,
"grad_norm": 0.5809155702590942,
"learning_rate": 3.549218979472636e-05,
"loss": 0.0149,
"step": 2410
},
{
"epoch": 4.58,
"grad_norm": 0.6363112926483154,
"learning_rate": 3.547402824343463e-05,
"loss": 0.0116,
"step": 2411
},
{
"epoch": 4.59,
"grad_norm": 0.5268595814704895,
"learning_rate": 3.5455864616627994e-05,
"loss": 0.0138,
"step": 2412
},
{
"epoch": 4.59,
"grad_norm": 0.4714985191822052,
"learning_rate": 3.543769892119331e-05,
"loss": 0.0099,
"step": 2413
},
{
"epoch": 4.59,
"grad_norm": 0.7840426564216614,
"learning_rate": 3.541953116401822e-05,
"loss": 0.0178,
"step": 2414
},
{
"epoch": 4.59,
"grad_norm": 0.646496057510376,
"learning_rate": 3.540136135199117e-05,
"loss": 0.0137,
"step": 2415
},
{
"epoch": 4.59,
"grad_norm": 0.5730254054069519,
"learning_rate": 3.538318949200136e-05,
"loss": 0.0175,
"step": 2416
},
{
"epoch": 4.6,
"grad_norm": 0.8631293177604675,
"learning_rate": 3.536501559093876e-05,
"loss": 0.015,
"step": 2417
},
{
"epoch": 4.6,
"grad_norm": 0.4079771339893341,
"learning_rate": 3.534683965569415e-05,
"loss": 0.0067,
"step": 2418
},
{
"epoch": 4.6,
"grad_norm": 1.2555129528045654,
"learning_rate": 3.5328661693159044e-05,
"loss": 0.0296,
"step": 2419
},
{
"epoch": 4.6,
"grad_norm": 0.8264919519424438,
"learning_rate": 3.5310481710225744e-05,
"loss": 0.0183,
"step": 2420
},
{
"epoch": 4.6,
"grad_norm": 0.5502712726593018,
"learning_rate": 3.529229971378731e-05,
"loss": 0.0172,
"step": 2421
},
{
"epoch": 4.6,
"grad_norm": 0.6614871621131897,
"learning_rate": 3.5274115710737576e-05,
"loss": 0.0203,
"step": 2422
},
{
"epoch": 4.61,
"grad_norm": 0.49781695008277893,
"learning_rate": 3.525592970797113e-05,
"loss": 0.0053,
"step": 2423
},
{
"epoch": 4.61,
"grad_norm": 0.871432363986969,
"learning_rate": 3.5237741712383306e-05,
"loss": 0.0342,
"step": 2424
},
{
"epoch": 4.61,
"grad_norm": 0.7911785840988159,
"learning_rate": 3.521955173087022e-05,
"loss": 0.0236,
"step": 2425
},
{
"epoch": 4.61,
"grad_norm": 0.5860491394996643,
"learning_rate": 3.5201359770328726e-05,
"loss": 0.0136,
"step": 2426
},
{
"epoch": 4.61,
"grad_norm": 1.5729241371154785,
"learning_rate": 3.518316583765642e-05,
"loss": 0.0095,
"step": 2427
},
{
"epoch": 4.62,
"grad_norm": 0.47271475195884705,
"learning_rate": 3.5164969939751655e-05,
"loss": 0.0117,
"step": 2428
},
{
"epoch": 4.62,
"grad_norm": 0.7784184813499451,
"learning_rate": 3.514677208351354e-05,
"loss": 0.0151,
"step": 2429
},
{
"epoch": 4.62,
"grad_norm": 0.7161446809768677,
"learning_rate": 3.512857227584191e-05,
"loss": 0.0158,
"step": 2430
},
{
"epoch": 4.62,
"grad_norm": 0.5870130062103271,
"learning_rate": 3.511037052363734e-05,
"loss": 0.0257,
"step": 2431
},
{
"epoch": 4.62,
"grad_norm": 0.5369093418121338,
"learning_rate": 3.509216683380115e-05,
"loss": 0.0074,
"step": 2432
},
{
"epoch": 4.63,
"grad_norm": 0.8127210140228271,
"learning_rate": 3.5073961213235405e-05,
"loss": 0.0185,
"step": 2433
},
{
"epoch": 4.63,
"grad_norm": 1.1477314233779907,
"learning_rate": 3.5055753668842865e-05,
"loss": 0.0214,
"step": 2434
},
{
"epoch": 4.63,
"grad_norm": 0.6497639417648315,
"learning_rate": 3.503754420752707e-05,
"loss": 0.0197,
"step": 2435
},
{
"epoch": 4.63,
"grad_norm": 0.866664707660675,
"learning_rate": 3.501933283619224e-05,
"loss": 0.0257,
"step": 2436
},
{
"epoch": 4.63,
"grad_norm": 0.7515466213226318,
"learning_rate": 3.500111956174335e-05,
"loss": 0.0109,
"step": 2437
},
{
"epoch": 4.63,
"grad_norm": 0.7075735926628113,
"learning_rate": 3.498290439108609e-05,
"loss": 0.0157,
"step": 2438
},
{
"epoch": 4.64,
"grad_norm": 1.0329500436782837,
"learning_rate": 3.4964687331126855e-05,
"loss": 0.0204,
"step": 2439
},
{
"epoch": 4.64,
"grad_norm": 0.5367659330368042,
"learning_rate": 3.494646838877277e-05,
"loss": 0.0123,
"step": 2440
},
{
"epoch": 4.64,
"grad_norm": 1.0324910879135132,
"learning_rate": 3.492824757093168e-05,
"loss": 0.0224,
"step": 2441
},
{
"epoch": 4.64,
"grad_norm": 0.65562903881073,
"learning_rate": 3.491002488451212e-05,
"loss": 0.0087,
"step": 2442
},
{
"epoch": 4.64,
"grad_norm": 0.9763492941856384,
"learning_rate": 3.489180033642334e-05,
"loss": 0.0187,
"step": 2443
},
{
"epoch": 4.65,
"grad_norm": 0.7696658968925476,
"learning_rate": 3.4873573933575324e-05,
"loss": 0.0116,
"step": 2444
},
{
"epoch": 4.65,
"grad_norm": 0.4130136966705322,
"learning_rate": 3.4855345682878716e-05,
"loss": 0.0092,
"step": 2445
},
{
"epoch": 4.65,
"grad_norm": 0.7712420225143433,
"learning_rate": 3.483711559124488e-05,
"loss": 0.0225,
"step": 2446
},
{
"epoch": 4.65,
"grad_norm": 0.5408477187156677,
"learning_rate": 3.4818883665585886e-05,
"loss": 0.0116,
"step": 2447
},
{
"epoch": 4.65,
"grad_norm": 0.6969491839408875,
"learning_rate": 3.48006499128145e-05,
"loss": 0.0189,
"step": 2448
},
{
"epoch": 4.66,
"grad_norm": 0.6516832113265991,
"learning_rate": 3.4782414339844166e-05,
"loss": 0.0095,
"step": 2449
},
{
"epoch": 4.66,
"grad_norm": 1.105002522468567,
"learning_rate": 3.476417695358902e-05,
"loss": 0.0106,
"step": 2450
},
{
"epoch": 4.66,
"grad_norm": 0.5930959582328796,
"learning_rate": 3.474593776096389e-05,
"loss": 0.0176,
"step": 2451
},
{
"epoch": 4.66,
"grad_norm": 0.4453127682209015,
"learning_rate": 3.4727696768884304e-05,
"loss": 0.0106,
"step": 2452
},
{
"epoch": 4.66,
"grad_norm": 0.6314166188240051,
"learning_rate": 3.470945398426644e-05,
"loss": 0.0221,
"step": 2453
},
{
"epoch": 4.67,
"grad_norm": 0.8016642332077026,
"learning_rate": 3.4691209414027186e-05,
"loss": 0.0129,
"step": 2454
},
{
"epoch": 4.67,
"grad_norm": 0.5322256684303284,
"learning_rate": 3.4672963065084084e-05,
"loss": 0.0118,
"step": 2455
},
{
"epoch": 4.67,
"grad_norm": 0.7646715641021729,
"learning_rate": 3.465471494435536e-05,
"loss": 0.0162,
"step": 2456
},
{
"epoch": 4.67,
"grad_norm": 0.7809215784072876,
"learning_rate": 3.463646505875994e-05,
"loss": 0.0168,
"step": 2457
},
{
"epoch": 4.67,
"grad_norm": 0.521217405796051,
"learning_rate": 3.4618213415217356e-05,
"loss": 0.014,
"step": 2458
},
{
"epoch": 4.67,
"grad_norm": 0.841955304145813,
"learning_rate": 3.459996002064786e-05,
"loss": 0.0141,
"step": 2459
},
{
"epoch": 4.68,
"grad_norm": 0.897569477558136,
"learning_rate": 3.4581704881972344e-05,
"loss": 0.0133,
"step": 2460
},
{
"epoch": 4.68,
"grad_norm": 1.1701143980026245,
"learning_rate": 3.456344800611238e-05,
"loss": 0.022,
"step": 2461
},
{
"epoch": 4.68,
"grad_norm": 0.6980761289596558,
"learning_rate": 3.454518939999017e-05,
"loss": 0.0162,
"step": 2462
},
{
"epoch": 4.68,
"grad_norm": 0.5344531536102295,
"learning_rate": 3.45269290705286e-05,
"loss": 0.0129,
"step": 2463
},
{
"epoch": 4.68,
"grad_norm": 0.7089115381240845,
"learning_rate": 3.450866702465119e-05,
"loss": 0.0096,
"step": 2464
},
{
"epoch": 4.69,
"grad_norm": 0.42874494194984436,
"learning_rate": 3.449040326928213e-05,
"loss": 0.008,
"step": 2465
},
{
"epoch": 4.69,
"grad_norm": 0.3710055649280548,
"learning_rate": 3.4472137811346215e-05,
"loss": 0.0175,
"step": 2466
},
{
"epoch": 4.69,
"grad_norm": 0.6988310217857361,
"learning_rate": 3.4453870657768955e-05,
"loss": 0.0077,
"step": 2467
},
{
"epoch": 4.69,
"grad_norm": 0.9593952298164368,
"learning_rate": 3.443560181547646e-05,
"loss": 0.022,
"step": 2468
},
{
"epoch": 4.69,
"grad_norm": 0.2613586187362671,
"learning_rate": 3.441733129139546e-05,
"loss": 0.0046,
"step": 2469
},
{
"epoch": 4.7,
"grad_norm": 0.7432543635368347,
"learning_rate": 3.439905909245337e-05,
"loss": 0.0155,
"step": 2470
},
{
"epoch": 4.7,
"grad_norm": 0.8921736478805542,
"learning_rate": 3.4380785225578217e-05,
"loss": 0.0182,
"step": 2471
},
{
"epoch": 4.7,
"grad_norm": 0.7975713610649109,
"learning_rate": 3.436250969769865e-05,
"loss": 0.011,
"step": 2472
},
{
"epoch": 4.7,
"grad_norm": 0.27677029371261597,
"learning_rate": 3.4344232515743965e-05,
"loss": 0.006,
"step": 2473
},
{
"epoch": 4.7,
"grad_norm": 0.728541910648346,
"learning_rate": 3.432595368664408e-05,
"loss": 0.0127,
"step": 2474
},
{
"epoch": 4.71,
"grad_norm": 0.6335504055023193,
"learning_rate": 3.430767321732954e-05,
"loss": 0.0107,
"step": 2475
},
{
"epoch": 4.71,
"grad_norm": 0.8502057790756226,
"learning_rate": 3.4289391114731507e-05,
"loss": 0.0165,
"step": 2476
},
{
"epoch": 4.71,
"grad_norm": 0.9295932054519653,
"learning_rate": 3.427110738578176e-05,
"loss": 0.0177,
"step": 2477
},
{
"epoch": 4.71,
"grad_norm": 0.7788729667663574,
"learning_rate": 3.425282203741271e-05,
"loss": 0.0168,
"step": 2478
},
{
"epoch": 4.71,
"grad_norm": 0.9282010197639465,
"learning_rate": 3.4234535076557346e-05,
"loss": 0.0253,
"step": 2479
},
{
"epoch": 4.71,
"grad_norm": 1.1082226037979126,
"learning_rate": 3.421624651014932e-05,
"loss": 0.0164,
"step": 2480
},
{
"epoch": 4.72,
"grad_norm": 0.8360443115234375,
"learning_rate": 3.4197956345122856e-05,
"loss": 0.0134,
"step": 2481
},
{
"epoch": 4.72,
"grad_norm": 0.4351302683353424,
"learning_rate": 3.417966458841279e-05,
"loss": 0.0103,
"step": 2482
},
{
"epoch": 4.72,
"grad_norm": 1.0798381567001343,
"learning_rate": 3.4161371246954565e-05,
"loss": 0.0239,
"step": 2483
},
{
"epoch": 4.72,
"grad_norm": 0.2297414392232895,
"learning_rate": 3.4143076327684246e-05,
"loss": 0.005,
"step": 2484
},
{
"epoch": 4.72,
"grad_norm": 0.5078562498092651,
"learning_rate": 3.4124779837538445e-05,
"loss": 0.0076,
"step": 2485
},
{
"epoch": 4.73,
"grad_norm": 0.9148598909378052,
"learning_rate": 3.410648178345442e-05,
"loss": 0.0159,
"step": 2486
},
{
"epoch": 4.73,
"grad_norm": 0.607685923576355,
"learning_rate": 3.408818217236999e-05,
"loss": 0.0108,
"step": 2487
},
{
"epoch": 4.73,
"grad_norm": 0.7032291293144226,
"learning_rate": 3.406988101122359e-05,
"loss": 0.0147,
"step": 2488
},
{
"epoch": 4.73,
"grad_norm": 0.7176268696784973,
"learning_rate": 3.4051578306954214e-05,
"loss": 0.0113,
"step": 2489
},
{
"epoch": 4.73,
"grad_norm": 0.6715800762176514,
"learning_rate": 3.403327406650147e-05,
"loss": 0.0119,
"step": 2490
},
{
"epoch": 4.74,
"grad_norm": 0.9420292377471924,
"learning_rate": 3.401496829680553e-05,
"loss": 0.0132,
"step": 2491
},
{
"epoch": 4.74,
"grad_norm": 0.6202637553215027,
"learning_rate": 3.399666100480714e-05,
"loss": 0.0117,
"step": 2492
},
{
"epoch": 4.74,
"grad_norm": 0.5467507839202881,
"learning_rate": 3.3978352197447664e-05,
"loss": 0.0194,
"step": 2493
},
{
"epoch": 4.74,
"grad_norm": 0.7403022646903992,
"learning_rate": 3.396004188166898e-05,
"loss": 0.0145,
"step": 2494
},
{
"epoch": 4.74,
"grad_norm": 1.1096773147583008,
"learning_rate": 3.394173006441358e-05,
"loss": 0.0255,
"step": 2495
},
{
"epoch": 4.75,
"grad_norm": 0.49350011348724365,
"learning_rate": 3.3923416752624496e-05,
"loss": 0.0071,
"step": 2496
},
{
"epoch": 4.75,
"grad_norm": 1.0116976499557495,
"learning_rate": 3.390510195324538e-05,
"loss": 0.0195,
"step": 2497
},
{
"epoch": 4.75,
"grad_norm": 0.5441023707389832,
"learning_rate": 3.388678567322039e-05,
"loss": 0.0079,
"step": 2498
},
{
"epoch": 4.75,
"grad_norm": 0.824561357498169,
"learning_rate": 3.386846791949427e-05,
"loss": 0.0156,
"step": 2499
},
{
"epoch": 4.75,
"grad_norm": 0.6066474914550781,
"learning_rate": 3.385014869901232e-05,
"loss": 0.0101,
"step": 2500
},
{
"epoch": 4.75,
"eval_blimp_filtered_avg": 0.7156716417910448,
"eval_blimp_filtered_std": 0.005020440385347697,
"step": 2500
},
{
"epoch": 4.75,
"eval_blimp_supplement_avg": 0.7823275862068966,
"eval_blimp_supplement_std": 0.018054146891183953,
"step": 2500
},
{
"epoch": 4.75,
"eval_vqa_filtered_avg": 0.4,
"eval_vqa_filtered_std": 0.04923659639173309,
"step": 2500
},
{
"epoch": 4.75,
"eval_winoground_filtered_avg": 0.52,
"eval_winoground_filtered_std": 0.05021167315686779,
"step": 2500
},
{
"epoch": 4.75,
"grad_norm": 0.6806913614273071,
"learning_rate": 3.3831828018720404e-05,
"loss": 0.0051,
"step": 2501
},
{
"epoch": 4.76,
"grad_norm": 0.5570813417434692,
"learning_rate": 3.381350588556491e-05,
"loss": 0.0087,
"step": 2502
},
{
"epoch": 4.76,
"grad_norm": 0.5015181303024292,
"learning_rate": 3.379518230649282e-05,
"loss": 0.0078,
"step": 2503
},
{
"epoch": 4.76,
"grad_norm": 0.8748692274093628,
"learning_rate": 3.377685728845164e-05,
"loss": 0.0283,
"step": 2504
},
{
"epoch": 4.76,
"grad_norm": 0.530643105506897,
"learning_rate": 3.3758530838389424e-05,
"loss": 0.0139,
"step": 2505
},
{
"epoch": 4.76,
"grad_norm": 0.8141795992851257,
"learning_rate": 3.374020296325477e-05,
"loss": 0.022,
"step": 2506
},
{
"epoch": 4.77,
"grad_norm": 0.7700246572494507,
"learning_rate": 3.372187366999681e-05,
"loss": 0.0142,
"step": 2507
},
{
"epoch": 4.77,
"grad_norm": 1.0700180530548096,
"learning_rate": 3.3703542965565235e-05,
"loss": 0.0234,
"step": 2508
},
{
"epoch": 4.77,
"grad_norm": 1.9604212045669556,
"learning_rate": 3.3685210856910234e-05,
"loss": 0.0216,
"step": 2509
},
{
"epoch": 4.77,
"grad_norm": 0.6713403463363647,
"learning_rate": 3.366687735098257e-05,
"loss": 0.0162,
"step": 2510
},
{
"epoch": 4.77,
"grad_norm": 0.6530247330665588,
"learning_rate": 3.3648542454733496e-05,
"loss": 0.0116,
"step": 2511
},
{
"epoch": 4.78,
"grad_norm": 1.0282354354858398,
"learning_rate": 3.363020617511483e-05,
"loss": 0.0108,
"step": 2512
},
{
"epoch": 4.78,
"grad_norm": 0.6313475966453552,
"learning_rate": 3.3611868519078895e-05,
"loss": 0.0161,
"step": 2513
},
{
"epoch": 4.78,
"grad_norm": 0.6454443335533142,
"learning_rate": 3.359352949357852e-05,
"loss": 0.0257,
"step": 2514
},
{
"epoch": 4.78,
"grad_norm": 0.6797105073928833,
"learning_rate": 3.3575189105567085e-05,
"loss": 0.0102,
"step": 2515
},
{
"epoch": 4.78,
"grad_norm": 1.3944296836853027,
"learning_rate": 3.3556847361998475e-05,
"loss": 0.0143,
"step": 2516
},
{
"epoch": 4.79,
"grad_norm": 0.7667531371116638,
"learning_rate": 3.353850426982709e-05,
"loss": 0.0222,
"step": 2517
},
{
"epoch": 4.79,
"grad_norm": 1.010321855545044,
"learning_rate": 3.3520159836007817e-05,
"loss": 0.0225,
"step": 2518
},
{
"epoch": 4.79,
"grad_norm": 0.9052011370658875,
"learning_rate": 3.3501814067496095e-05,
"loss": 0.015,
"step": 2519
},
{
"epoch": 4.79,
"grad_norm": 0.767512321472168,
"learning_rate": 3.3483466971247834e-05,
"loss": 0.0246,
"step": 2520
},
{
"epoch": 4.79,
"grad_norm": 0.922870397567749,
"learning_rate": 3.3465118554219466e-05,
"loss": 0.017,
"step": 2521
},
{
"epoch": 4.79,
"grad_norm": 0.5921939611434937,
"learning_rate": 3.344676882336791e-05,
"loss": 0.0094,
"step": 2522
},
{
"epoch": 4.8,
"grad_norm": 0.6729060411453247,
"learning_rate": 3.34284177856506e-05,
"loss": 0.0167,
"step": 2523
},
{
"epoch": 4.8,
"grad_norm": 0.8896371722221375,
"learning_rate": 3.341006544802545e-05,
"loss": 0.0133,
"step": 2524
},
{
"epoch": 4.8,
"grad_norm": 0.7445226311683655,
"learning_rate": 3.339171181745087e-05,
"loss": 0.0159,
"step": 2525
},
{
"epoch": 4.8,
"grad_norm": 0.7400672435760498,
"learning_rate": 3.3373356900885765e-05,
"loss": 0.0144,
"step": 2526
},
{
"epoch": 4.8,
"grad_norm": 1.0611234903335571,
"learning_rate": 3.335500070528955e-05,
"loss": 0.0318,
"step": 2527
},
{
"epoch": 4.81,
"grad_norm": 0.9864000678062439,
"learning_rate": 3.3336643237622055e-05,
"loss": 0.0167,
"step": 2528
},
{
"epoch": 4.81,
"grad_norm": 0.8758773803710938,
"learning_rate": 3.331828450484367e-05,
"loss": 0.0134,
"step": 2529
},
{
"epoch": 4.81,
"grad_norm": 1.00602126121521,
"learning_rate": 3.329992451391523e-05,
"loss": 0.0233,
"step": 2530
},
{
"epoch": 4.81,
"grad_norm": 0.7682434916496277,
"learning_rate": 3.328156327179805e-05,
"loss": 0.0206,
"step": 2531
},
{
"epoch": 4.81,
"grad_norm": 0.6251866817474365,
"learning_rate": 3.326320078545392e-05,
"loss": 0.0103,
"step": 2532
},
{
"epoch": 4.82,
"grad_norm": 0.7585045099258423,
"learning_rate": 3.3244837061845086e-05,
"loss": 0.0147,
"step": 2533
},
{
"epoch": 4.82,
"grad_norm": 0.5303366780281067,
"learning_rate": 3.322647210793429e-05,
"loss": 0.0119,
"step": 2534
},
{
"epoch": 4.82,
"grad_norm": 0.607226550579071,
"learning_rate": 3.3208105930684734e-05,
"loss": 0.0132,
"step": 2535
},
{
"epoch": 4.82,
"grad_norm": 0.7435906529426575,
"learning_rate": 3.3189738537060064e-05,
"loss": 0.0179,
"step": 2536
},
{
"epoch": 4.82,
"grad_norm": 0.4219205379486084,
"learning_rate": 3.317136993402442e-05,
"loss": 0.0183,
"step": 2537
},
{
"epoch": 4.83,
"grad_norm": 1.0047707557678223,
"learning_rate": 3.315300012854236e-05,
"loss": 0.0401,
"step": 2538
},
{
"epoch": 4.83,
"grad_norm": 0.7867347002029419,
"learning_rate": 3.313462912757894e-05,
"loss": 0.017,
"step": 2539
},
{
"epoch": 4.83,
"grad_norm": 0.4949563443660736,
"learning_rate": 3.311625693809965e-05,
"loss": 0.0113,
"step": 2540
},
{
"epoch": 4.83,
"grad_norm": 0.7276683449745178,
"learning_rate": 3.3097883567070405e-05,
"loss": 0.0221,
"step": 2541
},
{
"epoch": 4.83,
"grad_norm": 1.0241750478744507,
"learning_rate": 3.307950902145763e-05,
"loss": 0.0251,
"step": 2542
},
{
"epoch": 4.83,
"grad_norm": 1.0852843523025513,
"learning_rate": 3.3061133308228125e-05,
"loss": 0.0312,
"step": 2543
},
{
"epoch": 4.84,
"grad_norm": 0.8237437605857849,
"learning_rate": 3.3042756434349186e-05,
"loss": 0.0127,
"step": 2544
},
{
"epoch": 4.84,
"grad_norm": 1.2978965044021606,
"learning_rate": 3.302437840678852e-05,
"loss": 0.0346,
"step": 2545
},
{
"epoch": 4.84,
"grad_norm": 0.4865902066230774,
"learning_rate": 3.30059992325143e-05,
"loss": 0.0102,
"step": 2546
},
{
"epoch": 4.84,
"grad_norm": 0.7701038122177124,
"learning_rate": 3.298761891849508e-05,
"loss": 0.0187,
"step": 2547
},
{
"epoch": 4.84,
"grad_norm": 0.685307502746582,
"learning_rate": 3.2969237471699906e-05,
"loss": 0.0186,
"step": 2548
},
{
"epoch": 4.85,
"grad_norm": 0.7615230679512024,
"learning_rate": 3.2950854899098224e-05,
"loss": 0.0209,
"step": 2549
},
{
"epoch": 4.85,
"grad_norm": 0.6191115379333496,
"learning_rate": 3.29324712076599e-05,
"loss": 0.0084,
"step": 2550
},
{
"epoch": 4.85,
"grad_norm": 0.8195491433143616,
"learning_rate": 3.2914086404355246e-05,
"loss": 0.0167,
"step": 2551
},
{
"epoch": 4.85,
"grad_norm": 0.719419002532959,
"learning_rate": 3.289570049615499e-05,
"loss": 0.0102,
"step": 2552
},
{
"epoch": 4.85,
"grad_norm": 0.9063612818717957,
"learning_rate": 3.287731349003026e-05,
"loss": 0.0118,
"step": 2553
},
{
"epoch": 4.86,
"grad_norm": 1.145775556564331,
"learning_rate": 3.2858925392952624e-05,
"loss": 0.0156,
"step": 2554
},
{
"epoch": 4.86,
"grad_norm": 0.7481817603111267,
"learning_rate": 3.284053621189405e-05,
"loss": 0.0202,
"step": 2555
},
{
"epoch": 4.86,
"grad_norm": 0.9012960195541382,
"learning_rate": 3.2822145953826924e-05,
"loss": 0.0135,
"step": 2556
},
{
"epoch": 4.86,
"grad_norm": 0.8570747375488281,
"learning_rate": 3.280375462572403e-05,
"loss": 0.0196,
"step": 2557
},
{
"epoch": 4.86,
"grad_norm": 0.5664126873016357,
"learning_rate": 3.278536223455856e-05,
"loss": 0.0221,
"step": 2558
},
{
"epoch": 4.87,
"grad_norm": 0.6674412488937378,
"learning_rate": 3.2766968787304127e-05,
"loss": 0.0127,
"step": 2559
},
{
"epoch": 4.87,
"grad_norm": 0.846566379070282,
"learning_rate": 3.2748574290934725e-05,
"loss": 0.0245,
"step": 2560
},
{
"epoch": 4.87,
"grad_norm": 0.8954035043716431,
"learning_rate": 3.2730178752424754e-05,
"loss": 0.0355,
"step": 2561
},
{
"epoch": 4.87,
"grad_norm": 0.7568804621696472,
"learning_rate": 3.271178217874901e-05,
"loss": 0.0181,
"step": 2562
},
{
"epoch": 4.87,
"grad_norm": 0.37463417649269104,
"learning_rate": 3.269338457688267e-05,
"loss": 0.0048,
"step": 2563
},
{
"epoch": 4.87,
"grad_norm": 0.5295295715332031,
"learning_rate": 3.267498595380131e-05,
"loss": 0.0148,
"step": 2564
},
{
"epoch": 4.88,
"grad_norm": 1.0253456830978394,
"learning_rate": 3.265658631648091e-05,
"loss": 0.0148,
"step": 2565
},
{
"epoch": 4.88,
"grad_norm": 0.9699122905731201,
"learning_rate": 3.2638185671897794e-05,
"loss": 0.0173,
"step": 2566
},
{
"epoch": 4.88,
"grad_norm": 0.7683483958244324,
"learning_rate": 3.2619784027028705e-05,
"loss": 0.0141,
"step": 2567
},
{
"epoch": 4.88,
"grad_norm": 0.6850505471229553,
"learning_rate": 3.2601381388850746e-05,
"loss": 0.0154,
"step": 2568
},
{
"epoch": 4.88,
"grad_norm": 0.800439178943634,
"learning_rate": 3.258297776434141e-05,
"loss": 0.0205,
"step": 2569
},
{
"epoch": 4.89,
"grad_norm": 0.9860864877700806,
"learning_rate": 3.2564573160478546e-05,
"loss": 0.027,
"step": 2570
},
{
"epoch": 4.89,
"grad_norm": 0.5462270975112915,
"learning_rate": 3.254616758424037e-05,
"loss": 0.0102,
"step": 2571
},
{
"epoch": 4.89,
"grad_norm": 0.6027054190635681,
"learning_rate": 3.252776104260552e-05,
"loss": 0.009,
"step": 2572
},
{
"epoch": 4.89,
"grad_norm": 0.7554721236228943,
"learning_rate": 3.250935354255293e-05,
"loss": 0.0256,
"step": 2573
},
{
"epoch": 4.89,
"grad_norm": 1.0127109289169312,
"learning_rate": 3.249094509106194e-05,
"loss": 0.0197,
"step": 2574
},
{
"epoch": 4.9,
"grad_norm": 0.6435019373893738,
"learning_rate": 3.2472535695112235e-05,
"loss": 0.0111,
"step": 2575
},
{
"epoch": 4.9,
"grad_norm": 0.35749948024749756,
"learning_rate": 3.245412536168387e-05,
"loss": 0.0094,
"step": 2576
},
{
"epoch": 4.9,
"grad_norm": 0.6160987019538879,
"learning_rate": 3.243571409775725e-05,
"loss": 0.0118,
"step": 2577
},
{
"epoch": 4.9,
"grad_norm": 0.8736352920532227,
"learning_rate": 3.2417301910313107e-05,
"loss": 0.0227,
"step": 2578
},
{
"epoch": 4.9,
"grad_norm": 0.5913633704185486,
"learning_rate": 3.2398888806332567e-05,
"loss": 0.0222,
"step": 2579
},
{
"epoch": 4.9,
"grad_norm": 0.8007988333702087,
"learning_rate": 3.238047479279708e-05,
"loss": 0.0213,
"step": 2580
},
{
"epoch": 4.91,
"grad_norm": 0.7501394152641296,
"learning_rate": 3.236205987668843e-05,
"loss": 0.0196,
"step": 2581
},
{
"epoch": 4.91,
"grad_norm": 1.001498818397522,
"learning_rate": 3.234364406498879e-05,
"loss": 0.0178,
"step": 2582
},
{
"epoch": 4.91,
"grad_norm": 0.7225024700164795,
"learning_rate": 3.232522736468062e-05,
"loss": 0.0348,
"step": 2583
},
{
"epoch": 4.91,
"grad_norm": 0.857322096824646,
"learning_rate": 3.230680978274671e-05,
"loss": 0.0194,
"step": 2584
},
{
"epoch": 4.91,
"grad_norm": 0.9504652619361877,
"learning_rate": 3.228839132617026e-05,
"loss": 0.0138,
"step": 2585
},
{
"epoch": 4.92,
"grad_norm": 0.8126028180122375,
"learning_rate": 3.226997200193472e-05,
"loss": 0.0173,
"step": 2586
},
{
"epoch": 4.92,
"grad_norm": 0.7010256052017212,
"learning_rate": 3.225155181702391e-05,
"loss": 0.0206,
"step": 2587
},
{
"epoch": 4.92,
"grad_norm": 0.7558382749557495,
"learning_rate": 3.223313077842198e-05,
"loss": 0.0233,
"step": 2588
},
{
"epoch": 4.92,
"grad_norm": 0.9425675272941589,
"learning_rate": 3.221470889311337e-05,
"loss": 0.0162,
"step": 2589
},
{
"epoch": 4.92,
"grad_norm": 0.7752507328987122,
"learning_rate": 3.2196286168082884e-05,
"loss": 0.0334,
"step": 2590
},
{
"epoch": 4.93,
"grad_norm": 0.8690214157104492,
"learning_rate": 3.217786261031561e-05,
"loss": 0.0262,
"step": 2591
},
{
"epoch": 4.93,
"grad_norm": 0.6650963425636292,
"learning_rate": 3.215943822679697e-05,
"loss": 0.0172,
"step": 2592
},
{
"epoch": 4.93,
"grad_norm": 1.2970396280288696,
"learning_rate": 3.2141013024512685e-05,
"loss": 0.0432,
"step": 2593
},
{
"epoch": 4.93,
"grad_norm": 0.48316967487335205,
"learning_rate": 3.2122587010448796e-05,
"loss": 0.0107,
"step": 2594
},
{
"epoch": 4.93,
"grad_norm": 1.2123706340789795,
"learning_rate": 3.210416019159167e-05,
"loss": 0.0227,
"step": 2595
},
{
"epoch": 4.94,
"grad_norm": 0.8546351194381714,
"learning_rate": 3.208573257492795e-05,
"loss": 0.0337,
"step": 2596
},
{
"epoch": 4.94,
"grad_norm": 0.5967420339584351,
"learning_rate": 3.2067304167444564e-05,
"loss": 0.0169,
"step": 2597
},
{
"epoch": 4.94,
"grad_norm": 0.7887455224990845,
"learning_rate": 3.204887497612881e-05,
"loss": 0.027,
"step": 2598
},
{
"epoch": 4.94,
"grad_norm": 0.7494004964828491,
"learning_rate": 3.203044500796822e-05,
"loss": 0.0174,
"step": 2599
},
{
"epoch": 4.94,
"grad_norm": 0.49930045008659363,
"learning_rate": 3.201201426995066e-05,
"loss": 0.0139,
"step": 2600
},
{
"epoch": 4.94,
"eval_blimp_filtered_avg": 0.7183582089552238,
"eval_blimp_filtered_std": 0.005007470606586838,
"step": 2600
},
{
"epoch": 4.94,
"eval_blimp_supplement_avg": 0.7758620689655172,
"eval_blimp_supplement_std": 0.01806836394050218,
"step": 2600
},
{
"epoch": 4.94,
"eval_vqa_filtered_avg": 0.38,
"eval_vqa_filtered_std": 0.04878317312145633,
"step": 2600
},
{
"epoch": 4.94,
"eval_winoground_filtered_avg": 0.44,
"eval_winoground_filtered_std": 0.04988876515698589,
"step": 2600
},
{
"epoch": 4.94,
"grad_norm": 1.3112573623657227,
"learning_rate": 3.199358276906423e-05,
"loss": 0.0299,
"step": 2601
},
{
"epoch": 4.95,
"grad_norm": 0.68585205078125,
"learning_rate": 3.19751505122974e-05,
"loss": 0.0134,
"step": 2602
},
{
"epoch": 4.95,
"grad_norm": 0.511154294013977,
"learning_rate": 3.195671750663886e-05,
"loss": 0.0098,
"step": 2603
},
{
"epoch": 4.95,
"grad_norm": 0.5511214733123779,
"learning_rate": 3.193828375907761e-05,
"loss": 0.0169,
"step": 2604
},
{
"epoch": 4.95,
"grad_norm": 0.6784529685974121,
"learning_rate": 3.1919849276602945e-05,
"loss": 0.015,
"step": 2605
},
{
"epoch": 4.95,
"grad_norm": 1.4393839836120605,
"learning_rate": 3.190141406620442e-05,
"loss": 0.027,
"step": 2606
},
{
"epoch": 4.96,
"grad_norm": 0.49157747626304626,
"learning_rate": 3.188297813487185e-05,
"loss": 0.0109,
"step": 2607
},
{
"epoch": 4.96,
"grad_norm": 1.4795408248901367,
"learning_rate": 3.186454148959537e-05,
"loss": 0.0788,
"step": 2608
},
{
"epoch": 4.96,
"grad_norm": 0.6279906034469604,
"learning_rate": 3.184610413736535e-05,
"loss": 0.0147,
"step": 2609
},
{
"epoch": 4.96,
"grad_norm": 0.5698646903038025,
"learning_rate": 3.182766608517242e-05,
"loss": 0.019,
"step": 2610
},
{
"epoch": 4.96,
"grad_norm": 0.7896731495857239,
"learning_rate": 3.180922734000752e-05,
"loss": 0.0189,
"step": 2611
},
{
"epoch": 4.97,
"grad_norm": 1.2281535863876343,
"learning_rate": 3.179078790886181e-05,
"loss": 0.0215,
"step": 2612
},
{
"epoch": 4.97,
"grad_norm": 0.8027342557907104,
"learning_rate": 3.177234779872673e-05,
"loss": 0.0289,
"step": 2613
},
{
"epoch": 4.97,
"grad_norm": 0.4821428954601288,
"learning_rate": 3.175390701659397e-05,
"loss": 0.0094,
"step": 2614
},
{
"epoch": 4.97,
"grad_norm": 0.39466989040374756,
"learning_rate": 3.1735465569455486e-05,
"loss": 0.0099,
"step": 2615
},
{
"epoch": 4.97,
"grad_norm": 1.1873451471328735,
"learning_rate": 3.171702346430345e-05,
"loss": 0.024,
"step": 2616
},
{
"epoch": 4.98,
"grad_norm": 0.7595629096031189,
"learning_rate": 3.1698580708130346e-05,
"loss": 0.0234,
"step": 2617
},
{
"epoch": 4.98,
"grad_norm": 0.5430089831352234,
"learning_rate": 3.168013730792885e-05,
"loss": 0.0093,
"step": 2618
},
{
"epoch": 4.98,
"grad_norm": 0.3912401497364044,
"learning_rate": 3.166169327069191e-05,
"loss": 0.0088,
"step": 2619
},
{
"epoch": 4.98,
"grad_norm": 0.7474127411842346,
"learning_rate": 3.164324860341269e-05,
"loss": 0.0171,
"step": 2620
},
{
"epoch": 4.98,
"grad_norm": 0.9248905777931213,
"learning_rate": 3.1624803313084645e-05,
"loss": 0.0311,
"step": 2621
},
{
"epoch": 4.98,
"grad_norm": 0.7016698718070984,
"learning_rate": 3.160635740670141e-05,
"loss": 0.0159,
"step": 2622
},
{
"epoch": 4.99,
"grad_norm": 1.0913496017456055,
"learning_rate": 3.158791089125687e-05,
"loss": 0.0501,
"step": 2623
},
{
"epoch": 4.99,
"grad_norm": 0.7720309495925903,
"learning_rate": 3.1569463773745165e-05,
"loss": 0.0151,
"step": 2624
},
{
"epoch": 4.99,
"grad_norm": 0.4265575408935547,
"learning_rate": 3.1551016061160624e-05,
"loss": 0.01,
"step": 2625
},
{
"epoch": 4.99,
"grad_norm": 0.7054266333580017,
"learning_rate": 3.1532567760497825e-05,
"loss": 0.0218,
"step": 2626
},
{
"epoch": 4.99,
"grad_norm": 0.8669946193695068,
"learning_rate": 3.151411887875159e-05,
"loss": 0.0202,
"step": 2627
},
{
"epoch": 5.0,
"grad_norm": 0.6798685789108276,
"learning_rate": 3.1495669422916916e-05,
"loss": 0.0113,
"step": 2628
},
{
"epoch": 5.0,
"grad_norm": 0.3804970681667328,
"learning_rate": 3.147721939998903e-05,
"loss": 0.0067,
"step": 2629
},
{
"epoch": 5.0,
"grad_norm": 0.554598867893219,
"learning_rate": 3.145876881696341e-05,
"loss": 0.0192,
"step": 2630
},
{
"epoch": 5.0,
"grad_norm": 0.6888824105262756,
"learning_rate": 3.144031768083571e-05,
"loss": 0.0067,
"step": 2631
},
{
"epoch": 5.0,
"grad_norm": 0.5580077171325684,
"learning_rate": 3.142186599860179e-05,
"loss": 0.0304,
"step": 2632
},
{
"epoch": 5.01,
"grad_norm": 1.0857365131378174,
"learning_rate": 3.140341377725774e-05,
"loss": 0.0172,
"step": 2633
},
{
"epoch": 5.01,
"grad_norm": 0.5568354725837708,
"learning_rate": 3.1384961023799856e-05,
"loss": 0.0216,
"step": 2634
},
{
"epoch": 5.01,
"grad_norm": 0.6704714298248291,
"learning_rate": 3.1366507745224606e-05,
"loss": 0.0073,
"step": 2635
},
{
"epoch": 5.01,
"grad_norm": 0.6381349563598633,
"learning_rate": 3.134805394852869e-05,
"loss": 0.0263,
"step": 2636
},
{
"epoch": 5.01,
"grad_norm": 0.4878329336643219,
"learning_rate": 3.132959964070897e-05,
"loss": 0.0142,
"step": 2637
},
{
"epoch": 5.02,
"grad_norm": 0.5874066948890686,
"learning_rate": 3.1311144828762554e-05,
"loss": 0.0215,
"step": 2638
},
{
"epoch": 5.02,
"grad_norm": 0.7634973526000977,
"learning_rate": 3.129268951968667e-05,
"loss": 0.0135,
"step": 2639
},
{
"epoch": 5.02,
"grad_norm": 0.7094599604606628,
"learning_rate": 3.127423372047881e-05,
"loss": 0.0209,
"step": 2640
},
{
"epoch": 5.02,
"grad_norm": 0.46887609362602234,
"learning_rate": 3.125577743813659e-05,
"loss": 0.0058,
"step": 2641
},
{
"epoch": 5.02,
"grad_norm": 0.9768758416175842,
"learning_rate": 3.1237320679657856e-05,
"loss": 0.0379,
"step": 2642
},
{
"epoch": 5.02,
"grad_norm": 0.8575007915496826,
"learning_rate": 3.121886345204058e-05,
"loss": 0.0094,
"step": 2643
},
{
"epoch": 5.03,
"grad_norm": 0.7477152347564697,
"learning_rate": 3.1200405762282985e-05,
"loss": 0.0321,
"step": 2644
},
{
"epoch": 5.03,
"grad_norm": 0.5118290781974792,
"learning_rate": 3.11819476173834e-05,
"loss": 0.0223,
"step": 2645
},
{
"epoch": 5.03,
"grad_norm": 0.39241889119148254,
"learning_rate": 3.116348902434037e-05,
"loss": 0.008,
"step": 2646
},
{
"epoch": 5.03,
"grad_norm": 0.704013466835022,
"learning_rate": 3.114502999015259e-05,
"loss": 0.0092,
"step": 2647
},
{
"epoch": 5.03,
"grad_norm": 0.22399628162384033,
"learning_rate": 3.112657052181892e-05,
"loss": 0.0033,
"step": 2648
},
{
"epoch": 5.04,
"grad_norm": 0.8250674605369568,
"learning_rate": 3.1108110626338406e-05,
"loss": 0.015,
"step": 2649
},
{
"epoch": 5.04,
"grad_norm": 0.43121230602264404,
"learning_rate": 3.108965031071025e-05,
"loss": 0.0115,
"step": 2650
},
{
"epoch": 5.04,
"grad_norm": 0.34507110714912415,
"learning_rate": 3.1071189581933785e-05,
"loss": 0.0084,
"step": 2651
},
{
"epoch": 5.04,
"grad_norm": 1.0469059944152832,
"learning_rate": 3.1052728447008524e-05,
"loss": 0.0612,
"step": 2652
},
{
"epoch": 5.04,
"grad_norm": 0.4445590674877167,
"learning_rate": 3.1034266912934146e-05,
"loss": 0.0056,
"step": 2653
},
{
"epoch": 5.05,
"grad_norm": 0.6758363246917725,
"learning_rate": 3.101580498671046e-05,
"loss": 0.0348,
"step": 2654
},
{
"epoch": 5.05,
"grad_norm": 0.6954324841499329,
"learning_rate": 3.099734267533743e-05,
"loss": 0.0251,
"step": 2655
},
{
"epoch": 5.05,
"grad_norm": 0.4815708100795746,
"learning_rate": 3.0978879985815154e-05,
"loss": 0.014,
"step": 2656
},
{
"epoch": 5.05,
"grad_norm": 0.5608925819396973,
"learning_rate": 3.096041692514391e-05,
"loss": 0.0093,
"step": 2657
},
{
"epoch": 5.05,
"grad_norm": 0.6329244375228882,
"learning_rate": 3.0941953500324074e-05,
"loss": 0.0166,
"step": 2658
},
{
"epoch": 5.06,
"grad_norm": 0.3578009605407715,
"learning_rate": 3.0923489718356194e-05,
"loss": 0.008,
"step": 2659
},
{
"epoch": 5.06,
"grad_norm": 0.7515147924423218,
"learning_rate": 3.090502558624092e-05,
"loss": 0.02,
"step": 2660
},
{
"epoch": 5.06,
"grad_norm": 0.47370824217796326,
"learning_rate": 3.0886561110979067e-05,
"loss": 0.0083,
"step": 2661
},
{
"epoch": 5.06,
"grad_norm": 0.5509132146835327,
"learning_rate": 3.0868096299571544e-05,
"loss": 0.009,
"step": 2662
},
{
"epoch": 5.06,
"grad_norm": 0.7511709332466125,
"learning_rate": 3.084963115901943e-05,
"loss": 0.0113,
"step": 2663
},
{
"epoch": 5.06,
"grad_norm": 0.4746300280094147,
"learning_rate": 3.083116569632393e-05,
"loss": 0.0097,
"step": 2664
},
{
"epoch": 5.07,
"grad_norm": 0.30844685435295105,
"learning_rate": 3.0812699918486297e-05,
"loss": 0.0087,
"step": 2665
},
{
"epoch": 5.07,
"grad_norm": 0.44849836826324463,
"learning_rate": 3.079423383250798e-05,
"loss": 0.0149,
"step": 2666
},
{
"epoch": 5.07,
"grad_norm": 0.7418750524520874,
"learning_rate": 3.077576744539054e-05,
"loss": 0.0273,
"step": 2667
},
{
"epoch": 5.07,
"grad_norm": 0.6695336103439331,
"learning_rate": 3.0757300764135606e-05,
"loss": 0.0162,
"step": 2668
},
{
"epoch": 5.07,
"grad_norm": 0.49009013175964355,
"learning_rate": 3.073883379574496e-05,
"loss": 0.0061,
"step": 2669
},
{
"epoch": 5.08,
"grad_norm": 0.5873833894729614,
"learning_rate": 3.072036654722047e-05,
"loss": 0.0193,
"step": 2670
},
{
"epoch": 5.08,
"grad_norm": 0.808398962020874,
"learning_rate": 3.070189902556413e-05,
"loss": 0.0097,
"step": 2671
},
{
"epoch": 5.08,
"grad_norm": 0.9591725468635559,
"learning_rate": 3.0683431237778e-05,
"loss": 0.0104,
"step": 2672
},
{
"epoch": 5.08,
"grad_norm": 0.48921096324920654,
"learning_rate": 3.0664963190864296e-05,
"loss": 0.0117,
"step": 2673
},
{
"epoch": 5.08,
"grad_norm": 0.6877846121788025,
"learning_rate": 3.0646494891825285e-05,
"loss": 0.0257,
"step": 2674
},
{
"epoch": 5.09,
"grad_norm": 0.4908289313316345,
"learning_rate": 3.0628026347663355e-05,
"loss": 0.0091,
"step": 2675
},
{
"epoch": 5.09,
"grad_norm": 0.6457719206809998,
"learning_rate": 3.060955756538099e-05,
"loss": 0.0196,
"step": 2676
},
{
"epoch": 5.09,
"grad_norm": 0.6335395574569702,
"learning_rate": 3.059108855198073e-05,
"loss": 0.0071,
"step": 2677
},
{
"epoch": 5.09,
"grad_norm": 0.8716082572937012,
"learning_rate": 3.057261931446524e-05,
"loss": 0.0154,
"step": 2678
},
{
"epoch": 5.09,
"grad_norm": 0.6026990413665771,
"learning_rate": 3.055414985983726e-05,
"loss": 0.0071,
"step": 2679
},
{
"epoch": 5.1,
"grad_norm": 0.5927395224571228,
"learning_rate": 3.0535680195099604e-05,
"loss": 0.0344,
"step": 2680
},
{
"epoch": 5.1,
"grad_norm": 0.8427085280418396,
"learning_rate": 3.0517210327255184e-05,
"loss": 0.0116,
"step": 2681
},
{
"epoch": 5.1,
"grad_norm": 1.600317120552063,
"learning_rate": 3.0498740263306954e-05,
"loss": 0.0181,
"step": 2682
},
{
"epoch": 5.1,
"grad_norm": 0.38328924775123596,
"learning_rate": 3.0480270010257974e-05,
"loss": 0.0239,
"step": 2683
},
{
"epoch": 5.1,
"grad_norm": 0.4525313377380371,
"learning_rate": 3.0461799575111364e-05,
"loss": 0.0099,
"step": 2684
},
{
"epoch": 5.1,
"grad_norm": 0.6008482575416565,
"learning_rate": 3.044332896487031e-05,
"loss": 0.0067,
"step": 2685
},
{
"epoch": 5.11,
"grad_norm": 0.1723678559064865,
"learning_rate": 3.042485818653808e-05,
"loss": 0.0022,
"step": 2686
},
{
"epoch": 5.11,
"grad_norm": 0.286811888217926,
"learning_rate": 3.0406387247117988e-05,
"loss": 0.0035,
"step": 2687
},
{
"epoch": 5.11,
"grad_norm": 0.5824350714683533,
"learning_rate": 3.038791615361341e-05,
"loss": 0.009,
"step": 2688
},
{
"epoch": 5.11,
"grad_norm": 0.31421786546707153,
"learning_rate": 3.03694449130278e-05,
"loss": 0.0089,
"step": 2689
},
{
"epoch": 5.11,
"grad_norm": 0.8175370097160339,
"learning_rate": 3.0350973532364653e-05,
"loss": 0.0188,
"step": 2690
},
{
"epoch": 5.12,
"grad_norm": 0.590660810470581,
"learning_rate": 3.0332502018627505e-05,
"loss": 0.008,
"step": 2691
},
{
"epoch": 5.12,
"grad_norm": 0.527713418006897,
"learning_rate": 3.031403037881997e-05,
"loss": 0.0034,
"step": 2692
},
{
"epoch": 5.12,
"grad_norm": 0.7556295990943909,
"learning_rate": 3.029555861994568e-05,
"loss": 0.035,
"step": 2693
},
{
"epoch": 5.12,
"grad_norm": 0.38739511370658875,
"learning_rate": 3.0277086749008345e-05,
"loss": 0.0026,
"step": 2694
},
{
"epoch": 5.12,
"grad_norm": 0.9117938280105591,
"learning_rate": 3.0258614773011684e-05,
"loss": 0.0161,
"step": 2695
},
{
"epoch": 5.13,
"grad_norm": 0.531975507736206,
"learning_rate": 3.0240142698959483e-05,
"loss": 0.0045,
"step": 2696
},
{
"epoch": 5.13,
"grad_norm": 0.3335452377796173,
"learning_rate": 3.022167053385555e-05,
"loss": 0.0037,
"step": 2697
},
{
"epoch": 5.13,
"grad_norm": 0.8446972966194153,
"learning_rate": 3.020319828470373e-05,
"loss": 0.007,
"step": 2698
},
{
"epoch": 5.13,
"grad_norm": 0.6875817179679871,
"learning_rate": 3.01847259585079e-05,
"loss": 0.0155,
"step": 2699
},
{
"epoch": 5.13,
"grad_norm": 0.6529269218444824,
"learning_rate": 3.016625356227198e-05,
"loss": 0.0137,
"step": 2700
},
{
"epoch": 5.13,
"eval_blimp_filtered_avg": 0.716865671641791,
"eval_blimp_filtered_std": 0.005035522125395878,
"step": 2700
},
{
"epoch": 5.13,
"eval_blimp_supplement_avg": 0.771551724137931,
"eval_blimp_supplement_std": 0.018016315649767755,
"step": 2700
},
{
"epoch": 5.13,
"eval_vqa_filtered_avg": 0.35,
"eval_vqa_filtered_std": 0.047937248544110196,
"step": 2700
},
{
"epoch": 5.13,
"eval_winoground_filtered_avg": 0.5,
"eval_winoground_filtered_std": 0.050251890762960605,
"step": 2700
},
{
"epoch": 5.13,
"grad_norm": 0.47955256700515747,
"learning_rate": 3.0147781102999885e-05,
"loss": 0.0045,
"step": 2701
},
{
"epoch": 5.14,
"grad_norm": 0.43090352416038513,
"learning_rate": 3.012930858769559e-05,
"loss": 0.0104,
"step": 2702
},
{
"epoch": 5.14,
"grad_norm": 0.600084662437439,
"learning_rate": 3.0110836023363068e-05,
"loss": 0.0161,
"step": 2703
},
{
"epoch": 5.14,
"grad_norm": 0.24938753247261047,
"learning_rate": 3.0092363417006313e-05,
"loss": 0.0068,
"step": 2704
},
{
"epoch": 5.14,
"grad_norm": 0.8033124208450317,
"learning_rate": 3.0073890775629343e-05,
"loss": 0.0068,
"step": 2705
},
{
"epoch": 5.14,
"grad_norm": 0.7103681564331055,
"learning_rate": 3.0055418106236193e-05,
"loss": 0.0134,
"step": 2706
},
{
"epoch": 5.15,
"grad_norm": 0.3161657154560089,
"learning_rate": 3.0036945415830886e-05,
"loss": 0.0028,
"step": 2707
},
{
"epoch": 5.15,
"grad_norm": 1.1548048257827759,
"learning_rate": 3.0018472711417467e-05,
"loss": 0.015,
"step": 2708
},
{
"epoch": 5.15,
"grad_norm": 0.9178134799003601,
"learning_rate": 3e-05,
"loss": 0.0285,
"step": 2709
},
{
"epoch": 5.15,
"grad_norm": 0.9164400696754456,
"learning_rate": 2.998152728858253e-05,
"loss": 0.0231,
"step": 2710
},
{
"epoch": 5.15,
"grad_norm": 0.5622690916061401,
"learning_rate": 2.9963054584169125e-05,
"loss": 0.0052,
"step": 2711
},
{
"epoch": 5.16,
"grad_norm": 0.42683538794517517,
"learning_rate": 2.9944581893763815e-05,
"loss": 0.0053,
"step": 2712
},
{
"epoch": 5.16,
"grad_norm": 0.5729798078536987,
"learning_rate": 2.992610922437066e-05,
"loss": 0.012,
"step": 2713
},
{
"epoch": 5.16,
"grad_norm": 0.6264289021492004,
"learning_rate": 2.990763658299369e-05,
"loss": 0.0099,
"step": 2714
},
{
"epoch": 5.16,
"grad_norm": 0.2999776303768158,
"learning_rate": 2.9889163976636937e-05,
"loss": 0.0072,
"step": 2715
},
{
"epoch": 5.16,
"grad_norm": 0.49189528822898865,
"learning_rate": 2.9870691412304407e-05,
"loss": 0.0071,
"step": 2716
},
{
"epoch": 5.17,
"grad_norm": 0.7994298338890076,
"learning_rate": 2.985221889700012e-05,
"loss": 0.0101,
"step": 2717
},
{
"epoch": 5.17,
"grad_norm": 0.9241929650306702,
"learning_rate": 2.9833746437728027e-05,
"loss": 0.0221,
"step": 2718
},
{
"epoch": 5.17,
"grad_norm": 0.5734905004501343,
"learning_rate": 2.9815274041492096e-05,
"loss": 0.0077,
"step": 2719
},
{
"epoch": 5.17,
"grad_norm": 0.5160849690437317,
"learning_rate": 2.9796801715296276e-05,
"loss": 0.0237,
"step": 2720
},
{
"epoch": 5.17,
"grad_norm": 0.47447580099105835,
"learning_rate": 2.9778329466144455e-05,
"loss": 0.0106,
"step": 2721
},
{
"epoch": 5.17,
"grad_norm": 0.3402750492095947,
"learning_rate": 2.975985730104053e-05,
"loss": 0.0038,
"step": 2722
},
{
"epoch": 5.18,
"grad_norm": 0.29368576407432556,
"learning_rate": 2.974138522698832e-05,
"loss": 0.0026,
"step": 2723
},
{
"epoch": 5.18,
"grad_norm": 0.40511587262153625,
"learning_rate": 2.9722913250991656e-05,
"loss": 0.0062,
"step": 2724
},
{
"epoch": 5.18,
"grad_norm": 0.6562106013298035,
"learning_rate": 2.9704441380054327e-05,
"loss": 0.0079,
"step": 2725
},
{
"epoch": 5.18,
"grad_norm": 0.5135022401809692,
"learning_rate": 2.9685969621180037e-05,
"loss": 0.0242,
"step": 2726
},
{
"epoch": 5.18,
"grad_norm": 0.9988023042678833,
"learning_rate": 2.96674979813725e-05,
"loss": 0.0151,
"step": 2727
},
{
"epoch": 5.19,
"grad_norm": 1.0609445571899414,
"learning_rate": 2.964902646763535e-05,
"loss": 0.0159,
"step": 2728
},
{
"epoch": 5.19,
"grad_norm": 0.8205469250679016,
"learning_rate": 2.9630555086972202e-05,
"loss": 0.0177,
"step": 2729
},
{
"epoch": 5.19,
"grad_norm": 0.5736174583435059,
"learning_rate": 2.9612083846386587e-05,
"loss": 0.0094,
"step": 2730
},
{
"epoch": 5.19,
"grad_norm": 0.30881407856941223,
"learning_rate": 2.9593612752882027e-05,
"loss": 0.005,
"step": 2731
},
{
"epoch": 5.19,
"grad_norm": 0.19705606997013092,
"learning_rate": 2.9575141813461928e-05,
"loss": 0.0037,
"step": 2732
},
{
"epoch": 5.2,
"grad_norm": 0.6435357332229614,
"learning_rate": 2.955667103512969e-05,
"loss": 0.0153,
"step": 2733
},
{
"epoch": 5.2,
"grad_norm": 0.6848272085189819,
"learning_rate": 2.953820042488865e-05,
"loss": 0.0092,
"step": 2734
},
{
"epoch": 5.2,
"grad_norm": 0.6380735039710999,
"learning_rate": 2.9519729989742035e-05,
"loss": 0.0167,
"step": 2735
},
{
"epoch": 5.2,
"grad_norm": 0.29926490783691406,
"learning_rate": 2.950125973669305e-05,
"loss": 0.0066,
"step": 2736
},
{
"epoch": 5.2,
"grad_norm": 0.6781852841377258,
"learning_rate": 2.9482789672744824e-05,
"loss": 0.0102,
"step": 2737
},
{
"epoch": 5.21,
"grad_norm": 0.5853827595710754,
"learning_rate": 2.9464319804900394e-05,
"loss": 0.0071,
"step": 2738
},
{
"epoch": 5.21,
"grad_norm": 0.5854670405387878,
"learning_rate": 2.9445850140162735e-05,
"loss": 0.0076,
"step": 2739
},
{
"epoch": 5.21,
"grad_norm": 0.3603760302066803,
"learning_rate": 2.9427380685534766e-05,
"loss": 0.0044,
"step": 2740
},
{
"epoch": 5.21,
"grad_norm": 0.593079686164856,
"learning_rate": 2.9408911448019276e-05,
"loss": 0.014,
"step": 2741
},
{
"epoch": 5.21,
"grad_norm": 0.6322662234306335,
"learning_rate": 2.9390442434619016e-05,
"loss": 0.011,
"step": 2742
},
{
"epoch": 5.21,
"grad_norm": 0.545086681842804,
"learning_rate": 2.9371973652336653e-05,
"loss": 0.0144,
"step": 2743
},
{
"epoch": 5.22,
"grad_norm": 0.866096556186676,
"learning_rate": 2.935350510817472e-05,
"loss": 0.021,
"step": 2744
},
{
"epoch": 5.22,
"grad_norm": 0.33861085772514343,
"learning_rate": 2.9335036809135705e-05,
"loss": 0.013,
"step": 2745
},
{
"epoch": 5.22,
"grad_norm": 0.4894312918186188,
"learning_rate": 2.931656876222201e-05,
"loss": 0.0101,
"step": 2746
},
{
"epoch": 5.22,
"grad_norm": 0.890651285648346,
"learning_rate": 2.929810097443588e-05,
"loss": 0.0223,
"step": 2747
},
{
"epoch": 5.22,
"grad_norm": 0.5600990653038025,
"learning_rate": 2.927963345277954e-05,
"loss": 0.0121,
"step": 2748
},
{
"epoch": 5.23,
"grad_norm": 0.6349409222602844,
"learning_rate": 2.9261166204255047e-05,
"loss": 0.0095,
"step": 2749
},
{
"epoch": 5.23,
"grad_norm": 0.4856875538825989,
"learning_rate": 2.92426992358644e-05,
"loss": 0.0076,
"step": 2750
},
{
"epoch": 5.23,
"grad_norm": 1.047897219657898,
"learning_rate": 2.9224232554609467e-05,
"loss": 0.0207,
"step": 2751
},
{
"epoch": 5.23,
"grad_norm": 0.5296978950500488,
"learning_rate": 2.920576616749202e-05,
"loss": 0.0077,
"step": 2752
},
{
"epoch": 5.23,
"grad_norm": 0.4981590807437897,
"learning_rate": 2.918730008151371e-05,
"loss": 0.015,
"step": 2753
},
{
"epoch": 5.24,
"grad_norm": 0.626278281211853,
"learning_rate": 2.916883430367609e-05,
"loss": 0.0054,
"step": 2754
},
{
"epoch": 5.24,
"grad_norm": 0.3714407682418823,
"learning_rate": 2.9150368840980565e-05,
"loss": 0.0046,
"step": 2755
},
{
"epoch": 5.24,
"grad_norm": 0.3161694407463074,
"learning_rate": 2.9131903700428454e-05,
"loss": 0.0052,
"step": 2756
},
{
"epoch": 5.24,
"grad_norm": 0.613519549369812,
"learning_rate": 2.9113438889020952e-05,
"loss": 0.0069,
"step": 2757
},
{
"epoch": 5.24,
"grad_norm": 0.24029579758644104,
"learning_rate": 2.909497441375909e-05,
"loss": 0.0038,
"step": 2758
},
{
"epoch": 5.25,
"grad_norm": 0.4901004433631897,
"learning_rate": 2.9076510281643814e-05,
"loss": 0.0104,
"step": 2759
},
{
"epoch": 5.25,
"grad_norm": 0.6988447308540344,
"learning_rate": 2.9058046499675928e-05,
"loss": 0.0082,
"step": 2760
},
{
"epoch": 5.25,
"grad_norm": 0.4997389018535614,
"learning_rate": 2.9039583074856095e-05,
"loss": 0.0152,
"step": 2761
},
{
"epoch": 5.25,
"grad_norm": 0.3509935140609741,
"learning_rate": 2.902112001418484e-05,
"loss": 0.0071,
"step": 2762
},
{
"epoch": 5.25,
"grad_norm": 0.9302648305892944,
"learning_rate": 2.900265732466258e-05,
"loss": 0.0075,
"step": 2763
},
{
"epoch": 5.25,
"grad_norm": 0.5164839029312134,
"learning_rate": 2.8984195013289543e-05,
"loss": 0.0209,
"step": 2764
},
{
"epoch": 5.26,
"grad_norm": 0.429945170879364,
"learning_rate": 2.896573308706585e-05,
"loss": 0.008,
"step": 2765
},
{
"epoch": 5.26,
"grad_norm": 0.48670411109924316,
"learning_rate": 2.8947271552991477e-05,
"loss": 0.0141,
"step": 2766
},
{
"epoch": 5.26,
"grad_norm": 0.9812799692153931,
"learning_rate": 2.892881041806622e-05,
"loss": 0.0247,
"step": 2767
},
{
"epoch": 5.26,
"grad_norm": 0.7613944411277771,
"learning_rate": 2.8910349689289748e-05,
"loss": 0.0175,
"step": 2768
},
{
"epoch": 5.26,
"grad_norm": 0.3294127285480499,
"learning_rate": 2.8891889373661592e-05,
"loss": 0.0095,
"step": 2769
},
{
"epoch": 5.27,
"grad_norm": 0.8999878764152527,
"learning_rate": 2.8873429478181076e-05,
"loss": 0.0106,
"step": 2770
},
{
"epoch": 5.27,
"grad_norm": 0.3654761016368866,
"learning_rate": 2.8854970009847424e-05,
"loss": 0.0066,
"step": 2771
},
{
"epoch": 5.27,
"grad_norm": 0.6327349543571472,
"learning_rate": 2.883651097565964e-05,
"loss": 0.0066,
"step": 2772
},
{
"epoch": 5.27,
"grad_norm": 0.5260616540908813,
"learning_rate": 2.881805238261661e-05,
"loss": 0.0121,
"step": 2773
},
{
"epoch": 5.27,
"grad_norm": 0.701049268245697,
"learning_rate": 2.8799594237717027e-05,
"loss": 0.0192,
"step": 2774
},
{
"epoch": 5.28,
"grad_norm": 0.3985120952129364,
"learning_rate": 2.8781136547959423e-05,
"loss": 0.0077,
"step": 2775
},
{
"epoch": 5.28,
"grad_norm": 0.7325926423072815,
"learning_rate": 2.8762679320342156e-05,
"loss": 0.0078,
"step": 2776
},
{
"epoch": 5.28,
"grad_norm": 0.08369597047567368,
"learning_rate": 2.8744222561863417e-05,
"loss": 0.0012,
"step": 2777
},
{
"epoch": 5.28,
"grad_norm": 0.6579816937446594,
"learning_rate": 2.8725766279521197e-05,
"loss": 0.0035,
"step": 2778
},
{
"epoch": 5.28,
"grad_norm": 0.6968809962272644,
"learning_rate": 2.8707310480313325e-05,
"loss": 0.0099,
"step": 2779
},
{
"epoch": 5.29,
"grad_norm": 0.8602203130722046,
"learning_rate": 2.868885517123746e-05,
"loss": 0.0201,
"step": 2780
},
{
"epoch": 5.29,
"grad_norm": 0.5955621004104614,
"learning_rate": 2.8670400359291026e-05,
"loss": 0.0136,
"step": 2781
},
{
"epoch": 5.29,
"grad_norm": 0.4376932382583618,
"learning_rate": 2.8651946051471317e-05,
"loss": 0.0049,
"step": 2782
},
{
"epoch": 5.29,
"grad_norm": 0.5243163108825684,
"learning_rate": 2.86334922547754e-05,
"loss": 0.0085,
"step": 2783
},
{
"epoch": 5.29,
"grad_norm": 0.7298378348350525,
"learning_rate": 2.861503897620015e-05,
"loss": 0.0099,
"step": 2784
},
{
"epoch": 5.29,
"grad_norm": 0.24139423668384552,
"learning_rate": 2.8596586222742253e-05,
"loss": 0.0028,
"step": 2785
},
{
"epoch": 5.3,
"grad_norm": 0.7071788311004639,
"learning_rate": 2.8578134001398217e-05,
"loss": 0.0173,
"step": 2786
},
{
"epoch": 5.3,
"grad_norm": 0.3115825951099396,
"learning_rate": 2.85596823191643e-05,
"loss": 0.0052,
"step": 2787
},
{
"epoch": 5.3,
"grad_norm": 0.3365011215209961,
"learning_rate": 2.8541231183036585e-05,
"loss": 0.0049,
"step": 2788
},
{
"epoch": 5.3,
"grad_norm": 0.2221500277519226,
"learning_rate": 2.8522780600010973e-05,
"loss": 0.0028,
"step": 2789
},
{
"epoch": 5.3,
"grad_norm": 0.7803195118904114,
"learning_rate": 2.8504330577083092e-05,
"loss": 0.0099,
"step": 2790
},
{
"epoch": 5.31,
"grad_norm": 0.385833740234375,
"learning_rate": 2.848588112124841e-05,
"loss": 0.0054,
"step": 2791
},
{
"epoch": 5.31,
"grad_norm": 0.7232902646064758,
"learning_rate": 2.8467432239502173e-05,
"loss": 0.0177,
"step": 2792
},
{
"epoch": 5.31,
"grad_norm": 0.292402058839798,
"learning_rate": 2.844898393883938e-05,
"loss": 0.0047,
"step": 2793
},
{
"epoch": 5.31,
"grad_norm": 0.4307998716831207,
"learning_rate": 2.843053622625484e-05,
"loss": 0.0031,
"step": 2794
},
{
"epoch": 5.31,
"grad_norm": 0.6005372405052185,
"learning_rate": 2.841208910874313e-05,
"loss": 0.005,
"step": 2795
},
{
"epoch": 5.32,
"grad_norm": 0.6539878249168396,
"learning_rate": 2.8393642593298594e-05,
"loss": 0.0125,
"step": 2796
},
{
"epoch": 5.32,
"grad_norm": 0.48036593198776245,
"learning_rate": 2.8375196686915356e-05,
"loss": 0.0076,
"step": 2797
},
{
"epoch": 5.32,
"grad_norm": 0.3415619730949402,
"learning_rate": 2.8356751396587306e-05,
"loss": 0.0072,
"step": 2798
},
{
"epoch": 5.32,
"grad_norm": 0.4089036285877228,
"learning_rate": 2.8338306729308093e-05,
"loss": 0.005,
"step": 2799
},
{
"epoch": 5.32,
"grad_norm": 1.1154100894927979,
"learning_rate": 2.831986269207116e-05,
"loss": 0.01,
"step": 2800
},
{
"epoch": 5.32,
"eval_blimp_filtered_avg": 0.7192537313432836,
"eval_blimp_filtered_std": 0.0049934758779632766,
"step": 2800
},
{
"epoch": 5.32,
"eval_blimp_supplement_avg": 0.7866379310344828,
"eval_blimp_supplement_std": 0.017601141022014412,
"step": 2800
},
{
"epoch": 5.32,
"eval_vqa_filtered_avg": 0.33,
"eval_vqa_filtered_std": 0.047258156262526045,
"step": 2800
},
{
"epoch": 5.32,
"eval_winoground_filtered_avg": 0.5,
"eval_winoground_filtered_std": 0.050251890762960605,
"step": 2800
},
{
"epoch": 5.33,
"grad_norm": 0.7244845628738403,
"learning_rate": 2.830141929186966e-05,
"loss": 0.0148,
"step": 2801
},
{
"epoch": 5.33,
"grad_norm": 0.32194265723228455,
"learning_rate": 2.8282976535696546e-05,
"loss": 0.0091,
"step": 2802
},
{
"epoch": 5.33,
"grad_norm": 0.33089032769203186,
"learning_rate": 2.826453443054453e-05,
"loss": 0.003,
"step": 2803
},
{
"epoch": 5.33,
"grad_norm": 0.629823625087738,
"learning_rate": 2.8246092983406033e-05,
"loss": 0.0135,
"step": 2804
},
{
"epoch": 5.33,
"grad_norm": 0.5694602727890015,
"learning_rate": 2.8227652201273273e-05,
"loss": 0.0171,
"step": 2805
},
{
"epoch": 5.33,
"grad_norm": 0.3999047577381134,
"learning_rate": 2.8209212091138194e-05,
"loss": 0.0081,
"step": 2806
},
{
"epoch": 5.34,
"grad_norm": 0.6337502002716064,
"learning_rate": 2.8190772659992485e-05,
"loss": 0.0082,
"step": 2807
},
{
"epoch": 5.34,
"grad_norm": 0.6974929571151733,
"learning_rate": 2.817233391482758e-05,
"loss": 0.0124,
"step": 2808
},
{
"epoch": 5.34,
"grad_norm": 0.3280028998851776,
"learning_rate": 2.8153895862634665e-05,
"loss": 0.0068,
"step": 2809
},
{
"epoch": 5.34,
"grad_norm": 0.40549829602241516,
"learning_rate": 2.8135458510404636e-05,
"loss": 0.0094,
"step": 2810
},
{
"epoch": 5.34,
"grad_norm": 0.43101966381073,
"learning_rate": 2.811702186512815e-05,
"loss": 0.0095,
"step": 2811
},
{
"epoch": 5.35,
"grad_norm": 0.6215086579322815,
"learning_rate": 2.8098585933795597e-05,
"loss": 0.0197,
"step": 2812
},
{
"epoch": 5.35,
"grad_norm": 0.6214727759361267,
"learning_rate": 2.808015072339706e-05,
"loss": 0.0097,
"step": 2813
},
{
"epoch": 5.35,
"grad_norm": 0.5385430455207825,
"learning_rate": 2.8061716240922387e-05,
"loss": 0.0118,
"step": 2814
},
{
"epoch": 5.35,
"grad_norm": 0.6240503191947937,
"learning_rate": 2.8043282493361154e-05,
"loss": 0.0203,
"step": 2815
},
{
"epoch": 5.35,
"grad_norm": 0.35953447222709656,
"learning_rate": 2.8024849487702605e-05,
"loss": 0.0036,
"step": 2816
},
{
"epoch": 5.36,
"grad_norm": 0.5189916491508484,
"learning_rate": 2.800641723093577e-05,
"loss": 0.0065,
"step": 2817
},
{
"epoch": 5.36,
"grad_norm": 0.7398040890693665,
"learning_rate": 2.798798573004935e-05,
"loss": 0.0094,
"step": 2818
},
{
"epoch": 5.36,
"grad_norm": 0.4709341526031494,
"learning_rate": 2.796955499203178e-05,
"loss": 0.007,
"step": 2819
},
{
"epoch": 5.36,
"grad_norm": 0.5252180695533752,
"learning_rate": 2.7951125023871193e-05,
"loss": 0.0082,
"step": 2820
},
{
"epoch": 5.36,
"grad_norm": 1.099320650100708,
"learning_rate": 2.793269583255544e-05,
"loss": 0.0078,
"step": 2821
},
{
"epoch": 5.37,
"grad_norm": 0.3590094745159149,
"learning_rate": 2.7914267425072064e-05,
"loss": 0.0065,
"step": 2822
},
{
"epoch": 5.37,
"grad_norm": 0.3630542457103729,
"learning_rate": 2.7895839808408345e-05,
"loss": 0.0041,
"step": 2823
},
{
"epoch": 5.37,
"grad_norm": 0.6341803073883057,
"learning_rate": 2.7877412989551206e-05,
"loss": 0.0083,
"step": 2824
},
{
"epoch": 5.37,
"grad_norm": 0.27283141016960144,
"learning_rate": 2.785898697548732e-05,
"loss": 0.0046,
"step": 2825
},
{
"epoch": 5.37,
"grad_norm": 0.6050454378128052,
"learning_rate": 2.7840561773203044e-05,
"loss": 0.0113,
"step": 2826
},
{
"epoch": 5.37,
"grad_norm": 0.6951651573181152,
"learning_rate": 2.7822137389684396e-05,
"loss": 0.0086,
"step": 2827
},
{
"epoch": 5.38,
"grad_norm": 0.6231083273887634,
"learning_rate": 2.7803713831917117e-05,
"loss": 0.0053,
"step": 2828
},
{
"epoch": 5.38,
"grad_norm": 0.3886309564113617,
"learning_rate": 2.7785291106886632e-05,
"loss": 0.0052,
"step": 2829
},
{
"epoch": 5.38,
"grad_norm": 0.42390626668930054,
"learning_rate": 2.7766869221578027e-05,
"loss": 0.0093,
"step": 2830
},
{
"epoch": 5.38,
"grad_norm": 0.6966082453727722,
"learning_rate": 2.7748448182976086e-05,
"loss": 0.0102,
"step": 2831
},
{
"epoch": 5.38,
"grad_norm": 0.653001606464386,
"learning_rate": 2.773002799806529e-05,
"loss": 0.0133,
"step": 2832
},
{
"epoch": 5.39,
"grad_norm": 0.3783490061759949,
"learning_rate": 2.771160867382975e-05,
"loss": 0.0136,
"step": 2833
},
{
"epoch": 5.39,
"grad_norm": 0.4980427920818329,
"learning_rate": 2.7693190217253286e-05,
"loss": 0.0065,
"step": 2834
},
{
"epoch": 5.39,
"grad_norm": 0.4611315131187439,
"learning_rate": 2.76747726353194e-05,
"loss": 0.0108,
"step": 2835
},
{
"epoch": 5.39,
"grad_norm": 0.22137703001499176,
"learning_rate": 2.7656355935011216e-05,
"loss": 0.0051,
"step": 2836
},
{
"epoch": 5.39,
"grad_norm": 0.286083459854126,
"learning_rate": 2.763794012331156e-05,
"loss": 0.0088,
"step": 2837
},
{
"epoch": 5.4,
"grad_norm": 0.24880795180797577,
"learning_rate": 2.761952520720293e-05,
"loss": 0.0035,
"step": 2838
},
{
"epoch": 5.4,
"grad_norm": 0.5165143013000488,
"learning_rate": 2.7601111193667435e-05,
"loss": 0.0104,
"step": 2839
},
{
"epoch": 5.4,
"grad_norm": 0.5892491936683655,
"learning_rate": 2.75826980896869e-05,
"loss": 0.0317,
"step": 2840
},
{
"epoch": 5.4,
"grad_norm": 0.6384952664375305,
"learning_rate": 2.7564285902242763e-05,
"loss": 0.0092,
"step": 2841
},
{
"epoch": 5.4,
"grad_norm": 0.4227117896080017,
"learning_rate": 2.7545874638316134e-05,
"loss": 0.0259,
"step": 2842
},
{
"epoch": 5.4,
"grad_norm": 0.6741682291030884,
"learning_rate": 2.752746430488776e-05,
"loss": 0.0127,
"step": 2843
},
{
"epoch": 5.41,
"grad_norm": 0.27510708570480347,
"learning_rate": 2.7509054908938064e-05,
"loss": 0.0037,
"step": 2844
},
{
"epoch": 5.41,
"grad_norm": 0.6842405796051025,
"learning_rate": 2.7490646457447072e-05,
"loss": 0.008,
"step": 2845
},
{
"epoch": 5.41,
"grad_norm": 0.8336822390556335,
"learning_rate": 2.747223895739449e-05,
"loss": 0.0073,
"step": 2846
},
{
"epoch": 5.41,
"grad_norm": 0.6142996549606323,
"learning_rate": 2.745383241575963e-05,
"loss": 0.0241,
"step": 2847
},
{
"epoch": 5.41,
"grad_norm": 0.6426823735237122,
"learning_rate": 2.7435426839521465e-05,
"loss": 0.0125,
"step": 2848
},
{
"epoch": 5.42,
"grad_norm": 0.46845611929893494,
"learning_rate": 2.7417022235658605e-05,
"loss": 0.0049,
"step": 2849
},
{
"epoch": 5.42,
"grad_norm": 0.3934462368488312,
"learning_rate": 2.7398618611149255e-05,
"loss": 0.0051,
"step": 2850
},
{
"epoch": 5.42,
"grad_norm": 0.7731006145477295,
"learning_rate": 2.73802159729713e-05,
"loss": 0.0152,
"step": 2851
},
{
"epoch": 5.42,
"grad_norm": 0.5068455934524536,
"learning_rate": 2.736181432810221e-05,
"loss": 0.0088,
"step": 2852
},
{
"epoch": 5.42,
"grad_norm": 0.77860426902771,
"learning_rate": 2.73434136835191e-05,
"loss": 0.0182,
"step": 2853
},
{
"epoch": 5.43,
"grad_norm": 0.2874179184436798,
"learning_rate": 2.7325014046198686e-05,
"loss": 0.0061,
"step": 2854
},
{
"epoch": 5.43,
"grad_norm": 0.5737937688827515,
"learning_rate": 2.7306615423117343e-05,
"loss": 0.0167,
"step": 2855
},
{
"epoch": 5.43,
"grad_norm": 0.48524272441864014,
"learning_rate": 2.7288217821251e-05,
"loss": 0.0067,
"step": 2856
},
{
"epoch": 5.43,
"grad_norm": 0.46408611536026,
"learning_rate": 2.7269821247575247e-05,
"loss": 0.0123,
"step": 2857
},
{
"epoch": 5.43,
"grad_norm": 0.4562472403049469,
"learning_rate": 2.7251425709065283e-05,
"loss": 0.0057,
"step": 2858
},
{
"epoch": 5.44,
"grad_norm": 0.5342998504638672,
"learning_rate": 2.7233031212695882e-05,
"loss": 0.0071,
"step": 2859
},
{
"epoch": 5.44,
"grad_norm": 0.6310628652572632,
"learning_rate": 2.721463776544144e-05,
"loss": 0.0055,
"step": 2860
},
{
"epoch": 5.44,
"grad_norm": 0.5562223792076111,
"learning_rate": 2.7196245374275987e-05,
"loss": 0.0186,
"step": 2861
},
{
"epoch": 5.44,
"grad_norm": 0.46467289328575134,
"learning_rate": 2.7177854046173084e-05,
"loss": 0.0121,
"step": 2862
},
{
"epoch": 5.44,
"grad_norm": 0.53925621509552,
"learning_rate": 2.7159463788105955e-05,
"loss": 0.0154,
"step": 2863
},
{
"epoch": 5.44,
"grad_norm": 0.6220842003822327,
"learning_rate": 2.714107460704738e-05,
"loss": 0.0145,
"step": 2864
},
{
"epoch": 5.45,
"grad_norm": 0.5141053199768066,
"learning_rate": 2.7122686509969744e-05,
"loss": 0.0112,
"step": 2865
},
{
"epoch": 5.45,
"grad_norm": 0.5384431481361389,
"learning_rate": 2.710429950384501e-05,
"loss": 0.0137,
"step": 2866
},
{
"epoch": 5.45,
"grad_norm": 0.49669167399406433,
"learning_rate": 2.708591359564476e-05,
"loss": 0.0138,
"step": 2867
},
{
"epoch": 5.45,
"grad_norm": 0.6472033262252808,
"learning_rate": 2.7067528792340102e-05,
"loss": 0.0158,
"step": 2868
},
{
"epoch": 5.45,
"grad_norm": 0.45430317521095276,
"learning_rate": 2.704914510090179e-05,
"loss": 0.006,
"step": 2869
},
{
"epoch": 5.46,
"grad_norm": 0.590367317199707,
"learning_rate": 2.7030762528300102e-05,
"loss": 0.009,
"step": 2870
},
{
"epoch": 5.46,
"grad_norm": 0.6532981395721436,
"learning_rate": 2.7012381081504922e-05,
"loss": 0.0095,
"step": 2871
},
{
"epoch": 5.46,
"grad_norm": 0.8018893003463745,
"learning_rate": 2.6994000767485718e-05,
"loss": 0.0095,
"step": 2872
},
{
"epoch": 5.46,
"grad_norm": 0.671403169631958,
"learning_rate": 2.697562159321148e-05,
"loss": 0.0242,
"step": 2873
},
{
"epoch": 5.46,
"grad_norm": 0.44924625754356384,
"learning_rate": 2.695724356565082e-05,
"loss": 0.0025,
"step": 2874
},
{
"epoch": 5.47,
"grad_norm": 0.6504111289978027,
"learning_rate": 2.6938866691771883e-05,
"loss": 0.014,
"step": 2875
},
{
"epoch": 5.47,
"grad_norm": 0.8439309597015381,
"learning_rate": 2.692049097854238e-05,
"loss": 0.013,
"step": 2876
},
{
"epoch": 5.47,
"grad_norm": 0.30561748147010803,
"learning_rate": 2.6902116432929593e-05,
"loss": 0.0077,
"step": 2877
},
{
"epoch": 5.47,
"grad_norm": 0.32769307494163513,
"learning_rate": 2.6883743061900364e-05,
"loss": 0.0062,
"step": 2878
},
{
"epoch": 5.47,
"grad_norm": 0.5225582122802734,
"learning_rate": 2.6865370872421062e-05,
"loss": 0.0149,
"step": 2879
},
{
"epoch": 5.48,
"grad_norm": 0.7464303970336914,
"learning_rate": 2.6846999871457636e-05,
"loss": 0.0073,
"step": 2880
},
{
"epoch": 5.48,
"grad_norm": 0.3877076506614685,
"learning_rate": 2.6828630065975594e-05,
"loss": 0.0071,
"step": 2881
},
{
"epoch": 5.48,
"grad_norm": 0.6393812298774719,
"learning_rate": 2.681026146293994e-05,
"loss": 0.0126,
"step": 2882
},
{
"epoch": 5.48,
"grad_norm": 0.6272947788238525,
"learning_rate": 2.679189406931527e-05,
"loss": 0.0097,
"step": 2883
},
{
"epoch": 5.48,
"grad_norm": 0.45332270860671997,
"learning_rate": 2.6773527892065716e-05,
"loss": 0.0072,
"step": 2884
},
{
"epoch": 5.48,
"grad_norm": 0.3141744136810303,
"learning_rate": 2.675516293815492e-05,
"loss": 0.0039,
"step": 2885
},
{
"epoch": 5.49,
"grad_norm": 0.37585705518722534,
"learning_rate": 2.6736799214546092e-05,
"loss": 0.0035,
"step": 2886
},
{
"epoch": 5.49,
"grad_norm": 0.6157782077789307,
"learning_rate": 2.6718436728201956e-05,
"loss": 0.0128,
"step": 2887
},
{
"epoch": 5.49,
"grad_norm": 0.5655303597450256,
"learning_rate": 2.670007548608477e-05,
"loss": 0.0053,
"step": 2888
},
{
"epoch": 5.49,
"grad_norm": 0.6137311458587646,
"learning_rate": 2.6681715495156325e-05,
"loss": 0.0096,
"step": 2889
},
{
"epoch": 5.49,
"grad_norm": 0.40459829568862915,
"learning_rate": 2.6663356762377956e-05,
"loss": 0.0232,
"step": 2890
},
{
"epoch": 5.5,
"grad_norm": 0.7202457189559937,
"learning_rate": 2.6644999294710464e-05,
"loss": 0.0089,
"step": 2891
},
{
"epoch": 5.5,
"grad_norm": 0.2887400984764099,
"learning_rate": 2.6626643099114227e-05,
"loss": 0.0043,
"step": 2892
},
{
"epoch": 5.5,
"grad_norm": 0.4723149240016937,
"learning_rate": 2.660828818254914e-05,
"loss": 0.0093,
"step": 2893
},
{
"epoch": 5.5,
"grad_norm": 0.6761041879653931,
"learning_rate": 2.6589934551974554e-05,
"loss": 0.0199,
"step": 2894
},
{
"epoch": 5.5,
"grad_norm": 0.2740342915058136,
"learning_rate": 2.657158221434941e-05,
"loss": 0.0034,
"step": 2895
},
{
"epoch": 5.51,
"grad_norm": 0.6955597400665283,
"learning_rate": 2.6553231176632092e-05,
"loss": 0.0088,
"step": 2896
},
{
"epoch": 5.51,
"grad_norm": 0.5765194892883301,
"learning_rate": 2.6534881445780542e-05,
"loss": 0.0081,
"step": 2897
},
{
"epoch": 5.51,
"grad_norm": 0.29504600167274475,
"learning_rate": 2.6516533028752167e-05,
"loss": 0.0028,
"step": 2898
},
{
"epoch": 5.51,
"grad_norm": 0.22343313694000244,
"learning_rate": 2.649818593250391e-05,
"loss": 0.0031,
"step": 2899
},
{
"epoch": 5.51,
"grad_norm": 0.5304419994354248,
"learning_rate": 2.6479840163992178e-05,
"loss": 0.0073,
"step": 2900
},
{
"epoch": 5.51,
"eval_blimp_filtered_avg": 0.7201492537313433,
"eval_blimp_filtered_std": 0.004980074425737206,
"step": 2900
},
{
"epoch": 5.51,
"eval_blimp_supplement_avg": 0.7801724137931034,
"eval_blimp_supplement_std": 0.01801118401453912,
"step": 2900
},
{
"epoch": 5.51,
"eval_vqa_filtered_avg": 0.36,
"eval_vqa_filtered_std": 0.048241815132442176,
"step": 2900
},
{
"epoch": 5.51,
"eval_winoground_filtered_avg": 0.51,
"eval_winoground_filtered_std": 0.05024183937956912,
"step": 2900
},
{
"epoch": 5.52,
"grad_norm": 0.35768792033195496,
"learning_rate": 2.646149573017292e-05,
"loss": 0.0027,
"step": 2901
},
{
"epoch": 5.52,
"grad_norm": 0.16112962365150452,
"learning_rate": 2.6443152638001523e-05,
"loss": 0.0041,
"step": 2902
},
{
"epoch": 5.52,
"grad_norm": 0.286082923412323,
"learning_rate": 2.6424810894432913e-05,
"loss": 0.0038,
"step": 2903
},
{
"epoch": 5.52,
"grad_norm": 0.5121301412582397,
"learning_rate": 2.640647050642149e-05,
"loss": 0.0122,
"step": 2904
},
{
"epoch": 5.52,
"grad_norm": 0.7409411072731018,
"learning_rate": 2.6388131480921117e-05,
"loss": 0.0083,
"step": 2905
},
{
"epoch": 5.52,
"grad_norm": 0.5805025696754456,
"learning_rate": 2.636979382488517e-05,
"loss": 0.0064,
"step": 2906
},
{
"epoch": 5.53,
"grad_norm": 0.6522048115730286,
"learning_rate": 2.6351457545266512e-05,
"loss": 0.0086,
"step": 2907
},
{
"epoch": 5.53,
"grad_norm": 0.5614343285560608,
"learning_rate": 2.6333122649017438e-05,
"loss": 0.0112,
"step": 2908
},
{
"epoch": 5.53,
"grad_norm": 0.6549161076545715,
"learning_rate": 2.6314789143089768e-05,
"loss": 0.0105,
"step": 2909
},
{
"epoch": 5.53,
"grad_norm": 0.6155663728713989,
"learning_rate": 2.6296457034434773e-05,
"loss": 0.007,
"step": 2910
},
{
"epoch": 5.53,
"grad_norm": 0.7432111501693726,
"learning_rate": 2.627812633000319e-05,
"loss": 0.01,
"step": 2911
},
{
"epoch": 5.54,
"grad_norm": 0.1166112869977951,
"learning_rate": 2.625979703674523e-05,
"loss": 0.0031,
"step": 2912
},
{
"epoch": 5.54,
"grad_norm": 0.17321813106536865,
"learning_rate": 2.6241469161610584e-05,
"loss": 0.0014,
"step": 2913
},
{
"epoch": 5.54,
"grad_norm": 0.4977991282939911,
"learning_rate": 2.6223142711548358e-05,
"loss": 0.011,
"step": 2914
},
{
"epoch": 5.54,
"grad_norm": 0.24822646379470825,
"learning_rate": 2.620481769350718e-05,
"loss": 0.0029,
"step": 2915
},
{
"epoch": 5.54,
"grad_norm": 0.42407673597335815,
"learning_rate": 2.6186494114435094e-05,
"loss": 0.0035,
"step": 2916
},
{
"epoch": 5.55,
"grad_norm": 0.46724459528923035,
"learning_rate": 2.6168171981279608e-05,
"loss": 0.0098,
"step": 2917
},
{
"epoch": 5.55,
"grad_norm": 0.7849872708320618,
"learning_rate": 2.614985130098769e-05,
"loss": 0.0145,
"step": 2918
},
{
"epoch": 5.55,
"grad_norm": 0.25039568543434143,
"learning_rate": 2.6131532080505734e-05,
"loss": 0.0038,
"step": 2919
},
{
"epoch": 5.55,
"grad_norm": 0.48938724398612976,
"learning_rate": 2.6113214326779613e-05,
"loss": 0.0061,
"step": 2920
},
{
"epoch": 5.55,
"grad_norm": 1.1325254440307617,
"learning_rate": 2.6094898046754624e-05,
"loss": 0.0175,
"step": 2921
},
{
"epoch": 5.56,
"grad_norm": 0.5243287682533264,
"learning_rate": 2.6076583247375502e-05,
"loss": 0.0078,
"step": 2922
},
{
"epoch": 5.56,
"grad_norm": 0.30546247959136963,
"learning_rate": 2.605826993558643e-05,
"loss": 0.0049,
"step": 2923
},
{
"epoch": 5.56,
"grad_norm": 0.7112234830856323,
"learning_rate": 2.6039958118331034e-05,
"loss": 0.0088,
"step": 2924
},
{
"epoch": 5.56,
"grad_norm": 0.4688345193862915,
"learning_rate": 2.6021647802552347e-05,
"loss": 0.0094,
"step": 2925
},
{
"epoch": 5.56,
"grad_norm": 1.8279430866241455,
"learning_rate": 2.600333899519285e-05,
"loss": 0.0297,
"step": 2926
},
{
"epoch": 5.56,
"grad_norm": 0.8752596378326416,
"learning_rate": 2.598503170319448e-05,
"loss": 0.0134,
"step": 2927
},
{
"epoch": 5.57,
"grad_norm": 0.24282385408878326,
"learning_rate": 2.5966725933498533e-05,
"loss": 0.0026,
"step": 2928
},
{
"epoch": 5.57,
"grad_norm": 0.391885906457901,
"learning_rate": 2.5948421693045784e-05,
"loss": 0.0037,
"step": 2929
},
{
"epoch": 5.57,
"grad_norm": 0.42818641662597656,
"learning_rate": 2.593011898877642e-05,
"loss": 0.0064,
"step": 2930
},
{
"epoch": 5.57,
"grad_norm": 0.13558652997016907,
"learning_rate": 2.5911817827630012e-05,
"loss": 0.0039,
"step": 2931
},
{
"epoch": 5.57,
"grad_norm": 0.6378859877586365,
"learning_rate": 2.5893518216545587e-05,
"loss": 0.0137,
"step": 2932
},
{
"epoch": 5.58,
"grad_norm": 0.8934276103973389,
"learning_rate": 2.587522016246156e-05,
"loss": 0.0122,
"step": 2933
},
{
"epoch": 5.58,
"grad_norm": 0.43239331245422363,
"learning_rate": 2.5856923672315765e-05,
"loss": 0.0059,
"step": 2934
},
{
"epoch": 5.58,
"grad_norm": 0.5727044343948364,
"learning_rate": 2.583862875304543e-05,
"loss": 0.0065,
"step": 2935
},
{
"epoch": 5.58,
"grad_norm": 0.46607694029808044,
"learning_rate": 2.5820335411587217e-05,
"loss": 0.0122,
"step": 2936
},
{
"epoch": 5.58,
"grad_norm": 0.6143919825553894,
"learning_rate": 2.5802043654877146e-05,
"loss": 0.004,
"step": 2937
},
{
"epoch": 5.59,
"grad_norm": 0.3344140648841858,
"learning_rate": 2.5783753489850677e-05,
"loss": 0.0044,
"step": 2938
},
{
"epoch": 5.59,
"grad_norm": 1.013702392578125,
"learning_rate": 2.5765464923442655e-05,
"loss": 0.0085,
"step": 2939
},
{
"epoch": 5.59,
"grad_norm": 0.2905058264732361,
"learning_rate": 2.5747177962587296e-05,
"loss": 0.004,
"step": 2940
},
{
"epoch": 5.59,
"grad_norm": 0.5076903700828552,
"learning_rate": 2.5728892614218235e-05,
"loss": 0.004,
"step": 2941
},
{
"epoch": 5.59,
"grad_norm": 0.36029288172721863,
"learning_rate": 2.5710608885268495e-05,
"loss": 0.0064,
"step": 2942
},
{
"epoch": 5.6,
"grad_norm": 0.37828031182289124,
"learning_rate": 2.5692326782670463e-05,
"loss": 0.0104,
"step": 2943
},
{
"epoch": 5.6,
"grad_norm": 0.48675331473350525,
"learning_rate": 2.567404631335592e-05,
"loss": 0.0085,
"step": 2944
},
{
"epoch": 5.6,
"grad_norm": 0.6519892811775208,
"learning_rate": 2.565576748425604e-05,
"loss": 0.0134,
"step": 2945
},
{
"epoch": 5.6,
"grad_norm": 0.9044387936592102,
"learning_rate": 2.5637490302301355e-05,
"loss": 0.0063,
"step": 2946
},
{
"epoch": 5.6,
"grad_norm": 0.38932836055755615,
"learning_rate": 2.56192147744218e-05,
"loss": 0.0043,
"step": 2947
},
{
"epoch": 5.6,
"grad_norm": 0.39667364954948425,
"learning_rate": 2.5600940907546635e-05,
"loss": 0.0113,
"step": 2948
},
{
"epoch": 5.61,
"grad_norm": 0.2929546535015106,
"learning_rate": 2.558266870860454e-05,
"loss": 0.0042,
"step": 2949
},
{
"epoch": 5.61,
"grad_norm": 0.3363408148288727,
"learning_rate": 2.5564398184523553e-05,
"loss": 0.0042,
"step": 2950
},
{
"epoch": 5.61,
"grad_norm": 0.3681693375110626,
"learning_rate": 2.5546129342231043e-05,
"loss": 0.0042,
"step": 2951
},
{
"epoch": 5.61,
"grad_norm": 0.21494849026203156,
"learning_rate": 2.552786218865378e-05,
"loss": 0.0025,
"step": 2952
},
{
"epoch": 5.61,
"grad_norm": 0.2399398535490036,
"learning_rate": 2.5509596730717888e-05,
"loss": 0.0045,
"step": 2953
},
{
"epoch": 5.62,
"grad_norm": 0.623058557510376,
"learning_rate": 2.5491332975348815e-05,
"loss": 0.0111,
"step": 2954
},
{
"epoch": 5.62,
"grad_norm": 0.5116754174232483,
"learning_rate": 2.5473070929471404e-05,
"loss": 0.0065,
"step": 2955
},
{
"epoch": 5.62,
"grad_norm": 0.6679522395133972,
"learning_rate": 2.5454810600009835e-05,
"loss": 0.0149,
"step": 2956
},
{
"epoch": 5.62,
"grad_norm": 0.6347626447677612,
"learning_rate": 2.543655199388763e-05,
"loss": 0.0059,
"step": 2957
},
{
"epoch": 5.62,
"grad_norm": 0.39118558168411255,
"learning_rate": 2.541829511802765e-05,
"loss": 0.0022,
"step": 2958
},
{
"epoch": 5.63,
"grad_norm": 0.3903129994869232,
"learning_rate": 2.540003997935215e-05,
"loss": 0.0039,
"step": 2959
},
{
"epoch": 5.63,
"grad_norm": 0.48886796832084656,
"learning_rate": 2.538178658478265e-05,
"loss": 0.0042,
"step": 2960
},
{
"epoch": 5.63,
"grad_norm": 0.5054038166999817,
"learning_rate": 2.536353494124006e-05,
"loss": 0.0065,
"step": 2961
},
{
"epoch": 5.63,
"grad_norm": 0.3844326436519623,
"learning_rate": 2.5345285055644635e-05,
"loss": 0.0058,
"step": 2962
},
{
"epoch": 5.63,
"grad_norm": 0.5692445039749146,
"learning_rate": 2.5327036934915914e-05,
"loss": 0.0083,
"step": 2963
},
{
"epoch": 5.63,
"grad_norm": 0.46144339442253113,
"learning_rate": 2.5308790585972812e-05,
"loss": 0.01,
"step": 2964
},
{
"epoch": 5.64,
"grad_norm": 0.5184291005134583,
"learning_rate": 2.5290546015733563e-05,
"loss": 0.0074,
"step": 2965
},
{
"epoch": 5.64,
"grad_norm": 0.33393561840057373,
"learning_rate": 2.52723032311157e-05,
"loss": 0.009,
"step": 2966
},
{
"epoch": 5.64,
"grad_norm": 0.6981989145278931,
"learning_rate": 2.525406223903611e-05,
"loss": 0.0113,
"step": 2967
},
{
"epoch": 5.64,
"grad_norm": 1.1750779151916504,
"learning_rate": 2.5235823046410988e-05,
"loss": 0.0517,
"step": 2968
},
{
"epoch": 5.64,
"grad_norm": 0.6238927245140076,
"learning_rate": 2.521758566015584e-05,
"loss": 0.0067,
"step": 2969
},
{
"epoch": 5.65,
"grad_norm": 0.48219260573387146,
"learning_rate": 2.519935008718551e-05,
"loss": 0.0047,
"step": 2970
},
{
"epoch": 5.65,
"grad_norm": 0.7070558667182922,
"learning_rate": 2.5181116334414112e-05,
"loss": 0.0157,
"step": 2971
},
{
"epoch": 5.65,
"grad_norm": 0.5255359411239624,
"learning_rate": 2.516288440875512e-05,
"loss": 0.0078,
"step": 2972
},
{
"epoch": 5.65,
"grad_norm": 0.41906237602233887,
"learning_rate": 2.51446543171213e-05,
"loss": 0.0074,
"step": 2973
},
{
"epoch": 5.65,
"grad_norm": 0.4237635135650635,
"learning_rate": 2.5126426066424685e-05,
"loss": 0.0032,
"step": 2974
},
{
"epoch": 5.66,
"grad_norm": 1.0574781894683838,
"learning_rate": 2.5108199663576658e-05,
"loss": 0.0195,
"step": 2975
},
{
"epoch": 5.66,
"grad_norm": 0.5864211320877075,
"learning_rate": 2.5089975115487894e-05,
"loss": 0.0049,
"step": 2976
},
{
"epoch": 5.66,
"grad_norm": 0.6348299980163574,
"learning_rate": 2.507175242906833e-05,
"loss": 0.0073,
"step": 2977
},
{
"epoch": 5.66,
"grad_norm": 0.46735942363739014,
"learning_rate": 2.5053531611227236e-05,
"loss": 0.0039,
"step": 2978
},
{
"epoch": 5.66,
"grad_norm": 0.7203087210655212,
"learning_rate": 2.5035312668873153e-05,
"loss": 0.0083,
"step": 2979
},
{
"epoch": 5.67,
"grad_norm": 0.4927626848220825,
"learning_rate": 2.501709560891392e-05,
"loss": 0.0158,
"step": 2980
},
{
"epoch": 5.67,
"grad_norm": 0.5889272689819336,
"learning_rate": 2.4998880438256654e-05,
"loss": 0.012,
"step": 2981
},
{
"epoch": 5.67,
"grad_norm": 0.5449540019035339,
"learning_rate": 2.498066716380777e-05,
"loss": 0.0099,
"step": 2982
},
{
"epoch": 5.67,
"grad_norm": 0.2527967393398285,
"learning_rate": 2.4962455792472937e-05,
"loss": 0.0042,
"step": 2983
},
{
"epoch": 5.67,
"grad_norm": 0.5926845073699951,
"learning_rate": 2.494424633115713e-05,
"loss": 0.0087,
"step": 2984
},
{
"epoch": 5.67,
"grad_norm": 0.37962663173675537,
"learning_rate": 2.4926038786764607e-05,
"loss": 0.0043,
"step": 2985
},
{
"epoch": 5.68,
"grad_norm": 0.37678277492523193,
"learning_rate": 2.4907833166198848e-05,
"loss": 0.0061,
"step": 2986
},
{
"epoch": 5.68,
"grad_norm": 0.6339150071144104,
"learning_rate": 2.4889629476362658e-05,
"loss": 0.016,
"step": 2987
},
{
"epoch": 5.68,
"grad_norm": 0.4827187955379486,
"learning_rate": 2.4871427724158097e-05,
"loss": 0.0054,
"step": 2988
},
{
"epoch": 5.68,
"grad_norm": 0.7830893993377686,
"learning_rate": 2.4853227916486466e-05,
"loss": 0.0252,
"step": 2989
},
{
"epoch": 5.68,
"grad_norm": 0.36064523458480835,
"learning_rate": 2.483503006024834e-05,
"loss": 0.0202,
"step": 2990
},
{
"epoch": 5.69,
"grad_norm": 0.3324579894542694,
"learning_rate": 2.481683416234359e-05,
"loss": 0.002,
"step": 2991
},
{
"epoch": 5.69,
"grad_norm": 0.8264389038085938,
"learning_rate": 2.479864022967128e-05,
"loss": 0.012,
"step": 2992
},
{
"epoch": 5.69,
"grad_norm": 0.480388343334198,
"learning_rate": 2.478044826912979e-05,
"loss": 0.0119,
"step": 2993
},
{
"epoch": 5.69,
"grad_norm": 0.5527312755584717,
"learning_rate": 2.4762258287616695e-05,
"loss": 0.0089,
"step": 2994
},
{
"epoch": 5.69,
"grad_norm": 0.7051040530204773,
"learning_rate": 2.474407029202887e-05,
"loss": 0.0129,
"step": 2995
},
{
"epoch": 5.7,
"grad_norm": 0.7584980130195618,
"learning_rate": 2.4725884289262432e-05,
"loss": 0.0186,
"step": 2996
},
{
"epoch": 5.7,
"grad_norm": 0.6621697545051575,
"learning_rate": 2.470770028621269e-05,
"loss": 0.0101,
"step": 2997
},
{
"epoch": 5.7,
"grad_norm": 0.40191176533699036,
"learning_rate": 2.4689518289774258e-05,
"loss": 0.0088,
"step": 2998
},
{
"epoch": 5.7,
"grad_norm": 0.25088006258010864,
"learning_rate": 2.4671338306840964e-05,
"loss": 0.0046,
"step": 2999
},
{
"epoch": 5.7,
"grad_norm": 0.581766664981842,
"learning_rate": 2.4653160344305852e-05,
"loss": 0.0082,
"step": 3000
},
{
"epoch": 5.7,
"eval_blimp_filtered_avg": 0.7252238805970149,
"eval_blimp_filtered_std": 0.004940426265802327,
"step": 3000
},
{
"epoch": 5.7,
"eval_blimp_supplement_avg": 0.7780172413793104,
"eval_blimp_supplement_std": 0.01807494061750125,
"step": 3000
},
{
"epoch": 5.7,
"eval_vqa_filtered_avg": 0.34,
"eval_vqa_filtered_std": 0.04760952285695235,
"step": 3000
},
{
"epoch": 5.7,
"eval_winoground_filtered_avg": 0.54,
"eval_winoground_filtered_std": 0.05009082659620333,
"step": 3000
},
{
"epoch": 5.71,
"grad_norm": 0.5536354184150696,
"learning_rate": 2.463498440906124e-05,
"loss": 0.0206,
"step": 3001
},
{
"epoch": 5.71,
"grad_norm": 0.6416809558868408,
"learning_rate": 2.4616810507998647e-05,
"loss": 0.0076,
"step": 3002
},
{
"epoch": 5.71,
"grad_norm": 0.24707497656345367,
"learning_rate": 2.4598638648008834e-05,
"loss": 0.0015,
"step": 3003
},
{
"epoch": 5.71,
"grad_norm": 0.8356097340583801,
"learning_rate": 2.4580468835981778e-05,
"loss": 0.0118,
"step": 3004
},
{
"epoch": 5.71,
"grad_norm": 0.44287392497062683,
"learning_rate": 2.4562301078806705e-05,
"loss": 0.0059,
"step": 3005
},
{
"epoch": 5.71,
"grad_norm": 0.4870359003543854,
"learning_rate": 2.4544135383372015e-05,
"loss": 0.0062,
"step": 3006
},
{
"epoch": 5.72,
"grad_norm": 0.6240509152412415,
"learning_rate": 2.4525971756565372e-05,
"loss": 0.0135,
"step": 3007
},
{
"epoch": 5.72,
"grad_norm": 0.5650778412818909,
"learning_rate": 2.4507810205273647e-05,
"loss": 0.0079,
"step": 3008
},
{
"epoch": 5.72,
"grad_norm": 0.25420212745666504,
"learning_rate": 2.4489650736382892e-05,
"loss": 0.0015,
"step": 3009
},
{
"epoch": 5.72,
"grad_norm": 0.3340098261833191,
"learning_rate": 2.4471493356778407e-05,
"loss": 0.0022,
"step": 3010
},
{
"epoch": 5.72,
"grad_norm": 0.3140926659107208,
"learning_rate": 2.4453338073344694e-05,
"loss": 0.0038,
"step": 3011
},
{
"epoch": 5.73,
"grad_norm": 0.5158955454826355,
"learning_rate": 2.4435184892965436e-05,
"loss": 0.0119,
"step": 3012
},
{
"epoch": 5.73,
"grad_norm": 0.7740733027458191,
"learning_rate": 2.441703382252353e-05,
"loss": 0.018,
"step": 3013
},
{
"epoch": 5.73,
"grad_norm": 0.5009097456932068,
"learning_rate": 2.4398884868901102e-05,
"loss": 0.0165,
"step": 3014
},
{
"epoch": 5.73,
"grad_norm": 0.6356241106987,
"learning_rate": 2.4380738038979425e-05,
"loss": 0.0062,
"step": 3015
},
{
"epoch": 5.73,
"grad_norm": 0.8227708339691162,
"learning_rate": 2.4362593339639013e-05,
"loss": 0.0168,
"step": 3016
},
{
"epoch": 5.74,
"grad_norm": 0.2607153356075287,
"learning_rate": 2.434445077775953e-05,
"loss": 0.0031,
"step": 3017
},
{
"epoch": 5.74,
"grad_norm": 0.37923452258110046,
"learning_rate": 2.4326310360219865e-05,
"loss": 0.0048,
"step": 3018
},
{
"epoch": 5.74,
"grad_norm": 0.6764603853225708,
"learning_rate": 2.4308172093898087e-05,
"loss": 0.012,
"step": 3019
},
{
"epoch": 5.74,
"grad_norm": 0.5025920867919922,
"learning_rate": 2.4290035985671426e-05,
"loss": 0.011,
"step": 3020
},
{
"epoch": 5.74,
"grad_norm": 0.5170683860778809,
"learning_rate": 2.4271902042416313e-05,
"loss": 0.0067,
"step": 3021
},
{
"epoch": 5.75,
"grad_norm": 0.4370313882827759,
"learning_rate": 2.425377027100838e-05,
"loss": 0.0077,
"step": 3022
},
{
"epoch": 5.75,
"grad_norm": 0.4232877492904663,
"learning_rate": 2.4235640678322377e-05,
"loss": 0.0075,
"step": 3023
},
{
"epoch": 5.75,
"grad_norm": 0.5512738227844238,
"learning_rate": 2.421751327123228e-05,
"loss": 0.0056,
"step": 3024
},
{
"epoch": 5.75,
"grad_norm": 0.7510284781455994,
"learning_rate": 2.419938805661122e-05,
"loss": 0.0182,
"step": 3025
},
{
"epoch": 5.75,
"grad_norm": 0.829243004322052,
"learning_rate": 2.418126504133149e-05,
"loss": 0.0126,
"step": 3026
},
{
"epoch": 5.75,
"grad_norm": 0.8013080954551697,
"learning_rate": 2.416314423226454e-05,
"loss": 0.0197,
"step": 3027
},
{
"epoch": 5.76,
"grad_norm": 0.53065025806427,
"learning_rate": 2.4145025636281035e-05,
"loss": 0.0056,
"step": 3028
},
{
"epoch": 5.76,
"grad_norm": 0.49869245290756226,
"learning_rate": 2.4126909260250723e-05,
"loss": 0.0085,
"step": 3029
},
{
"epoch": 5.76,
"grad_norm": 0.4031859040260315,
"learning_rate": 2.4108795111042565e-05,
"loss": 0.0059,
"step": 3030
},
{
"epoch": 5.76,
"grad_norm": 1.0594654083251953,
"learning_rate": 2.4090683195524684e-05,
"loss": 0.0103,
"step": 3031
},
{
"epoch": 5.76,
"grad_norm": 0.3268260359764099,
"learning_rate": 2.4072573520564305e-05,
"loss": 0.0048,
"step": 3032
},
{
"epoch": 5.77,
"grad_norm": 0.1961507499217987,
"learning_rate": 2.405446609302784e-05,
"loss": 0.0046,
"step": 3033
},
{
"epoch": 5.77,
"grad_norm": 0.33655256032943726,
"learning_rate": 2.403636091978087e-05,
"loss": 0.0059,
"step": 3034
},
{
"epoch": 5.77,
"grad_norm": 0.5173361301422119,
"learning_rate": 2.4018258007688066e-05,
"loss": 0.0054,
"step": 3035
},
{
"epoch": 5.77,
"grad_norm": 0.55496746301651,
"learning_rate": 2.400015736361327e-05,
"loss": 0.0078,
"step": 3036
},
{
"epoch": 5.77,
"grad_norm": 0.21624338626861572,
"learning_rate": 2.3982058994419482e-05,
"loss": 0.003,
"step": 3037
},
{
"epoch": 5.78,
"grad_norm": 0.6227338910102844,
"learning_rate": 2.39639629069688e-05,
"loss": 0.0077,
"step": 3038
},
{
"epoch": 5.78,
"grad_norm": 0.5210617184638977,
"learning_rate": 2.394586910812249e-05,
"loss": 0.0046,
"step": 3039
},
{
"epoch": 5.78,
"grad_norm": 0.3098476827144623,
"learning_rate": 2.3927777604740953e-05,
"loss": 0.0028,
"step": 3040
},
{
"epoch": 5.78,
"grad_norm": 0.4937729239463806,
"learning_rate": 2.3909688403683675e-05,
"loss": 0.01,
"step": 3041
},
{
"epoch": 5.78,
"grad_norm": 0.48026394844055176,
"learning_rate": 2.3891601511809324e-05,
"loss": 0.0081,
"step": 3042
},
{
"epoch": 5.79,
"grad_norm": 0.3703099191188812,
"learning_rate": 2.3873516935975647e-05,
"loss": 0.0106,
"step": 3043
},
{
"epoch": 5.79,
"grad_norm": 0.5925240516662598,
"learning_rate": 2.3855434683039544e-05,
"loss": 0.008,
"step": 3044
},
{
"epoch": 5.79,
"grad_norm": 0.6185641288757324,
"learning_rate": 2.383735475985704e-05,
"loss": 0.0131,
"step": 3045
},
{
"epoch": 5.79,
"grad_norm": 0.4271090626716614,
"learning_rate": 2.3819277173283223e-05,
"loss": 0.0069,
"step": 3046
},
{
"epoch": 5.79,
"grad_norm": 0.20361751317977905,
"learning_rate": 2.3801201930172363e-05,
"loss": 0.0027,
"step": 3047
},
{
"epoch": 5.79,
"grad_norm": 0.4757899343967438,
"learning_rate": 2.3783129037377804e-05,
"loss": 0.004,
"step": 3048
},
{
"epoch": 5.8,
"grad_norm": 0.48286470770835876,
"learning_rate": 2.3765058501752003e-05,
"loss": 0.0041,
"step": 3049
},
{
"epoch": 5.8,
"grad_norm": 0.2777113616466522,
"learning_rate": 2.3746990330146518e-05,
"loss": 0.0047,
"step": 3050
},
{
"epoch": 5.8,
"grad_norm": 0.5065255165100098,
"learning_rate": 2.3728924529412043e-05,
"loss": 0.0129,
"step": 3051
},
{
"epoch": 5.8,
"grad_norm": 0.5484614968299866,
"learning_rate": 2.3710861106398316e-05,
"loss": 0.0156,
"step": 3052
},
{
"epoch": 5.8,
"grad_norm": 0.3452228307723999,
"learning_rate": 2.3692800067954226e-05,
"loss": 0.0047,
"step": 3053
},
{
"epoch": 5.81,
"grad_norm": 0.5226677656173706,
"learning_rate": 2.3674741420927745e-05,
"loss": 0.0042,
"step": 3054
},
{
"epoch": 5.81,
"grad_norm": 0.34155282378196716,
"learning_rate": 2.3656685172165913e-05,
"loss": 0.0062,
"step": 3055
},
{
"epoch": 5.81,
"grad_norm": 0.6832013130187988,
"learning_rate": 2.3638631328514887e-05,
"loss": 0.0252,
"step": 3056
},
{
"epoch": 5.81,
"grad_norm": 0.27093780040740967,
"learning_rate": 2.3620579896819912e-05,
"loss": 0.0039,
"step": 3057
},
{
"epoch": 5.81,
"grad_norm": 0.6265966892242432,
"learning_rate": 2.3602530883925306e-05,
"loss": 0.0119,
"step": 3058
},
{
"epoch": 5.82,
"grad_norm": 0.33013275265693665,
"learning_rate": 2.3584484296674463e-05,
"loss": 0.0094,
"step": 3059
},
{
"epoch": 5.82,
"grad_norm": 0.4164813756942749,
"learning_rate": 2.356644014190989e-05,
"loss": 0.0091,
"step": 3060
},
{
"epoch": 5.82,
"grad_norm": 0.5117319822311401,
"learning_rate": 2.354839842647312e-05,
"loss": 0.0061,
"step": 3061
},
{
"epoch": 5.82,
"grad_norm": 0.3875889480113983,
"learning_rate": 2.3530359157204818e-05,
"loss": 0.0066,
"step": 3062
},
{
"epoch": 5.82,
"grad_norm": 0.4512465000152588,
"learning_rate": 2.35123223409447e-05,
"loss": 0.0067,
"step": 3063
},
{
"epoch": 5.83,
"grad_norm": 0.6785383224487305,
"learning_rate": 2.3494287984531515e-05,
"loss": 0.0109,
"step": 3064
},
{
"epoch": 5.83,
"grad_norm": 0.47951585054397583,
"learning_rate": 2.347625609480315e-05,
"loss": 0.0045,
"step": 3065
},
{
"epoch": 5.83,
"grad_norm": 0.31853848695755005,
"learning_rate": 2.3458226678596485e-05,
"loss": 0.0033,
"step": 3066
},
{
"epoch": 5.83,
"grad_norm": 0.87542724609375,
"learning_rate": 2.3440199742747508e-05,
"loss": 0.0092,
"step": 3067
},
{
"epoch": 5.83,
"grad_norm": 0.43346911668777466,
"learning_rate": 2.3422175294091268e-05,
"loss": 0.0035,
"step": 3068
},
{
"epoch": 5.83,
"grad_norm": 0.47433724999427795,
"learning_rate": 2.3404153339461837e-05,
"loss": 0.007,
"step": 3069
},
{
"epoch": 5.84,
"grad_norm": 0.298163503408432,
"learning_rate": 2.3386133885692368e-05,
"loss": 0.0025,
"step": 3070
},
{
"epoch": 5.84,
"grad_norm": 0.4712491035461426,
"learning_rate": 2.3368116939615062e-05,
"loss": 0.0091,
"step": 3071
},
{
"epoch": 5.84,
"grad_norm": 0.24634528160095215,
"learning_rate": 2.3350102508061167e-05,
"loss": 0.006,
"step": 3072
},
{
"epoch": 5.84,
"grad_norm": 0.8681941628456116,
"learning_rate": 2.3332090597860963e-05,
"loss": 0.0159,
"step": 3073
},
{
"epoch": 5.84,
"grad_norm": 0.5991348028182983,
"learning_rate": 2.331408121584381e-05,
"loss": 0.0076,
"step": 3074
},
{
"epoch": 5.85,
"grad_norm": 0.35161665081977844,
"learning_rate": 2.329607436883807e-05,
"loss": 0.0027,
"step": 3075
},
{
"epoch": 5.85,
"grad_norm": 0.4838812053203583,
"learning_rate": 2.327807006367116e-05,
"loss": 0.0068,
"step": 3076
},
{
"epoch": 5.85,
"grad_norm": 0.3934789299964905,
"learning_rate": 2.3260068307169553e-05,
"loss": 0.0048,
"step": 3077
},
{
"epoch": 5.85,
"grad_norm": 0.4447924792766571,
"learning_rate": 2.3242069106158718e-05,
"loss": 0.0075,
"step": 3078
},
{
"epoch": 5.85,
"grad_norm": 0.6664283275604248,
"learning_rate": 2.322407246746318e-05,
"loss": 0.0179,
"step": 3079
},
{
"epoch": 5.86,
"grad_norm": 0.43585553765296936,
"learning_rate": 2.3206078397906495e-05,
"loss": 0.0053,
"step": 3080
},
{
"epoch": 5.86,
"grad_norm": 1.4159910678863525,
"learning_rate": 2.318808690431123e-05,
"loss": 0.0066,
"step": 3081
},
{
"epoch": 5.86,
"grad_norm": 0.48957493901252747,
"learning_rate": 2.317009799349897e-05,
"loss": 0.0123,
"step": 3082
},
{
"epoch": 5.86,
"grad_norm": 0.33846062421798706,
"learning_rate": 2.3152111672290357e-05,
"loss": 0.0044,
"step": 3083
},
{
"epoch": 5.86,
"grad_norm": 0.45614397525787354,
"learning_rate": 2.3134127947504997e-05,
"loss": 0.0079,
"step": 3084
},
{
"epoch": 5.87,
"grad_norm": 0.505937933921814,
"learning_rate": 2.3116146825961562e-05,
"loss": 0.0133,
"step": 3085
},
{
"epoch": 5.87,
"grad_norm": 0.4856806695461273,
"learning_rate": 2.309816831447772e-05,
"loss": 0.0052,
"step": 3086
},
{
"epoch": 5.87,
"grad_norm": 0.5389716625213623,
"learning_rate": 2.3080192419870127e-05,
"loss": 0.0102,
"step": 3087
},
{
"epoch": 5.87,
"grad_norm": 0.1981583833694458,
"learning_rate": 2.3062219148954474e-05,
"loss": 0.0022,
"step": 3088
},
{
"epoch": 5.87,
"grad_norm": 0.3050638735294342,
"learning_rate": 2.304424850854546e-05,
"loss": 0.0049,
"step": 3089
},
{
"epoch": 5.87,
"grad_norm": 0.8551312685012817,
"learning_rate": 2.3026280505456752e-05,
"loss": 0.0106,
"step": 3090
},
{
"epoch": 5.88,
"grad_norm": 0.4892871677875519,
"learning_rate": 2.3008315146501066e-05,
"loss": 0.0099,
"step": 3091
},
{
"epoch": 5.88,
"grad_norm": 0.721844494342804,
"learning_rate": 2.2990352438490066e-05,
"loss": 0.0209,
"step": 3092
},
{
"epoch": 5.88,
"grad_norm": 0.7635865807533264,
"learning_rate": 2.2972392388234446e-05,
"loss": 0.0126,
"step": 3093
},
{
"epoch": 5.88,
"grad_norm": 0.24350158870220184,
"learning_rate": 2.2954435002543885e-05,
"loss": 0.0031,
"step": 3094
},
{
"epoch": 5.88,
"grad_norm": 0.3142109811306,
"learning_rate": 2.2936480288227035e-05,
"loss": 0.0056,
"step": 3095
},
{
"epoch": 5.89,
"grad_norm": 0.38689592480659485,
"learning_rate": 2.291852825209155e-05,
"loss": 0.0048,
"step": 3096
},
{
"epoch": 5.89,
"grad_norm": 0.6465576887130737,
"learning_rate": 2.2900578900944086e-05,
"loss": 0.0135,
"step": 3097
},
{
"epoch": 5.89,
"grad_norm": 0.5101389288902283,
"learning_rate": 2.2882632241590233e-05,
"loss": 0.008,
"step": 3098
},
{
"epoch": 5.89,
"grad_norm": 0.8862186074256897,
"learning_rate": 2.28646882808346e-05,
"loss": 0.0159,
"step": 3099
},
{
"epoch": 5.89,
"grad_norm": 0.3941279351711273,
"learning_rate": 2.284674702548077e-05,
"loss": 0.0063,
"step": 3100
},
{
"epoch": 5.89,
"eval_blimp_filtered_avg": 0.7197014925373134,
"eval_blimp_filtered_std": 0.004976435828958852,
"step": 3100
},
{
"epoch": 5.89,
"eval_blimp_supplement_avg": 0.7801724137931034,
"eval_blimp_supplement_std": 0.017917007350357304,
"step": 3100
},
{
"epoch": 5.89,
"eval_vqa_filtered_avg": 0.34,
"eval_vqa_filtered_std": 0.04760952285695235,
"step": 3100
},
{
"epoch": 5.89,
"eval_winoground_filtered_avg": 0.53,
"eval_winoground_filtered_std": 0.05016135580465919,
"step": 3100
},
{
"epoch": 5.9,
"grad_norm": 0.6813041567802429,
"learning_rate": 2.282880848233128e-05,
"loss": 0.0118,
"step": 3101
},
{
"epoch": 5.9,
"grad_norm": 0.36645177006721497,
"learning_rate": 2.281087265818766e-05,
"loss": 0.004,
"step": 3102
},
{
"epoch": 5.9,
"grad_norm": 0.6210029125213623,
"learning_rate": 2.279293955985039e-05,
"loss": 0.0081,
"step": 3103
},
{
"epoch": 5.9,
"grad_norm": 0.32746678590774536,
"learning_rate": 2.277500919411893e-05,
"loss": 0.0086,
"step": 3104
},
{
"epoch": 5.9,
"grad_norm": 0.3149266541004181,
"learning_rate": 2.2757081567791695e-05,
"loss": 0.0065,
"step": 3105
},
{
"epoch": 5.9,
"grad_norm": 0.18683135509490967,
"learning_rate": 2.273915668766608e-05,
"loss": 0.0031,
"step": 3106
},
{
"epoch": 5.91,
"grad_norm": 0.5401168465614319,
"learning_rate": 2.2721234560538403e-05,
"loss": 0.0119,
"step": 3107
},
{
"epoch": 5.91,
"grad_norm": 0.4191358983516693,
"learning_rate": 2.270331519320397e-05,
"loss": 0.0077,
"step": 3108
},
{
"epoch": 5.91,
"grad_norm": 0.5179948806762695,
"learning_rate": 2.268539859245705e-05,
"loss": 0.0049,
"step": 3109
},
{
"epoch": 5.91,
"grad_norm": 0.5850247144699097,
"learning_rate": 2.266748476509081e-05,
"loss": 0.0106,
"step": 3110
},
{
"epoch": 5.91,
"grad_norm": 0.6729183793067932,
"learning_rate": 2.2649573717897415e-05,
"loss": 0.016,
"step": 3111
},
{
"epoch": 5.92,
"grad_norm": 0.1664877086877823,
"learning_rate": 2.263166545766797e-05,
"loss": 0.0015,
"step": 3112
},
{
"epoch": 5.92,
"grad_norm": 0.5753315687179565,
"learning_rate": 2.2613759991192498e-05,
"loss": 0.0048,
"step": 3113
},
{
"epoch": 5.92,
"grad_norm": 0.46321824193000793,
"learning_rate": 2.2595857325259992e-05,
"loss": 0.0073,
"step": 3114
},
{
"epoch": 5.92,
"grad_norm": 0.6212788820266724,
"learning_rate": 2.2577957466658347e-05,
"loss": 0.0136,
"step": 3115
},
{
"epoch": 5.92,
"grad_norm": 0.1692100167274475,
"learning_rate": 2.2560060422174433e-05,
"loss": 0.0024,
"step": 3116
},
{
"epoch": 5.93,
"grad_norm": 0.39871180057525635,
"learning_rate": 2.2542166198594036e-05,
"loss": 0.012,
"step": 3117
},
{
"epoch": 5.93,
"grad_norm": 0.6243621110916138,
"learning_rate": 2.2524274802701865e-05,
"loss": 0.0068,
"step": 3118
},
{
"epoch": 5.93,
"grad_norm": 0.5113845467567444,
"learning_rate": 2.2506386241281564e-05,
"loss": 0.0132,
"step": 3119
},
{
"epoch": 5.93,
"grad_norm": 0.42898932099342346,
"learning_rate": 2.2488500521115716e-05,
"loss": 0.007,
"step": 3120
},
{
"epoch": 5.93,
"grad_norm": 0.34930282831192017,
"learning_rate": 2.2470617648985786e-05,
"loss": 0.0054,
"step": 3121
},
{
"epoch": 5.94,
"grad_norm": 1.1428364515304565,
"learning_rate": 2.245273763167221e-05,
"loss": 0.0042,
"step": 3122
},
{
"epoch": 5.94,
"grad_norm": 0.38643041253089905,
"learning_rate": 2.2434860475954316e-05,
"loss": 0.0033,
"step": 3123
},
{
"epoch": 5.94,
"grad_norm": 0.7714603543281555,
"learning_rate": 2.241698618861033e-05,
"loss": 0.0201,
"step": 3124
},
{
"epoch": 5.94,
"grad_norm": 0.19342990219593048,
"learning_rate": 2.2399114776417432e-05,
"loss": 0.0021,
"step": 3125
},
{
"epoch": 5.94,
"grad_norm": 0.5309851765632629,
"learning_rate": 2.2381246246151675e-05,
"loss": 0.0065,
"step": 3126
},
{
"epoch": 5.94,
"grad_norm": 0.4924316704273224,
"learning_rate": 2.236338060458804e-05,
"loss": 0.0049,
"step": 3127
},
{
"epoch": 5.95,
"grad_norm": 0.4532407820224762,
"learning_rate": 2.23455178585004e-05,
"loss": 0.0096,
"step": 3128
},
{
"epoch": 5.95,
"grad_norm": 0.593385636806488,
"learning_rate": 2.2327658014661552e-05,
"loss": 0.0121,
"step": 3129
},
{
"epoch": 5.95,
"grad_norm": 0.2774548828601837,
"learning_rate": 2.230980107984315e-05,
"loss": 0.0035,
"step": 3130
},
{
"epoch": 5.95,
"grad_norm": 0.6404510736465454,
"learning_rate": 2.2291947060815784e-05,
"loss": 0.0101,
"step": 3131
},
{
"epoch": 5.95,
"grad_norm": 0.4716700613498688,
"learning_rate": 2.2274095964348945e-05,
"loss": 0.0054,
"step": 3132
},
{
"epoch": 5.96,
"grad_norm": 0.47315603494644165,
"learning_rate": 2.2256247797210965e-05,
"loss": 0.0096,
"step": 3133
},
{
"epoch": 5.96,
"grad_norm": 0.541944682598114,
"learning_rate": 2.2238402566169117e-05,
"loss": 0.0234,
"step": 3134
},
{
"epoch": 5.96,
"grad_norm": 0.6196478605270386,
"learning_rate": 2.2220560277989548e-05,
"loss": 0.0128,
"step": 3135
},
{
"epoch": 5.96,
"grad_norm": 0.3430854380130768,
"learning_rate": 2.220272093943726e-05,
"loss": 0.0024,
"step": 3136
},
{
"epoch": 5.96,
"grad_norm": 0.52972412109375,
"learning_rate": 2.2184884557276174e-05,
"loss": 0.0054,
"step": 3137
},
{
"epoch": 5.97,
"grad_norm": 0.30342450737953186,
"learning_rate": 2.216705113826908e-05,
"loss": 0.0032,
"step": 3138
},
{
"epoch": 5.97,
"grad_norm": 0.2564369738101959,
"learning_rate": 2.2149220689177627e-05,
"loss": 0.0031,
"step": 3139
},
{
"epoch": 5.97,
"grad_norm": 0.4639342129230499,
"learning_rate": 2.213139321676236e-05,
"loss": 0.0158,
"step": 3140
},
{
"epoch": 5.97,
"grad_norm": 0.7798018455505371,
"learning_rate": 2.211356872778268e-05,
"loss": 0.0068,
"step": 3141
},
{
"epoch": 5.97,
"grad_norm": 0.40375497937202454,
"learning_rate": 2.2095747228996856e-05,
"loss": 0.0041,
"step": 3142
},
{
"epoch": 5.98,
"grad_norm": 0.2676216959953308,
"learning_rate": 2.207792872716205e-05,
"loss": 0.0025,
"step": 3143
},
{
"epoch": 5.98,
"grad_norm": 0.44410884380340576,
"learning_rate": 2.206011322903425e-05,
"loss": 0.0028,
"step": 3144
},
{
"epoch": 5.98,
"grad_norm": 0.6605573296546936,
"learning_rate": 2.2042300741368325e-05,
"loss": 0.0158,
"step": 3145
},
{
"epoch": 5.98,
"grad_norm": 0.699411928653717,
"learning_rate": 2.202449127091802e-05,
"loss": 0.0073,
"step": 3146
},
{
"epoch": 5.98,
"grad_norm": 0.44974446296691895,
"learning_rate": 2.2006684824435884e-05,
"loss": 0.0047,
"step": 3147
},
{
"epoch": 5.98,
"grad_norm": 0.6488699316978455,
"learning_rate": 2.1988881408673377e-05,
"loss": 0.0112,
"step": 3148
},
{
"epoch": 5.99,
"grad_norm": 0.4221762418746948,
"learning_rate": 2.1971081030380776e-05,
"loss": 0.0041,
"step": 3149
},
{
"epoch": 5.99,
"grad_norm": 0.6200349926948547,
"learning_rate": 2.195328369630722e-05,
"loss": 0.0109,
"step": 3150
},
{
"epoch": 5.99,
"grad_norm": 0.6280919313430786,
"learning_rate": 2.1935489413200676e-05,
"loss": 0.0206,
"step": 3151
},
{
"epoch": 5.99,
"grad_norm": 0.613147497177124,
"learning_rate": 2.1917698187807987e-05,
"loss": 0.0079,
"step": 3152
},
{
"epoch": 5.99,
"grad_norm": 0.26535120606422424,
"learning_rate": 2.189991002687479e-05,
"loss": 0.0048,
"step": 3153
},
{
"epoch": 6.0,
"grad_norm": 0.5430494546890259,
"learning_rate": 2.188212493714561e-05,
"loss": 0.0113,
"step": 3154
},
{
"epoch": 6.0,
"grad_norm": 0.383083701133728,
"learning_rate": 2.1864342925363784e-05,
"loss": 0.0086,
"step": 3155
},
{
"epoch": 6.0,
"grad_norm": 0.6285627484321594,
"learning_rate": 2.184656399827146e-05,
"loss": 0.0036,
"step": 3156
},
{
"epoch": 6.0,
"grad_norm": 0.29397639632225037,
"learning_rate": 2.1828788162609656e-05,
"loss": 0.0044,
"step": 3157
},
{
"epoch": 6.0,
"grad_norm": 0.7248390913009644,
"learning_rate": 2.181101542511821e-05,
"loss": 0.0027,
"step": 3158
},
{
"epoch": 6.01,
"grad_norm": 0.19152431190013885,
"learning_rate": 2.179324579253575e-05,
"loss": 0.0021,
"step": 3159
},
{
"epoch": 6.01,
"grad_norm": 0.603239893913269,
"learning_rate": 2.1775479271599774e-05,
"loss": 0.0053,
"step": 3160
},
{
"epoch": 6.01,
"grad_norm": 0.42730075120925903,
"learning_rate": 2.175771586904657e-05,
"loss": 0.0041,
"step": 3161
},
{
"epoch": 6.01,
"grad_norm": 0.1671006828546524,
"learning_rate": 2.173995559161126e-05,
"loss": 0.0034,
"step": 3162
},
{
"epoch": 6.01,
"grad_norm": 0.2735035717487335,
"learning_rate": 2.1722198446027758e-05,
"loss": 0.0028,
"step": 3163
},
{
"epoch": 6.02,
"grad_norm": 0.09199666231870651,
"learning_rate": 2.170444443902882e-05,
"loss": 0.0008,
"step": 3164
},
{
"epoch": 6.02,
"grad_norm": 0.13025526702404022,
"learning_rate": 2.1686693577345986e-05,
"loss": 0.0025,
"step": 3165
},
{
"epoch": 6.02,
"grad_norm": 0.4617486596107483,
"learning_rate": 2.1668945867709636e-05,
"loss": 0.0028,
"step": 3166
},
{
"epoch": 6.02,
"grad_norm": 0.2593457102775574,
"learning_rate": 2.165120131684891e-05,
"loss": 0.0032,
"step": 3167
},
{
"epoch": 6.02,
"grad_norm": 0.15547211468219757,
"learning_rate": 2.163345993149178e-05,
"loss": 0.003,
"step": 3168
},
{
"epoch": 6.02,
"grad_norm": 0.21250589191913605,
"learning_rate": 2.161572171836504e-05,
"loss": 0.0015,
"step": 3169
},
{
"epoch": 6.03,
"grad_norm": 0.5834636688232422,
"learning_rate": 2.1597986684194215e-05,
"loss": 0.0135,
"step": 3170
},
{
"epoch": 6.03,
"grad_norm": 0.2283647507429123,
"learning_rate": 2.1580254835703687e-05,
"loss": 0.0029,
"step": 3171
},
{
"epoch": 6.03,
"grad_norm": 0.25487494468688965,
"learning_rate": 2.1562526179616603e-05,
"loss": 0.0019,
"step": 3172
},
{
"epoch": 6.03,
"grad_norm": 0.3889104127883911,
"learning_rate": 2.1544800722654894e-05,
"loss": 0.0061,
"step": 3173
},
{
"epoch": 6.03,
"grad_norm": 0.48635241389274597,
"learning_rate": 2.152707847153929e-05,
"loss": 0.0033,
"step": 3174
},
{
"epoch": 6.04,
"grad_norm": 0.4855099022388458,
"learning_rate": 2.1509359432989323e-05,
"loss": 0.0088,
"step": 3175
},
{
"epoch": 6.04,
"grad_norm": 0.4310423731803894,
"learning_rate": 2.1491643613723248e-05,
"loss": 0.0016,
"step": 3176
},
{
"epoch": 6.04,
"grad_norm": 0.07505448162555695,
"learning_rate": 2.147393102045817e-05,
"loss": 0.0009,
"step": 3177
},
{
"epoch": 6.04,
"grad_norm": 0.42079681158065796,
"learning_rate": 2.1456221659909932e-05,
"loss": 0.0041,
"step": 3178
},
{
"epoch": 6.04,
"grad_norm": 0.41127970814704895,
"learning_rate": 2.1438515538793147e-05,
"loss": 0.0103,
"step": 3179
},
{
"epoch": 6.05,
"grad_norm": 0.1803799569606781,
"learning_rate": 2.1420812663821218e-05,
"loss": 0.0016,
"step": 3180
},
{
"epoch": 6.05,
"grad_norm": 0.5722612142562866,
"learning_rate": 2.1403113041706317e-05,
"loss": 0.0078,
"step": 3181
},
{
"epoch": 6.05,
"grad_norm": 0.4573041498661041,
"learning_rate": 2.1385416679159363e-05,
"loss": 0.0051,
"step": 3182
},
{
"epoch": 6.05,
"grad_norm": 0.24095070362091064,
"learning_rate": 2.136772358289006e-05,
"loss": 0.0018,
"step": 3183
},
{
"epoch": 6.05,
"grad_norm": 0.3806236684322357,
"learning_rate": 2.1350033759606863e-05,
"loss": 0.0026,
"step": 3184
},
{
"epoch": 6.06,
"grad_norm": 0.5477047562599182,
"learning_rate": 2.1332347216016994e-05,
"loss": 0.0043,
"step": 3185
},
{
"epoch": 6.06,
"grad_norm": 0.24739281833171844,
"learning_rate": 2.131466395882641e-05,
"loss": 0.0052,
"step": 3186
},
{
"epoch": 6.06,
"grad_norm": 0.3290814459323883,
"learning_rate": 2.129698399473986e-05,
"loss": 0.0065,
"step": 3187
},
{
"epoch": 6.06,
"grad_norm": 0.274578332901001,
"learning_rate": 2.1279307330460794e-05,
"loss": 0.0035,
"step": 3188
},
{
"epoch": 6.06,
"grad_norm": 0.35903090238571167,
"learning_rate": 2.1261633972691474e-05,
"loss": 0.0053,
"step": 3189
},
{
"epoch": 6.06,
"grad_norm": 0.5447088479995728,
"learning_rate": 2.1243963928132843e-05,
"loss": 0.008,
"step": 3190
},
{
"epoch": 6.07,
"grad_norm": 0.7220551371574402,
"learning_rate": 2.1226297203484624e-05,
"loss": 0.0106,
"step": 3191
},
{
"epoch": 6.07,
"grad_norm": 0.4653252065181732,
"learning_rate": 2.1208633805445296e-05,
"loss": 0.0063,
"step": 3192
},
{
"epoch": 6.07,
"grad_norm": 0.7600398659706116,
"learning_rate": 2.1190973740712034e-05,
"loss": 0.0097,
"step": 3193
},
{
"epoch": 6.07,
"grad_norm": 0.13786649703979492,
"learning_rate": 2.117331701598078e-05,
"loss": 0.0014,
"step": 3194
},
{
"epoch": 6.07,
"grad_norm": 0.26394274830818176,
"learning_rate": 2.11556636379462e-05,
"loss": 0.003,
"step": 3195
},
{
"epoch": 6.08,
"grad_norm": 0.655761182308197,
"learning_rate": 2.113801361330169e-05,
"loss": 0.0034,
"step": 3196
},
{
"epoch": 6.08,
"grad_norm": 0.07911576330661774,
"learning_rate": 2.1120366948739374e-05,
"loss": 0.0006,
"step": 3197
},
{
"epoch": 6.08,
"grad_norm": 0.6608574390411377,
"learning_rate": 2.110272365095012e-05,
"loss": 0.0054,
"step": 3198
},
{
"epoch": 6.08,
"grad_norm": 0.18311186134815216,
"learning_rate": 2.1085083726623476e-05,
"loss": 0.0012,
"step": 3199
},
{
"epoch": 6.08,
"grad_norm": 0.5217009782791138,
"learning_rate": 2.1067447182447756e-05,
"loss": 0.0066,
"step": 3200
},
{
"epoch": 6.08,
"eval_blimp_filtered_avg": 0.7177611940298507,
"eval_blimp_filtered_std": 0.004970803188977296,
"step": 3200
},
{
"epoch": 6.08,
"eval_blimp_supplement_avg": 0.7737068965517241,
"eval_blimp_supplement_std": 0.018186218886732296,
"step": 3200
},
{
"epoch": 6.08,
"eval_vqa_filtered_avg": 0.3,
"eval_vqa_filtered_std": 0.046056618647183814,
"step": 3200
},
{
"epoch": 6.08,
"eval_winoground_filtered_avg": 0.54,
"eval_winoground_filtered_std": 0.05009082659620333,
"step": 3200
},
{
"epoch": 6.09,
"grad_norm": 0.6344010829925537,
"learning_rate": 2.104981402510999e-05,
"loss": 0.0039,
"step": 3201
},
{
"epoch": 6.09,
"grad_norm": 0.5448365807533264,
"learning_rate": 2.103218426129588e-05,
"loss": 0.003,
"step": 3202
},
{
"epoch": 6.09,
"grad_norm": 0.701843798160553,
"learning_rate": 2.1014557897689884e-05,
"loss": 0.0091,
"step": 3203
},
{
"epoch": 6.09,
"grad_norm": 0.23374047875404358,
"learning_rate": 2.099693494097517e-05,
"loss": 0.0025,
"step": 3204
},
{
"epoch": 6.09,
"grad_norm": 0.12455445528030396,
"learning_rate": 2.0979315397833576e-05,
"loss": 0.0013,
"step": 3205
},
{
"epoch": 6.1,
"grad_norm": 0.29580822587013245,
"learning_rate": 2.096169927494569e-05,
"loss": 0.0098,
"step": 3206
},
{
"epoch": 6.1,
"grad_norm": 0.305984765291214,
"learning_rate": 2.0944086578990774e-05,
"loss": 0.0019,
"step": 3207
},
{
"epoch": 6.1,
"grad_norm": 0.4234779179096222,
"learning_rate": 2.092647731664681e-05,
"loss": 0.0046,
"step": 3208
},
{
"epoch": 6.1,
"grad_norm": 0.23357465863227844,
"learning_rate": 2.0908871494590456e-05,
"loss": 0.0041,
"step": 3209
},
{
"epoch": 6.1,
"grad_norm": 0.22897683084011078,
"learning_rate": 2.0891269119497097e-05,
"loss": 0.0029,
"step": 3210
},
{
"epoch": 6.1,
"grad_norm": 0.3738779127597809,
"learning_rate": 2.0873670198040773e-05,
"loss": 0.012,
"step": 3211
},
{
"epoch": 6.11,
"grad_norm": 0.4216354787349701,
"learning_rate": 2.085607473689426e-05,
"loss": 0.0086,
"step": 3212
},
{
"epoch": 6.11,
"grad_norm": 0.0873207151889801,
"learning_rate": 2.0838482742728962e-05,
"loss": 0.0008,
"step": 3213
},
{
"epoch": 6.11,
"grad_norm": 0.8321446180343628,
"learning_rate": 2.0820894222215024e-05,
"loss": 0.0034,
"step": 3214
},
{
"epoch": 6.11,
"grad_norm": 0.23226317763328552,
"learning_rate": 2.0803309182021264e-05,
"loss": 0.0049,
"step": 3215
},
{
"epoch": 6.11,
"grad_norm": 0.18355317413806915,
"learning_rate": 2.078572762881514e-05,
"loss": 0.0017,
"step": 3216
},
{
"epoch": 6.12,
"grad_norm": 0.24743328988552094,
"learning_rate": 2.076814956926284e-05,
"loss": 0.0047,
"step": 3217
},
{
"epoch": 6.12,
"grad_norm": 0.24436771869659424,
"learning_rate": 2.0750575010029198e-05,
"loss": 0.0041,
"step": 3218
},
{
"epoch": 6.12,
"grad_norm": 0.274209588766098,
"learning_rate": 2.0733003957777722e-05,
"loss": 0.0043,
"step": 3219
},
{
"epoch": 6.12,
"grad_norm": 0.22842419147491455,
"learning_rate": 2.0715436419170595e-05,
"loss": 0.0023,
"step": 3220
},
{
"epoch": 6.12,
"grad_norm": 0.3472060263156891,
"learning_rate": 2.0697872400868683e-05,
"loss": 0.0027,
"step": 3221
},
{
"epoch": 6.13,
"grad_norm": 0.5275905728340149,
"learning_rate": 2.0680311909531486e-05,
"loss": 0.0053,
"step": 3222
},
{
"epoch": 6.13,
"grad_norm": 0.5880752205848694,
"learning_rate": 2.0662754951817186e-05,
"loss": 0.0124,
"step": 3223
},
{
"epoch": 6.13,
"grad_norm": 0.5676242113113403,
"learning_rate": 2.064520153438264e-05,
"loss": 0.0051,
"step": 3224
},
{
"epoch": 6.13,
"grad_norm": 0.4593947231769562,
"learning_rate": 2.0627651663883314e-05,
"loss": 0.0044,
"step": 3225
},
{
"epoch": 6.13,
"grad_norm": 0.12419800460338593,
"learning_rate": 2.0610105346973387e-05,
"loss": 0.0018,
"step": 3226
},
{
"epoch": 6.13,
"grad_norm": 0.4667418301105499,
"learning_rate": 2.059256259030566e-05,
"loss": 0.0106,
"step": 3227
},
{
"epoch": 6.14,
"grad_norm": 0.3733164966106415,
"learning_rate": 2.0575023400531572e-05,
"loss": 0.0015,
"step": 3228
},
{
"epoch": 6.14,
"grad_norm": 0.28068041801452637,
"learning_rate": 2.0557487784301245e-05,
"loss": 0.0017,
"step": 3229
},
{
"epoch": 6.14,
"grad_norm": 0.3788168132305145,
"learning_rate": 2.0539955748263414e-05,
"loss": 0.0077,
"step": 3230
},
{
"epoch": 6.14,
"grad_norm": 0.3907839357852936,
"learning_rate": 2.0522427299065476e-05,
"loss": 0.0024,
"step": 3231
},
{
"epoch": 6.14,
"grad_norm": 0.6124275326728821,
"learning_rate": 2.0504902443353453e-05,
"loss": 0.0054,
"step": 3232
},
{
"epoch": 6.15,
"grad_norm": 0.1440645158290863,
"learning_rate": 2.048738118777203e-05,
"loss": 0.001,
"step": 3233
},
{
"epoch": 6.15,
"grad_norm": 0.42051059007644653,
"learning_rate": 2.0469863538964487e-05,
"loss": 0.0077,
"step": 3234
},
{
"epoch": 6.15,
"grad_norm": 0.36677461862564087,
"learning_rate": 2.0452349503572767e-05,
"loss": 0.0038,
"step": 3235
},
{
"epoch": 6.15,
"grad_norm": 0.11448579281568527,
"learning_rate": 2.0434839088237455e-05,
"loss": 0.0009,
"step": 3236
},
{
"epoch": 6.15,
"grad_norm": 0.49695947766304016,
"learning_rate": 2.041733229959771e-05,
"loss": 0.0031,
"step": 3237
},
{
"epoch": 6.16,
"grad_norm": 0.2697499692440033,
"learning_rate": 2.0399829144291385e-05,
"loss": 0.003,
"step": 3238
},
{
"epoch": 6.16,
"grad_norm": 0.5743992924690247,
"learning_rate": 2.0382329628954888e-05,
"loss": 0.0021,
"step": 3239
},
{
"epoch": 6.16,
"grad_norm": 0.49111178517341614,
"learning_rate": 2.0364833760223297e-05,
"loss": 0.0078,
"step": 3240
},
{
"epoch": 6.16,
"grad_norm": 0.5234203934669495,
"learning_rate": 2.034734154473029e-05,
"loss": 0.0028,
"step": 3241
},
{
"epoch": 6.16,
"grad_norm": 0.33945608139038086,
"learning_rate": 2.032985298910815e-05,
"loss": 0.004,
"step": 3242
},
{
"epoch": 6.17,
"grad_norm": 0.24588191509246826,
"learning_rate": 2.031236809998778e-05,
"loss": 0.0017,
"step": 3243
},
{
"epoch": 6.17,
"grad_norm": 0.23402190208435059,
"learning_rate": 2.029488688399871e-05,
"loss": 0.0024,
"step": 3244
},
{
"epoch": 6.17,
"grad_norm": 0.19059251248836517,
"learning_rate": 2.0277409347769043e-05,
"loss": 0.0022,
"step": 3245
},
{
"epoch": 6.17,
"grad_norm": 0.7030087113380432,
"learning_rate": 2.0259935497925514e-05,
"loss": 0.0074,
"step": 3246
},
{
"epoch": 6.17,
"grad_norm": 0.10905256867408752,
"learning_rate": 2.0242465341093462e-05,
"loss": 0.0011,
"step": 3247
},
{
"epoch": 6.17,
"grad_norm": 0.3776775896549225,
"learning_rate": 2.0224998883896788e-05,
"loss": 0.0063,
"step": 3248
},
{
"epoch": 6.18,
"grad_norm": 0.4319923222064972,
"learning_rate": 2.0207536132958034e-05,
"loss": 0.0031,
"step": 3249
},
{
"epoch": 6.18,
"grad_norm": 0.22646470367908478,
"learning_rate": 2.019007709489834e-05,
"loss": 0.0023,
"step": 3250
},
{
"epoch": 6.18,
"grad_norm": 0.21215522289276123,
"learning_rate": 2.017262177633738e-05,
"loss": 0.0025,
"step": 3251
},
{
"epoch": 6.18,
"grad_norm": 0.3644309341907501,
"learning_rate": 2.0155170183893486e-05,
"loss": 0.0043,
"step": 3252
},
{
"epoch": 6.18,
"grad_norm": 0.32146161794662476,
"learning_rate": 2.0137722324183533e-05,
"loss": 0.0018,
"step": 3253
},
{
"epoch": 6.19,
"grad_norm": 0.6544110774993896,
"learning_rate": 2.0120278203823003e-05,
"loss": 0.01,
"step": 3254
},
{
"epoch": 6.19,
"grad_norm": 0.6944030523300171,
"learning_rate": 2.0102837829425942e-05,
"loss": 0.0037,
"step": 3255
},
{
"epoch": 6.19,
"grad_norm": 0.5521427392959595,
"learning_rate": 2.0085401207605006e-05,
"loss": 0.0093,
"step": 3256
},
{
"epoch": 6.19,
"grad_norm": 0.25941741466522217,
"learning_rate": 2.0067968344971377e-05,
"loss": 0.002,
"step": 3257
},
{
"epoch": 6.19,
"grad_norm": 0.39784881472587585,
"learning_rate": 2.0050539248134867e-05,
"loss": 0.0047,
"step": 3258
},
{
"epoch": 6.2,
"grad_norm": 0.0985315591096878,
"learning_rate": 2.003311392370384e-05,
"loss": 0.0019,
"step": 3259
},
{
"epoch": 6.2,
"grad_norm": 0.6722962856292725,
"learning_rate": 2.0015692378285206e-05,
"loss": 0.0019,
"step": 3260
},
{
"epoch": 6.2,
"grad_norm": 0.12845318019390106,
"learning_rate": 1.999827461848448e-05,
"loss": 0.0015,
"step": 3261
},
{
"epoch": 6.2,
"grad_norm": 0.08231335878372192,
"learning_rate": 1.998086065090571e-05,
"loss": 0.0012,
"step": 3262
},
{
"epoch": 6.2,
"grad_norm": 0.11343070864677429,
"learning_rate": 1.996345048215153e-05,
"loss": 0.0017,
"step": 3263
},
{
"epoch": 6.21,
"grad_norm": 0.336061954498291,
"learning_rate": 1.9946044118823112e-05,
"loss": 0.0063,
"step": 3264
},
{
"epoch": 6.21,
"grad_norm": 0.545028805732727,
"learning_rate": 1.992864156752021e-05,
"loss": 0.0042,
"step": 3265
},
{
"epoch": 6.21,
"grad_norm": 0.28023281693458557,
"learning_rate": 1.991124283484111e-05,
"loss": 0.0041,
"step": 3266
},
{
"epoch": 6.21,
"grad_norm": 0.11884608119726181,
"learning_rate": 1.989384792738267e-05,
"loss": 0.0007,
"step": 3267
},
{
"epoch": 6.21,
"grad_norm": 0.31619003415107727,
"learning_rate": 1.9876456851740267e-05,
"loss": 0.0027,
"step": 3268
},
{
"epoch": 6.21,
"grad_norm": 0.35391175746917725,
"learning_rate": 1.9859069614507857e-05,
"loss": 0.0018,
"step": 3269
},
{
"epoch": 6.22,
"grad_norm": 0.47565433382987976,
"learning_rate": 1.9841686222277944e-05,
"loss": 0.0028,
"step": 3270
},
{
"epoch": 6.22,
"grad_norm": 0.854709804058075,
"learning_rate": 1.982430668164153e-05,
"loss": 0.006,
"step": 3271
},
{
"epoch": 6.22,
"grad_norm": 0.39493897557258606,
"learning_rate": 1.9806930999188207e-05,
"loss": 0.0198,
"step": 3272
},
{
"epoch": 6.22,
"grad_norm": 0.5420381426811218,
"learning_rate": 1.9789559181506078e-05,
"loss": 0.007,
"step": 3273
},
{
"epoch": 6.22,
"grad_norm": 0.3632787764072418,
"learning_rate": 1.9772191235181777e-05,
"loss": 0.0044,
"step": 3274
},
{
"epoch": 6.23,
"grad_norm": 0.2589487135410309,
"learning_rate": 1.9754827166800493e-05,
"loss": 0.0015,
"step": 3275
},
{
"epoch": 6.23,
"grad_norm": 0.18724770843982697,
"learning_rate": 1.973746698294592e-05,
"loss": 0.0035,
"step": 3276
},
{
"epoch": 6.23,
"grad_norm": 0.14987881481647491,
"learning_rate": 1.972011069020029e-05,
"loss": 0.0018,
"step": 3277
},
{
"epoch": 6.23,
"grad_norm": 0.38624802231788635,
"learning_rate": 1.9702758295144354e-05,
"loss": 0.005,
"step": 3278
},
{
"epoch": 6.23,
"grad_norm": 0.14111268520355225,
"learning_rate": 1.968540980435741e-05,
"loss": 0.0007,
"step": 3279
},
{
"epoch": 6.24,
"grad_norm": 0.10986458510160446,
"learning_rate": 1.966806522441723e-05,
"loss": 0.001,
"step": 3280
},
{
"epoch": 6.24,
"grad_norm": 0.11996032297611237,
"learning_rate": 1.965072456190014e-05,
"loss": 0.0015,
"step": 3281
},
{
"epoch": 6.24,
"grad_norm": 0.1899401992559433,
"learning_rate": 1.9633387823380987e-05,
"loss": 0.0034,
"step": 3282
},
{
"epoch": 6.24,
"grad_norm": 0.12810049951076508,
"learning_rate": 1.961605501543308e-05,
"loss": 0.002,
"step": 3283
},
{
"epoch": 6.24,
"grad_norm": 0.1065487191081047,
"learning_rate": 1.9598726144628294e-05,
"loss": 0.0009,
"step": 3284
},
{
"epoch": 6.25,
"grad_norm": 0.13630525767803192,
"learning_rate": 1.9581401217536988e-05,
"loss": 0.0025,
"step": 3285
},
{
"epoch": 6.25,
"grad_norm": 0.09486868977546692,
"learning_rate": 1.9564080240728017e-05,
"loss": 0.0007,
"step": 3286
},
{
"epoch": 6.25,
"grad_norm": 0.2978106737136841,
"learning_rate": 1.954676322076875e-05,
"loss": 0.0018,
"step": 3287
},
{
"epoch": 6.25,
"grad_norm": 0.7745591998100281,
"learning_rate": 1.9529450164225045e-05,
"loss": 0.01,
"step": 3288
},
{
"epoch": 6.25,
"grad_norm": 0.24473340809345245,
"learning_rate": 1.951214107766127e-05,
"loss": 0.0025,
"step": 3289
},
{
"epoch": 6.25,
"grad_norm": 0.48880141973495483,
"learning_rate": 1.9494835967640292e-05,
"loss": 0.0053,
"step": 3290
},
{
"epoch": 6.26,
"grad_norm": 0.24796532094478607,
"learning_rate": 1.9477534840723444e-05,
"loss": 0.0012,
"step": 3291
},
{
"epoch": 6.26,
"grad_norm": 0.42322441935539246,
"learning_rate": 1.9460237703470568e-05,
"loss": 0.0026,
"step": 3292
},
{
"epoch": 6.26,
"grad_norm": 0.38934600353240967,
"learning_rate": 1.9442944562440014e-05,
"loss": 0.0055,
"step": 3293
},
{
"epoch": 6.26,
"grad_norm": 0.2952500283718109,
"learning_rate": 1.9425655424188563e-05,
"loss": 0.0023,
"step": 3294
},
{
"epoch": 6.26,
"grad_norm": 0.2418069988489151,
"learning_rate": 1.940837029527152e-05,
"loss": 0.0014,
"step": 3295
},
{
"epoch": 6.27,
"grad_norm": 0.20799657702445984,
"learning_rate": 1.9391089182242677e-05,
"loss": 0.0023,
"step": 3296
},
{
"epoch": 6.27,
"grad_norm": 0.08170506358146667,
"learning_rate": 1.9373812091654254e-05,
"loss": 0.0011,
"step": 3297
},
{
"epoch": 6.27,
"grad_norm": 1.3647176027297974,
"learning_rate": 1.9356539030056998e-05,
"loss": 0.009,
"step": 3298
},
{
"epoch": 6.27,
"grad_norm": 0.39549294114112854,
"learning_rate": 1.9339270004000104e-05,
"loss": 0.0077,
"step": 3299
},
{
"epoch": 6.27,
"grad_norm": 0.6177412867546082,
"learning_rate": 1.9322005020031234e-05,
"loss": 0.0071,
"step": 3300
},
{
"epoch": 6.27,
"eval_blimp_filtered_avg": 0.7164179104477612,
"eval_blimp_filtered_std": 0.004977159240127459,
"step": 3300
},
{
"epoch": 6.27,
"eval_blimp_supplement_avg": 0.7737068965517241,
"eval_blimp_supplement_std": 0.017983790562922024,
"step": 3300
},
{
"epoch": 6.27,
"eval_vqa_filtered_avg": 0.32,
"eval_vqa_filtered_std": 0.046882617226215034,
"step": 3300
},
{
"epoch": 6.27,
"eval_winoground_filtered_avg": 0.54,
"eval_winoground_filtered_std": 0.05009082659620333,
"step": 3300
},
{
"epoch": 6.28,
"grad_norm": 0.30938997864723206,
"learning_rate": 1.9304744084696524e-05,
"loss": 0.0042,
"step": 3301
},
{
"epoch": 6.28,
"grad_norm": 0.2177630215883255,
"learning_rate": 1.928748720454059e-05,
"loss": 0.0027,
"step": 3302
},
{
"epoch": 6.28,
"grad_norm": 0.7774405479431152,
"learning_rate": 1.9270234386106468e-05,
"loss": 0.0103,
"step": 3303
},
{
"epoch": 6.28,
"grad_norm": 0.861967146396637,
"learning_rate": 1.92529856359357e-05,
"loss": 0.006,
"step": 3304
},
{
"epoch": 6.28,
"grad_norm": 0.7675650119781494,
"learning_rate": 1.923574096056827e-05,
"loss": 0.012,
"step": 3305
},
{
"epoch": 6.29,
"grad_norm": 0.3599194288253784,
"learning_rate": 1.9218500366542594e-05,
"loss": 0.0044,
"step": 3306
},
{
"epoch": 6.29,
"grad_norm": 0.09892160445451736,
"learning_rate": 1.9201263860395577e-05,
"loss": 0.0013,
"step": 3307
},
{
"epoch": 6.29,
"grad_norm": 0.35983601212501526,
"learning_rate": 1.9184031448662564e-05,
"loss": 0.0017,
"step": 3308
},
{
"epoch": 6.29,
"grad_norm": 0.16213543713092804,
"learning_rate": 1.916680313787732e-05,
"loss": 0.0026,
"step": 3309
},
{
"epoch": 6.29,
"grad_norm": 0.4860267639160156,
"learning_rate": 1.9149578934572085e-05,
"loss": 0.0146,
"step": 3310
},
{
"epoch": 6.29,
"grad_norm": 0.5300498008728027,
"learning_rate": 1.913235884527753e-05,
"loss": 0.0057,
"step": 3311
},
{
"epoch": 6.3,
"grad_norm": 0.41538554430007935,
"learning_rate": 1.911514287652277e-05,
"loss": 0.0062,
"step": 3312
},
{
"epoch": 6.3,
"grad_norm": 0.2275433987379074,
"learning_rate": 1.909793103483537e-05,
"loss": 0.0016,
"step": 3313
},
{
"epoch": 6.3,
"grad_norm": 0.6437553763389587,
"learning_rate": 1.9080723326741287e-05,
"loss": 0.0171,
"step": 3314
},
{
"epoch": 6.3,
"grad_norm": 0.1121813952922821,
"learning_rate": 1.9063519758764955e-05,
"loss": 0.0009,
"step": 3315
},
{
"epoch": 6.3,
"grad_norm": 0.687263011932373,
"learning_rate": 1.904632033742924e-05,
"loss": 0.0129,
"step": 3316
},
{
"epoch": 6.31,
"grad_norm": 0.16878388822078705,
"learning_rate": 1.902912506925539e-05,
"loss": 0.0009,
"step": 3317
},
{
"epoch": 6.31,
"grad_norm": 0.10072501003742218,
"learning_rate": 1.9011933960763126e-05,
"loss": 0.0011,
"step": 3318
},
{
"epoch": 6.31,
"grad_norm": 0.12781405448913574,
"learning_rate": 1.8994747018470575e-05,
"loss": 0.0015,
"step": 3319
},
{
"epoch": 6.31,
"grad_norm": 0.3725898861885071,
"learning_rate": 1.897756424889427e-05,
"loss": 0.0039,
"step": 3320
},
{
"epoch": 6.31,
"grad_norm": 0.36559370160102844,
"learning_rate": 1.896038565854918e-05,
"loss": 0.0025,
"step": 3321
},
{
"epoch": 6.32,
"grad_norm": 0.3733101487159729,
"learning_rate": 1.8943211253948692e-05,
"loss": 0.0044,
"step": 3322
},
{
"epoch": 6.32,
"grad_norm": 0.6218533515930176,
"learning_rate": 1.8926041041604588e-05,
"loss": 0.0021,
"step": 3323
},
{
"epoch": 6.32,
"grad_norm": 0.5452737808227539,
"learning_rate": 1.8908875028027066e-05,
"loss": 0.0067,
"step": 3324
},
{
"epoch": 6.32,
"grad_norm": 0.5408026576042175,
"learning_rate": 1.8891713219724757e-05,
"loss": 0.0051,
"step": 3325
},
{
"epoch": 6.32,
"grad_norm": 0.213319331407547,
"learning_rate": 1.887455562320465e-05,
"loss": 0.0027,
"step": 3326
},
{
"epoch": 6.33,
"grad_norm": 0.2139444500207901,
"learning_rate": 1.8857402244972183e-05,
"loss": 0.0013,
"step": 3327
},
{
"epoch": 6.33,
"grad_norm": 0.11615992337465286,
"learning_rate": 1.8840253091531178e-05,
"loss": 0.0011,
"step": 3328
},
{
"epoch": 6.33,
"grad_norm": 0.16389843821525574,
"learning_rate": 1.8823108169383836e-05,
"loss": 0.002,
"step": 3329
},
{
"epoch": 6.33,
"grad_norm": 0.3731285631656647,
"learning_rate": 1.8805967485030773e-05,
"loss": 0.0031,
"step": 3330
},
{
"epoch": 6.33,
"grad_norm": 0.508953332901001,
"learning_rate": 1.878883104497102e-05,
"loss": 0.0112,
"step": 3331
},
{
"epoch": 6.33,
"grad_norm": 0.22628939151763916,
"learning_rate": 1.8771698855701946e-05,
"loss": 0.003,
"step": 3332
},
{
"epoch": 6.34,
"grad_norm": 0.21960903704166412,
"learning_rate": 1.8754570923719346e-05,
"loss": 0.0015,
"step": 3333
},
{
"epoch": 6.34,
"grad_norm": 0.08086056262254715,
"learning_rate": 1.8737447255517402e-05,
"loss": 0.0005,
"step": 3334
},
{
"epoch": 6.34,
"grad_norm": 0.2818944752216339,
"learning_rate": 1.8720327857588646e-05,
"loss": 0.0024,
"step": 3335
},
{
"epoch": 6.34,
"grad_norm": 0.4846373498439789,
"learning_rate": 1.8703212736424038e-05,
"loss": 0.0034,
"step": 3336
},
{
"epoch": 6.34,
"grad_norm": 0.17349940538406372,
"learning_rate": 1.8686101898512872e-05,
"loss": 0.0021,
"step": 3337
},
{
"epoch": 6.35,
"grad_norm": 0.5364410281181335,
"learning_rate": 1.8668995350342846e-05,
"loss": 0.006,
"step": 3338
},
{
"epoch": 6.35,
"grad_norm": 0.6806755661964417,
"learning_rate": 1.865189309840004e-05,
"loss": 0.0103,
"step": 3339
},
{
"epoch": 6.35,
"grad_norm": 0.22201111912727356,
"learning_rate": 1.863479514916886e-05,
"loss": 0.0021,
"step": 3340
},
{
"epoch": 6.35,
"grad_norm": 0.5623649954795837,
"learning_rate": 1.8617701509132124e-05,
"loss": 0.0094,
"step": 3341
},
{
"epoch": 6.35,
"grad_norm": 0.5879457592964172,
"learning_rate": 1.860061218477101e-05,
"loss": 0.0051,
"step": 3342
},
{
"epoch": 6.36,
"grad_norm": 0.5466235280036926,
"learning_rate": 1.8583527182565033e-05,
"loss": 0.0086,
"step": 3343
},
{
"epoch": 6.36,
"grad_norm": 0.23046714067459106,
"learning_rate": 1.85664465089921e-05,
"loss": 0.0032,
"step": 3344
},
{
"epoch": 6.36,
"grad_norm": 0.4773412346839905,
"learning_rate": 1.8549370170528458e-05,
"loss": 0.0045,
"step": 3345
},
{
"epoch": 6.36,
"grad_norm": 0.7632269263267517,
"learning_rate": 1.8532298173648724e-05,
"loss": 0.0094,
"step": 3346
},
{
"epoch": 6.36,
"grad_norm": 0.514247715473175,
"learning_rate": 1.8515230524825846e-05,
"loss": 0.0028,
"step": 3347
},
{
"epoch": 6.37,
"grad_norm": 0.41898754239082336,
"learning_rate": 1.8498167230531165e-05,
"loss": 0.0059,
"step": 3348
},
{
"epoch": 6.37,
"grad_norm": 0.24857832491397858,
"learning_rate": 1.8481108297234314e-05,
"loss": 0.0037,
"step": 3349
},
{
"epoch": 6.37,
"grad_norm": 0.3656007945537567,
"learning_rate": 1.8464053731403317e-05,
"loss": 0.001,
"step": 3350
},
{
"epoch": 6.37,
"grad_norm": 0.5132533311843872,
"learning_rate": 1.8447003539504544e-05,
"loss": 0.0036,
"step": 3351
},
{
"epoch": 6.37,
"grad_norm": 0.3220248520374298,
"learning_rate": 1.842995772800266e-05,
"loss": 0.0041,
"step": 3352
},
{
"epoch": 6.37,
"grad_norm": 0.8172900080680847,
"learning_rate": 1.8412916303360718e-05,
"loss": 0.005,
"step": 3353
},
{
"epoch": 6.38,
"grad_norm": 0.6418620944023132,
"learning_rate": 1.8395879272040098e-05,
"loss": 0.013,
"step": 3354
},
{
"epoch": 6.38,
"grad_norm": 0.10984694957733154,
"learning_rate": 1.8378846640500487e-05,
"loss": 0.0021,
"step": 3355
},
{
"epoch": 6.38,
"grad_norm": 0.0737818032503128,
"learning_rate": 1.836181841519993e-05,
"loss": 0.0007,
"step": 3356
},
{
"epoch": 6.38,
"grad_norm": 0.08951135724782944,
"learning_rate": 1.8344794602594805e-05,
"loss": 0.0005,
"step": 3357
},
{
"epoch": 6.38,
"grad_norm": 0.2723471224308014,
"learning_rate": 1.832777520913978e-05,
"loss": 0.0026,
"step": 3358
},
{
"epoch": 6.39,
"grad_norm": 0.1450112909078598,
"learning_rate": 1.8310760241287894e-05,
"loss": 0.0008,
"step": 3359
},
{
"epoch": 6.39,
"grad_norm": 0.8828959465026855,
"learning_rate": 1.8293749705490494e-05,
"loss": 0.0163,
"step": 3360
},
{
"epoch": 6.39,
"grad_norm": 0.04302699863910675,
"learning_rate": 1.8276743608197217e-05,
"loss": 0.0005,
"step": 3361
},
{
"epoch": 6.39,
"grad_norm": 0.21324734389781952,
"learning_rate": 1.825974195585606e-05,
"loss": 0.0024,
"step": 3362
},
{
"epoch": 6.39,
"grad_norm": 0.1333737075328827,
"learning_rate": 1.8242744754913297e-05,
"loss": 0.0033,
"step": 3363
},
{
"epoch": 6.4,
"grad_norm": 0.2180042862892151,
"learning_rate": 1.8225752011813538e-05,
"loss": 0.0081,
"step": 3364
},
{
"epoch": 6.4,
"grad_norm": 0.3381204903125763,
"learning_rate": 1.8208763732999717e-05,
"loss": 0.0037,
"step": 3365
},
{
"epoch": 6.4,
"grad_norm": 0.15292266011238098,
"learning_rate": 1.8191779924913022e-05,
"loss": 0.0016,
"step": 3366
},
{
"epoch": 6.4,
"grad_norm": 0.26734766364097595,
"learning_rate": 1.8174800593993005e-05,
"loss": 0.0032,
"step": 3367
},
{
"epoch": 6.4,
"grad_norm": 0.254456490278244,
"learning_rate": 1.8157825746677482e-05,
"loss": 0.0028,
"step": 3368
},
{
"epoch": 6.4,
"grad_norm": 0.5537685751914978,
"learning_rate": 1.8140855389402582e-05,
"loss": 0.0033,
"step": 3369
},
{
"epoch": 6.41,
"grad_norm": 0.24642276763916016,
"learning_rate": 1.8123889528602726e-05,
"loss": 0.0083,
"step": 3370
},
{
"epoch": 6.41,
"grad_norm": 0.1486857682466507,
"learning_rate": 1.8106928170710655e-05,
"loss": 0.0009,
"step": 3371
},
{
"epoch": 6.41,
"grad_norm": 0.17765386402606964,
"learning_rate": 1.808997132215736e-05,
"loss": 0.0017,
"step": 3372
},
{
"epoch": 6.41,
"grad_norm": 0.2948109805583954,
"learning_rate": 1.807301898937215e-05,
"loss": 0.0037,
"step": 3373
},
{
"epoch": 6.41,
"grad_norm": 0.412345826625824,
"learning_rate": 1.805607117878264e-05,
"loss": 0.0024,
"step": 3374
},
{
"epoch": 6.42,
"grad_norm": 0.6901482343673706,
"learning_rate": 1.8039127896814668e-05,
"loss": 0.006,
"step": 3375
},
{
"epoch": 6.42,
"grad_norm": 0.17434167861938477,
"learning_rate": 1.802218914989242e-05,
"loss": 0.0024,
"step": 3376
},
{
"epoch": 6.42,
"grad_norm": 0.5307456851005554,
"learning_rate": 1.8005254944438344e-05,
"loss": 0.0046,
"step": 3377
},
{
"epoch": 6.42,
"grad_norm": 0.42831435799598694,
"learning_rate": 1.798832528687314e-05,
"loss": 0.003,
"step": 3378
},
{
"epoch": 6.42,
"grad_norm": 0.15407989919185638,
"learning_rate": 1.79714001836158e-05,
"loss": 0.0016,
"step": 3379
},
{
"epoch": 6.43,
"grad_norm": 0.3334437906742096,
"learning_rate": 1.7954479641083613e-05,
"loss": 0.0016,
"step": 3380
},
{
"epoch": 6.43,
"grad_norm": 0.18225297331809998,
"learning_rate": 1.7937563665692095e-05,
"loss": 0.0029,
"step": 3381
},
{
"epoch": 6.43,
"grad_norm": 0.8456035256385803,
"learning_rate": 1.792065226385506e-05,
"loss": 0.0047,
"step": 3382
},
{
"epoch": 6.43,
"grad_norm": 0.20761826634407043,
"learning_rate": 1.79037454419846e-05,
"loss": 0.0023,
"step": 3383
},
{
"epoch": 6.43,
"grad_norm": 0.3336322009563446,
"learning_rate": 1.788684320649102e-05,
"loss": 0.0031,
"step": 3384
},
{
"epoch": 6.44,
"grad_norm": 0.15063083171844482,
"learning_rate": 1.7869945563782937e-05,
"loss": 0.0026,
"step": 3385
},
{
"epoch": 6.44,
"grad_norm": 0.3705213963985443,
"learning_rate": 1.7853052520267195e-05,
"loss": 0.0123,
"step": 3386
},
{
"epoch": 6.44,
"grad_norm": 0.20011015236377716,
"learning_rate": 1.7836164082348908e-05,
"loss": 0.0015,
"step": 3387
},
{
"epoch": 6.44,
"grad_norm": 0.043872084468603134,
"learning_rate": 1.7819280256431465e-05,
"loss": 0.0005,
"step": 3388
},
{
"epoch": 6.44,
"grad_norm": 1.0054519176483154,
"learning_rate": 1.7802401048916448e-05,
"loss": 0.0031,
"step": 3389
},
{
"epoch": 6.44,
"grad_norm": 0.05100753530859947,
"learning_rate": 1.778552646620374e-05,
"loss": 0.0007,
"step": 3390
},
{
"epoch": 6.45,
"grad_norm": 0.3258419632911682,
"learning_rate": 1.776865651469146e-05,
"loss": 0.0067,
"step": 3391
},
{
"epoch": 6.45,
"grad_norm": 0.2779000997543335,
"learning_rate": 1.775179120077595e-05,
"loss": 0.0041,
"step": 3392
},
{
"epoch": 6.45,
"grad_norm": 0.4258268177509308,
"learning_rate": 1.7734930530851812e-05,
"loss": 0.0085,
"step": 3393
},
{
"epoch": 6.45,
"grad_norm": 0.3310245871543884,
"learning_rate": 1.7718074511311894e-05,
"loss": 0.0045,
"step": 3394
},
{
"epoch": 6.45,
"grad_norm": 0.20218446850776672,
"learning_rate": 1.7701223148547253e-05,
"loss": 0.0011,
"step": 3395
},
{
"epoch": 6.46,
"grad_norm": 0.0695488378405571,
"learning_rate": 1.7684376448947205e-05,
"loss": 0.0007,
"step": 3396
},
{
"epoch": 6.46,
"grad_norm": 0.41271236538887024,
"learning_rate": 1.7667534418899303e-05,
"loss": 0.0036,
"step": 3397
},
{
"epoch": 6.46,
"grad_norm": 0.2153390794992447,
"learning_rate": 1.7650697064789295e-05,
"loss": 0.0018,
"step": 3398
},
{
"epoch": 6.46,
"grad_norm": 0.7010805606842041,
"learning_rate": 1.7633864393001186e-05,
"loss": 0.0092,
"step": 3399
},
{
"epoch": 6.46,
"grad_norm": 0.0464182123541832,
"learning_rate": 1.7617036409917213e-05,
"loss": 0.0003,
"step": 3400
},
{
"epoch": 6.46,
"eval_blimp_filtered_avg": 0.7173134328358209,
"eval_blimp_filtered_std": 0.0049801873830755,
"step": 3400
},
{
"epoch": 6.46,
"eval_blimp_supplement_avg": 0.7780172413793104,
"eval_blimp_supplement_std": 0.01792887809713568,
"step": 3400
},
{
"epoch": 6.46,
"eval_vqa_filtered_avg": 0.34,
"eval_vqa_filtered_std": 0.04760952285695235,
"step": 3400
},
{
"epoch": 6.46,
"eval_winoground_filtered_avg": 0.5,
"eval_winoground_filtered_std": 0.050251890762960605,
"step": 3400
},
{
"epoch": 6.47,
"grad_norm": 0.1838081330060959,
"learning_rate": 1.7600213121917803e-05,
"loss": 0.0039,
"step": 3401
},
{
"epoch": 6.47,
"grad_norm": 0.10559318959712982,
"learning_rate": 1.7583394535381613e-05,
"loss": 0.0008,
"step": 3402
},
{
"epoch": 6.47,
"grad_norm": 0.1280212253332138,
"learning_rate": 1.7566580656685553e-05,
"loss": 0.0013,
"step": 3403
},
{
"epoch": 6.47,
"grad_norm": 0.12722331285476685,
"learning_rate": 1.7549771492204683e-05,
"loss": 0.0013,
"step": 3404
},
{
"epoch": 6.47,
"grad_norm": 0.368865966796875,
"learning_rate": 1.753296704831233e-05,
"loss": 0.002,
"step": 3405
},
{
"epoch": 6.48,
"grad_norm": 0.3276905119419098,
"learning_rate": 1.7516167331380024e-05,
"loss": 0.0015,
"step": 3406
},
{
"epoch": 6.48,
"grad_norm": 0.525767982006073,
"learning_rate": 1.7499372347777475e-05,
"loss": 0.0047,
"step": 3407
},
{
"epoch": 6.48,
"grad_norm": 0.36628180742263794,
"learning_rate": 1.7482582103872608e-05,
"loss": 0.0024,
"step": 3408
},
{
"epoch": 6.48,
"grad_norm": 0.198746919631958,
"learning_rate": 1.746579660603159e-05,
"loss": 0.0018,
"step": 3409
},
{
"epoch": 6.48,
"grad_norm": 0.4113011062145233,
"learning_rate": 1.7449015860618715e-05,
"loss": 0.0037,
"step": 3410
},
{
"epoch": 6.48,
"grad_norm": 0.4734998643398285,
"learning_rate": 1.743223987399656e-05,
"loss": 0.0022,
"step": 3411
},
{
"epoch": 6.49,
"grad_norm": 0.43428391218185425,
"learning_rate": 1.7415468652525802e-05,
"loss": 0.0088,
"step": 3412
},
{
"epoch": 6.49,
"grad_norm": 0.4206155240535736,
"learning_rate": 1.7398702202565407e-05,
"loss": 0.0051,
"step": 3413
},
{
"epoch": 6.49,
"grad_norm": 0.14306442439556122,
"learning_rate": 1.7381940530472476e-05,
"loss": 0.0014,
"step": 3414
},
{
"epoch": 6.49,
"grad_norm": 0.9414021372795105,
"learning_rate": 1.7365183642602302e-05,
"loss": 0.0048,
"step": 3415
},
{
"epoch": 6.49,
"grad_norm": 0.6979029178619385,
"learning_rate": 1.734843154530837e-05,
"loss": 0.0103,
"step": 3416
},
{
"epoch": 6.5,
"grad_norm": 0.6270625591278076,
"learning_rate": 1.7331684244942384e-05,
"loss": 0.0064,
"step": 3417
},
{
"epoch": 6.5,
"grad_norm": 0.4985022246837616,
"learning_rate": 1.7314941747854145e-05,
"loss": 0.0043,
"step": 3418
},
{
"epoch": 6.5,
"grad_norm": 0.30819937586784363,
"learning_rate": 1.7298204060391725e-05,
"loss": 0.0045,
"step": 3419
},
{
"epoch": 6.5,
"grad_norm": 0.5023206472396851,
"learning_rate": 1.728147118890133e-05,
"loss": 0.0037,
"step": 3420
},
{
"epoch": 6.5,
"grad_norm": 0.2249852418899536,
"learning_rate": 1.7264743139727306e-05,
"loss": 0.0028,
"step": 3421
},
{
"epoch": 6.51,
"grad_norm": 0.20597155392169952,
"learning_rate": 1.724801991921224e-05,
"loss": 0.0018,
"step": 3422
},
{
"epoch": 6.51,
"grad_norm": 0.4936918020248413,
"learning_rate": 1.7231301533696844e-05,
"loss": 0.0035,
"step": 3423
},
{
"epoch": 6.51,
"grad_norm": 0.43640658259391785,
"learning_rate": 1.721458798952001e-05,
"loss": 0.0034,
"step": 3424
},
{
"epoch": 6.51,
"grad_norm": 0.09195563942193985,
"learning_rate": 1.7197879293018776e-05,
"loss": 0.0008,
"step": 3425
},
{
"epoch": 6.51,
"grad_norm": 0.15116122364997864,
"learning_rate": 1.718117545052839e-05,
"loss": 0.0029,
"step": 3426
},
{
"epoch": 6.52,
"grad_norm": 1.0241228342056274,
"learning_rate": 1.7164476468382185e-05,
"loss": 0.0054,
"step": 3427
},
{
"epoch": 6.52,
"grad_norm": 0.25943541526794434,
"learning_rate": 1.714778235291173e-05,
"loss": 0.0032,
"step": 3428
},
{
"epoch": 6.52,
"grad_norm": 0.3694850504398346,
"learning_rate": 1.71310931104467e-05,
"loss": 0.006,
"step": 3429
},
{
"epoch": 6.52,
"grad_norm": 0.18269109725952148,
"learning_rate": 1.7114408747314925e-05,
"loss": 0.0014,
"step": 3430
},
{
"epoch": 6.52,
"grad_norm": 0.09916549175977707,
"learning_rate": 1.7097729269842403e-05,
"loss": 0.001,
"step": 3431
},
{
"epoch": 6.52,
"grad_norm": 0.34271401166915894,
"learning_rate": 1.7081054684353272e-05,
"loss": 0.0024,
"step": 3432
},
{
"epoch": 6.53,
"grad_norm": 0.4961518347263336,
"learning_rate": 1.706438499716981e-05,
"loss": 0.0127,
"step": 3433
},
{
"epoch": 6.53,
"grad_norm": 0.3890461027622223,
"learning_rate": 1.7047720214612447e-05,
"loss": 0.0021,
"step": 3434
},
{
"epoch": 6.53,
"grad_norm": 0.12355988472700119,
"learning_rate": 1.7031060342999742e-05,
"loss": 0.0013,
"step": 3435
},
{
"epoch": 6.53,
"grad_norm": 0.34317779541015625,
"learning_rate": 1.7014405388648387e-05,
"loss": 0.0029,
"step": 3436
},
{
"epoch": 6.53,
"grad_norm": 0.2528637945652008,
"learning_rate": 1.699775535787326e-05,
"loss": 0.0051,
"step": 3437
},
{
"epoch": 6.54,
"grad_norm": 0.10284775495529175,
"learning_rate": 1.6981110256987282e-05,
"loss": 0.0005,
"step": 3438
},
{
"epoch": 6.54,
"grad_norm": 0.20277611911296844,
"learning_rate": 1.6964470092301595e-05,
"loss": 0.0032,
"step": 3439
},
{
"epoch": 6.54,
"grad_norm": 0.3469488322734833,
"learning_rate": 1.6947834870125417e-05,
"loss": 0.0027,
"step": 3440
},
{
"epoch": 6.54,
"grad_norm": 0.19436410069465637,
"learning_rate": 1.6931204596766105e-05,
"loss": 0.0012,
"step": 3441
},
{
"epoch": 6.54,
"grad_norm": 0.8355365991592407,
"learning_rate": 1.691457927852914e-05,
"loss": 0.0075,
"step": 3442
},
{
"epoch": 6.55,
"grad_norm": 0.04030472785234451,
"learning_rate": 1.689795892171812e-05,
"loss": 0.0004,
"step": 3443
},
{
"epoch": 6.55,
"grad_norm": 0.7233259081840515,
"learning_rate": 1.6881343532634773e-05,
"loss": 0.0095,
"step": 3444
},
{
"epoch": 6.55,
"grad_norm": 0.42218050360679626,
"learning_rate": 1.6864733117578922e-05,
"loss": 0.0025,
"step": 3445
},
{
"epoch": 6.55,
"grad_norm": 0.4359194040298462,
"learning_rate": 1.6848127682848552e-05,
"loss": 0.006,
"step": 3446
},
{
"epoch": 6.55,
"grad_norm": 0.1852865070104599,
"learning_rate": 1.6831527234739677e-05,
"loss": 0.0009,
"step": 3447
},
{
"epoch": 6.56,
"grad_norm": 0.36179882287979126,
"learning_rate": 1.6814931779546514e-05,
"loss": 0.0021,
"step": 3448
},
{
"epoch": 6.56,
"grad_norm": 0.5460297465324402,
"learning_rate": 1.679834132356132e-05,
"loss": 0.0021,
"step": 3449
},
{
"epoch": 6.56,
"grad_norm": 0.4444520175457001,
"learning_rate": 1.6781755873074483e-05,
"loss": 0.004,
"step": 3450
},
{
"epoch": 6.56,
"grad_norm": 0.548761248588562,
"learning_rate": 1.676517543437449e-05,
"loss": 0.0073,
"step": 3451
},
{
"epoch": 6.56,
"grad_norm": 0.3507717549800873,
"learning_rate": 1.674860001374793e-05,
"loss": 0.0062,
"step": 3452
},
{
"epoch": 6.56,
"grad_norm": 0.21076738834381104,
"learning_rate": 1.6732029617479474e-05,
"loss": 0.0029,
"step": 3453
},
{
"epoch": 6.57,
"grad_norm": 0.6583250761032104,
"learning_rate": 1.67154642518519e-05,
"loss": 0.0042,
"step": 3454
},
{
"epoch": 6.57,
"grad_norm": 0.08454226702451706,
"learning_rate": 1.669890392314611e-05,
"loss": 0.0015,
"step": 3455
},
{
"epoch": 6.57,
"grad_norm": 0.38162732124328613,
"learning_rate": 1.6682348637641018e-05,
"loss": 0.0014,
"step": 3456
},
{
"epoch": 6.57,
"grad_norm": 0.2161172330379486,
"learning_rate": 1.6665798401613706e-05,
"loss": 0.0036,
"step": 3457
},
{
"epoch": 6.57,
"grad_norm": 0.2715476155281067,
"learning_rate": 1.6649253221339302e-05,
"loss": 0.0039,
"step": 3458
},
{
"epoch": 6.58,
"grad_norm": 0.3748125731945038,
"learning_rate": 1.6632713103091018e-05,
"loss": 0.0055,
"step": 3459
},
{
"epoch": 6.58,
"grad_norm": 0.3601429760456085,
"learning_rate": 1.6616178053140155e-05,
"loss": 0.0023,
"step": 3460
},
{
"epoch": 6.58,
"grad_norm": 0.12365395575761795,
"learning_rate": 1.6599648077756084e-05,
"loss": 0.001,
"step": 3461
},
{
"epoch": 6.58,
"grad_norm": 0.1983584314584732,
"learning_rate": 1.6583123183206247e-05,
"loss": 0.0012,
"step": 3462
},
{
"epoch": 6.58,
"grad_norm": 0.3980696201324463,
"learning_rate": 1.6566603375756212e-05,
"loss": 0.0024,
"step": 3463
},
{
"epoch": 6.59,
"grad_norm": 0.11084765195846558,
"learning_rate": 1.6550088661669524e-05,
"loss": 0.0005,
"step": 3464
},
{
"epoch": 6.59,
"grad_norm": 0.4835197925567627,
"learning_rate": 1.6533579047207883e-05,
"loss": 0.005,
"step": 3465
},
{
"epoch": 6.59,
"grad_norm": 0.19755221903324127,
"learning_rate": 1.651707453863102e-05,
"loss": 0.0016,
"step": 3466
},
{
"epoch": 6.59,
"grad_norm": 0.30488455295562744,
"learning_rate": 1.6500575142196704e-05,
"loss": 0.0045,
"step": 3467
},
{
"epoch": 6.59,
"grad_norm": 0.3656032383441925,
"learning_rate": 1.6484080864160816e-05,
"loss": 0.0027,
"step": 3468
},
{
"epoch": 6.6,
"grad_norm": 0.204855278134346,
"learning_rate": 1.6467591710777276e-05,
"loss": 0.0018,
"step": 3469
},
{
"epoch": 6.6,
"grad_norm": 0.5943580269813538,
"learning_rate": 1.6451107688298046e-05,
"loss": 0.0131,
"step": 3470
},
{
"epoch": 6.6,
"grad_norm": 0.28167977929115295,
"learning_rate": 1.6434628802973146e-05,
"loss": 0.0018,
"step": 3471
},
{
"epoch": 6.6,
"grad_norm": 0.22183337807655334,
"learning_rate": 1.6418155061050694e-05,
"loss": 0.0023,
"step": 3472
},
{
"epoch": 6.6,
"grad_norm": 0.06578106433153152,
"learning_rate": 1.6401686468776775e-05,
"loss": 0.0009,
"step": 3473
},
{
"epoch": 6.6,
"grad_norm": 0.10122206062078476,
"learning_rate": 1.6385223032395598e-05,
"loss": 0.001,
"step": 3474
},
{
"epoch": 6.61,
"grad_norm": 0.11490953713655472,
"learning_rate": 1.6368764758149385e-05,
"loss": 0.0009,
"step": 3475
},
{
"epoch": 6.61,
"grad_norm": 0.3950178325176239,
"learning_rate": 1.6352311652278388e-05,
"loss": 0.0044,
"step": 3476
},
{
"epoch": 6.61,
"grad_norm": 0.16902953386306763,
"learning_rate": 1.6335863721020926e-05,
"loss": 0.0017,
"step": 3477
},
{
"epoch": 6.61,
"grad_norm": 0.4501452147960663,
"learning_rate": 1.6319420970613336e-05,
"loss": 0.0106,
"step": 3478
},
{
"epoch": 6.61,
"grad_norm": 0.4471287727355957,
"learning_rate": 1.6302983407290006e-05,
"loss": 0.0049,
"step": 3479
},
{
"epoch": 6.62,
"grad_norm": 0.3136860430240631,
"learning_rate": 1.628655103728332e-05,
"loss": 0.0025,
"step": 3480
},
{
"epoch": 6.62,
"grad_norm": 0.352350115776062,
"learning_rate": 1.6270123866823782e-05,
"loss": 0.0036,
"step": 3481
},
{
"epoch": 6.62,
"grad_norm": 0.13372322916984558,
"learning_rate": 1.62537019021398e-05,
"loss": 0.0032,
"step": 3482
},
{
"epoch": 6.62,
"grad_norm": 0.8312336206436157,
"learning_rate": 1.623728514945793e-05,
"loss": 0.0074,
"step": 3483
},
{
"epoch": 6.62,
"grad_norm": 0.11125674843788147,
"learning_rate": 1.6220873615002646e-05,
"loss": 0.0022,
"step": 3484
},
{
"epoch": 6.63,
"grad_norm": 0.545231282711029,
"learning_rate": 1.620446730499652e-05,
"loss": 0.0079,
"step": 3485
},
{
"epoch": 6.63,
"grad_norm": 0.10154607146978378,
"learning_rate": 1.6188066225660113e-05,
"loss": 0.0009,
"step": 3486
},
{
"epoch": 6.63,
"grad_norm": 0.33039742708206177,
"learning_rate": 1.6171670383212e-05,
"loss": 0.0045,
"step": 3487
},
{
"epoch": 6.63,
"grad_norm": 0.4560626149177551,
"learning_rate": 1.6155279783868782e-05,
"loss": 0.0067,
"step": 3488
},
{
"epoch": 6.63,
"grad_norm": 0.2209363877773285,
"learning_rate": 1.6138894433845064e-05,
"loss": 0.0043,
"step": 3489
},
{
"epoch": 6.63,
"grad_norm": 0.30562809109687805,
"learning_rate": 1.612251433935346e-05,
"loss": 0.0028,
"step": 3490
},
{
"epoch": 6.64,
"grad_norm": 0.49249887466430664,
"learning_rate": 1.6106139506604573e-05,
"loss": 0.0027,
"step": 3491
},
{
"epoch": 6.64,
"grad_norm": 0.2484356015920639,
"learning_rate": 1.6089769941807086e-05,
"loss": 0.0011,
"step": 3492
},
{
"epoch": 6.64,
"grad_norm": 0.20757447183132172,
"learning_rate": 1.607340565116757e-05,
"loss": 0.0034,
"step": 3493
},
{
"epoch": 6.64,
"grad_norm": 0.4600831866264343,
"learning_rate": 1.60570466408907e-05,
"loss": 0.0032,
"step": 3494
},
{
"epoch": 6.64,
"grad_norm": 0.5020560026168823,
"learning_rate": 1.6040692917179088e-05,
"loss": 0.0056,
"step": 3495
},
{
"epoch": 6.65,
"grad_norm": 0.12944163382053375,
"learning_rate": 1.602434448623336e-05,
"loss": 0.0017,
"step": 3496
},
{
"epoch": 6.65,
"grad_norm": 0.27128374576568604,
"learning_rate": 1.600800135425214e-05,
"loss": 0.0037,
"step": 3497
},
{
"epoch": 6.65,
"grad_norm": 0.24607905745506287,
"learning_rate": 1.599166352743203e-05,
"loss": 0.0021,
"step": 3498
},
{
"epoch": 6.65,
"grad_norm": 0.17547596991062164,
"learning_rate": 1.5975331011967628e-05,
"loss": 0.0008,
"step": 3499
},
{
"epoch": 6.65,
"grad_norm": 0.1530647873878479,
"learning_rate": 1.595900381405151e-05,
"loss": 0.0039,
"step": 3500
},
{
"epoch": 6.65,
"eval_blimp_filtered_avg": 0.7176119402985075,
"eval_blimp_filtered_std": 0.0049939039503697985,
"step": 3500
},
{
"epoch": 6.65,
"eval_blimp_supplement_avg": 0.7780172413793104,
"eval_blimp_supplement_std": 0.017994231080819156,
"step": 3500
},
{
"epoch": 6.65,
"eval_vqa_filtered_avg": 0.32,
"eval_vqa_filtered_std": 0.046882617226215034,
"step": 3500
},
{
"epoch": 6.65,
"eval_winoground_filtered_avg": 0.5,
"eval_winoground_filtered_std": 0.050251890762960605,
"step": 3500
},
{
"epoch": 6.66,
"grad_norm": 0.49169105291366577,
"learning_rate": 1.594268193987428e-05,
"loss": 0.0048,
"step": 3501
},
{
"epoch": 6.66,
"grad_norm": 0.20428015291690826,
"learning_rate": 1.5926365395624436e-05,
"loss": 0.0018,
"step": 3502
},
{
"epoch": 6.66,
"grad_norm": 0.08583717048168182,
"learning_rate": 1.5910054187488543e-05,
"loss": 0.0006,
"step": 3503
},
{
"epoch": 6.66,
"grad_norm": 0.43405020236968994,
"learning_rate": 1.5893748321651097e-05,
"loss": 0.0184,
"step": 3504
},
{
"epoch": 6.66,
"grad_norm": 0.7371810078620911,
"learning_rate": 1.5877447804294572e-05,
"loss": 0.0103,
"step": 3505
},
{
"epoch": 6.67,
"grad_norm": 0.28194859623908997,
"learning_rate": 1.5861152641599413e-05,
"loss": 0.0026,
"step": 3506
},
{
"epoch": 6.67,
"grad_norm": 0.6520022749900818,
"learning_rate": 1.5844862839744077e-05,
"loss": 0.002,
"step": 3507
},
{
"epoch": 6.67,
"grad_norm": 0.2972497344017029,
"learning_rate": 1.5828578404904903e-05,
"loss": 0.0034,
"step": 3508
},
{
"epoch": 6.67,
"grad_norm": 0.07361922413110733,
"learning_rate": 1.581229934325629e-05,
"loss": 0.0008,
"step": 3509
},
{
"epoch": 6.67,
"grad_norm": 0.4877745509147644,
"learning_rate": 1.5796025660970513e-05,
"loss": 0.003,
"step": 3510
},
{
"epoch": 6.67,
"grad_norm": 0.6851876974105835,
"learning_rate": 1.5779757364217882e-05,
"loss": 0.0027,
"step": 3511
},
{
"epoch": 6.68,
"grad_norm": 0.20010878145694733,
"learning_rate": 1.5763494459166626e-05,
"loss": 0.0017,
"step": 3512
},
{
"epoch": 6.68,
"grad_norm": 0.21758009493350983,
"learning_rate": 1.574723695198291e-05,
"loss": 0.0018,
"step": 3513
},
{
"epoch": 6.68,
"grad_norm": 0.26136043667793274,
"learning_rate": 1.573098484883091e-05,
"loss": 0.0015,
"step": 3514
},
{
"epoch": 6.68,
"grad_norm": 0.8989701867103577,
"learning_rate": 1.57147381558727e-05,
"loss": 0.0038,
"step": 3515
},
{
"epoch": 6.68,
"grad_norm": 0.13290181756019592,
"learning_rate": 1.569849687926834e-05,
"loss": 0.0011,
"step": 3516
},
{
"epoch": 6.69,
"grad_norm": 0.36345598101615906,
"learning_rate": 1.56822610251758e-05,
"loss": 0.003,
"step": 3517
},
{
"epoch": 6.69,
"grad_norm": 0.6586345434188843,
"learning_rate": 1.5666030599751047e-05,
"loss": 0.0089,
"step": 3518
},
{
"epoch": 6.69,
"grad_norm": 0.3388601839542389,
"learning_rate": 1.5649805609147912e-05,
"loss": 0.0051,
"step": 3519
},
{
"epoch": 6.69,
"grad_norm": 0.09850093722343445,
"learning_rate": 1.5633586059518244e-05,
"loss": 0.0005,
"step": 3520
},
{
"epoch": 6.69,
"grad_norm": 0.45225560665130615,
"learning_rate": 1.561737195701179e-05,
"loss": 0.0023,
"step": 3521
},
{
"epoch": 6.7,
"grad_norm": 0.6281595230102539,
"learning_rate": 1.560116330777623e-05,
"loss": 0.0049,
"step": 3522
},
{
"epoch": 6.7,
"grad_norm": 0.12063998728990555,
"learning_rate": 1.5584960117957186e-05,
"loss": 0.0009,
"step": 3523
},
{
"epoch": 6.7,
"grad_norm": 0.5553130507469177,
"learning_rate": 1.556876239369821e-05,
"loss": 0.006,
"step": 3524
},
{
"epoch": 6.7,
"grad_norm": 0.3684433400630951,
"learning_rate": 1.5552570141140767e-05,
"loss": 0.0055,
"step": 3525
},
{
"epoch": 6.7,
"grad_norm": 0.10913853347301483,
"learning_rate": 1.553638336642426e-05,
"loss": 0.0005,
"step": 3526
},
{
"epoch": 6.71,
"grad_norm": 0.14461100101470947,
"learning_rate": 1.5520202075686046e-05,
"loss": 0.0021,
"step": 3527
},
{
"epoch": 6.71,
"grad_norm": 0.4109615683555603,
"learning_rate": 1.5504026275061326e-05,
"loss": 0.0113,
"step": 3528
},
{
"epoch": 6.71,
"grad_norm": 0.33119291067123413,
"learning_rate": 1.5487855970683293e-05,
"loss": 0.0046,
"step": 3529
},
{
"epoch": 6.71,
"grad_norm": 0.1462479531764984,
"learning_rate": 1.5471691168683025e-05,
"loss": 0.0016,
"step": 3530
},
{
"epoch": 6.71,
"grad_norm": 0.2668493688106537,
"learning_rate": 1.5455531875189508e-05,
"loss": 0.0026,
"step": 3531
},
{
"epoch": 6.71,
"grad_norm": 0.34774523973464966,
"learning_rate": 1.5439378096329652e-05,
"loss": 0.0074,
"step": 3532
},
{
"epoch": 6.72,
"grad_norm": 0.05152219906449318,
"learning_rate": 1.542322983822827e-05,
"loss": 0.0004,
"step": 3533
},
{
"epoch": 6.72,
"grad_norm": 0.3965175449848175,
"learning_rate": 1.5407087107008086e-05,
"loss": 0.0022,
"step": 3534
},
{
"epoch": 6.72,
"grad_norm": 0.2895212769508362,
"learning_rate": 1.5390949908789722e-05,
"loss": 0.0024,
"step": 3535
},
{
"epoch": 6.72,
"grad_norm": 0.2494373917579651,
"learning_rate": 1.537481824969171e-05,
"loss": 0.0015,
"step": 3536
},
{
"epoch": 6.72,
"grad_norm": 0.3762430250644684,
"learning_rate": 1.5358692135830465e-05,
"loss": 0.0023,
"step": 3537
},
{
"epoch": 6.73,
"grad_norm": 0.730263888835907,
"learning_rate": 1.5342571573320344e-05,
"loss": 0.0127,
"step": 3538
},
{
"epoch": 6.73,
"grad_norm": 0.3032275438308716,
"learning_rate": 1.532645656827352e-05,
"loss": 0.002,
"step": 3539
},
{
"epoch": 6.73,
"grad_norm": 0.24880768358707428,
"learning_rate": 1.531034712680015e-05,
"loss": 0.0019,
"step": 3540
},
{
"epoch": 6.73,
"grad_norm": 0.5902478694915771,
"learning_rate": 1.529424325500822e-05,
"loss": 0.0024,
"step": 3541
},
{
"epoch": 6.73,
"grad_norm": 0.4602148234844208,
"learning_rate": 1.5278144959003624e-05,
"loss": 0.0046,
"step": 3542
},
{
"epoch": 6.74,
"grad_norm": 0.1478835493326187,
"learning_rate": 1.5262052244890143e-05,
"loss": 0.0019,
"step": 3543
},
{
"epoch": 6.74,
"grad_norm": 0.262362539768219,
"learning_rate": 1.5245965118769429e-05,
"loss": 0.0031,
"step": 3544
},
{
"epoch": 6.74,
"grad_norm": 0.2350742071866989,
"learning_rate": 1.5229883586741035e-05,
"loss": 0.0024,
"step": 3545
},
{
"epoch": 6.74,
"grad_norm": 0.31284821033477783,
"learning_rate": 1.5213807654902367e-05,
"loss": 0.0045,
"step": 3546
},
{
"epoch": 6.74,
"grad_norm": 0.3196871280670166,
"learning_rate": 1.5197737329348756e-05,
"loss": 0.0011,
"step": 3547
},
{
"epoch": 6.75,
"grad_norm": 0.3176352083683014,
"learning_rate": 1.5181672616173332e-05,
"loss": 0.0023,
"step": 3548
},
{
"epoch": 6.75,
"grad_norm": 0.09277216345071793,
"learning_rate": 1.5165613521467172e-05,
"loss": 0.0008,
"step": 3549
},
{
"epoch": 6.75,
"grad_norm": 0.355557918548584,
"learning_rate": 1.5149560051319184e-05,
"loss": 0.0009,
"step": 3550
},
{
"epoch": 6.75,
"grad_norm": 0.8923730850219727,
"learning_rate": 1.5133512211816145e-05,
"loss": 0.0198,
"step": 3551
},
{
"epoch": 6.75,
"grad_norm": 0.7113198637962341,
"learning_rate": 1.5117470009042693e-05,
"loss": 0.0037,
"step": 3552
},
{
"epoch": 6.75,
"grad_norm": 0.10487859696149826,
"learning_rate": 1.510143344908137e-05,
"loss": 0.0008,
"step": 3553
},
{
"epoch": 6.76,
"grad_norm": 0.3983052968978882,
"learning_rate": 1.50854025380125e-05,
"loss": 0.0019,
"step": 3554
},
{
"epoch": 6.76,
"grad_norm": 0.16239959001541138,
"learning_rate": 1.5069377281914353e-05,
"loss": 0.0018,
"step": 3555
},
{
"epoch": 6.76,
"grad_norm": 0.18855832517147064,
"learning_rate": 1.5053357686862996e-05,
"loss": 0.0017,
"step": 3556
},
{
"epoch": 6.76,
"grad_norm": 0.3614652156829834,
"learning_rate": 1.5037343758932366e-05,
"loss": 0.0033,
"step": 3557
},
{
"epoch": 6.76,
"grad_norm": 0.391305148601532,
"learning_rate": 1.5021335504194265e-05,
"loss": 0.0076,
"step": 3558
},
{
"epoch": 6.77,
"grad_norm": 0.30007970333099365,
"learning_rate": 1.50053329287183e-05,
"loss": 0.0032,
"step": 3559
},
{
"epoch": 6.77,
"grad_norm": 0.1186726838350296,
"learning_rate": 1.4989336038571987e-05,
"loss": 0.0009,
"step": 3560
},
{
"epoch": 6.77,
"grad_norm": 0.1658875048160553,
"learning_rate": 1.4973344839820644e-05,
"loss": 0.0014,
"step": 3561
},
{
"epoch": 6.77,
"grad_norm": 0.24117697775363922,
"learning_rate": 1.4957359338527446e-05,
"loss": 0.0011,
"step": 3562
},
{
"epoch": 6.77,
"grad_norm": 0.45898425579071045,
"learning_rate": 1.4941379540753389e-05,
"loss": 0.0047,
"step": 3563
},
{
"epoch": 6.78,
"grad_norm": 0.4130459725856781,
"learning_rate": 1.4925405452557357e-05,
"loss": 0.0071,
"step": 3564
},
{
"epoch": 6.78,
"grad_norm": 0.07529676705598831,
"learning_rate": 1.4909437079995987e-05,
"loss": 0.0006,
"step": 3565
},
{
"epoch": 6.78,
"grad_norm": 0.2320372611284256,
"learning_rate": 1.4893474429123833e-05,
"loss": 0.0021,
"step": 3566
},
{
"epoch": 6.78,
"grad_norm": 0.1590714305639267,
"learning_rate": 1.4877517505993226e-05,
"loss": 0.0009,
"step": 3567
},
{
"epoch": 6.78,
"grad_norm": 0.12127666175365448,
"learning_rate": 1.4861566316654347e-05,
"loss": 0.0015,
"step": 3568
},
{
"epoch": 6.79,
"grad_norm": 0.4866175949573517,
"learning_rate": 1.4845620867155191e-05,
"loss": 0.0066,
"step": 3569
},
{
"epoch": 6.79,
"grad_norm": 0.25652506947517395,
"learning_rate": 1.482968116354159e-05,
"loss": 0.0018,
"step": 3570
},
{
"epoch": 6.79,
"grad_norm": 0.25513944029808044,
"learning_rate": 1.481374721185719e-05,
"loss": 0.0014,
"step": 3571
},
{
"epoch": 6.79,
"grad_norm": 1.394303560256958,
"learning_rate": 1.4797819018143438e-05,
"loss": 0.006,
"step": 3572
},
{
"epoch": 6.79,
"grad_norm": 0.2836877405643463,
"learning_rate": 1.4781896588439659e-05,
"loss": 0.0034,
"step": 3573
},
{
"epoch": 6.79,
"grad_norm": 0.3531421422958374,
"learning_rate": 1.47659799287829e-05,
"loss": 0.0021,
"step": 3574
},
{
"epoch": 6.8,
"grad_norm": 0.20916636288166046,
"learning_rate": 1.4750069045208108e-05,
"loss": 0.0021,
"step": 3575
},
{
"epoch": 6.8,
"grad_norm": 0.32649677991867065,
"learning_rate": 1.4734163943747986e-05,
"loss": 0.0035,
"step": 3576
},
{
"epoch": 6.8,
"grad_norm": 0.24576488137245178,
"learning_rate": 1.4718264630433068e-05,
"loss": 0.0033,
"step": 3577
},
{
"epoch": 6.8,
"grad_norm": 0.10142157226800919,
"learning_rate": 1.470237111129168e-05,
"loss": 0.0007,
"step": 3578
},
{
"epoch": 6.8,
"grad_norm": 0.5828949809074402,
"learning_rate": 1.468648339234996e-05,
"loss": 0.0016,
"step": 3579
},
{
"epoch": 6.81,
"grad_norm": 0.12382181733846664,
"learning_rate": 1.4670601479631845e-05,
"loss": 0.0008,
"step": 3580
},
{
"epoch": 6.81,
"grad_norm": 0.3638239800930023,
"learning_rate": 1.4654725379159068e-05,
"loss": 0.0027,
"step": 3581
},
{
"epoch": 6.81,
"grad_norm": 0.037189967930316925,
"learning_rate": 1.463885509695116e-05,
"loss": 0.0004,
"step": 3582
},
{
"epoch": 6.81,
"grad_norm": 0.25927358865737915,
"learning_rate": 1.462299063902544e-05,
"loss": 0.0026,
"step": 3583
},
{
"epoch": 6.81,
"grad_norm": 0.16505324840545654,
"learning_rate": 1.460713201139705e-05,
"loss": 0.0009,
"step": 3584
},
{
"epoch": 6.82,
"grad_norm": 0.7466572523117065,
"learning_rate": 1.4591279220078856e-05,
"loss": 0.0105,
"step": 3585
},
{
"epoch": 6.82,
"grad_norm": 0.33323386311531067,
"learning_rate": 1.4575432271081587e-05,
"loss": 0.0054,
"step": 3586
},
{
"epoch": 6.82,
"grad_norm": 0.12661555409431458,
"learning_rate": 1.4559591170413707e-05,
"loss": 0.0009,
"step": 3587
},
{
"epoch": 6.82,
"grad_norm": 0.26080089807510376,
"learning_rate": 1.4543755924081478e-05,
"loss": 0.0014,
"step": 3588
},
{
"epoch": 6.82,
"grad_norm": 0.14986099302768707,
"learning_rate": 1.4527926538088938e-05,
"loss": 0.001,
"step": 3589
},
{
"epoch": 6.83,
"grad_norm": 0.4122147858142853,
"learning_rate": 1.4512103018437908e-05,
"loss": 0.0045,
"step": 3590
},
{
"epoch": 6.83,
"grad_norm": 0.6606019139289856,
"learning_rate": 1.4496285371127983e-05,
"loss": 0.0243,
"step": 3591
},
{
"epoch": 6.83,
"grad_norm": 0.0817459300160408,
"learning_rate": 1.4480473602156523e-05,
"loss": 0.0006,
"step": 3592
},
{
"epoch": 6.83,
"grad_norm": 0.24235716462135315,
"learning_rate": 1.4464667717518693e-05,
"loss": 0.0024,
"step": 3593
},
{
"epoch": 6.83,
"grad_norm": 0.3966996371746063,
"learning_rate": 1.4448867723207365e-05,
"loss": 0.003,
"step": 3594
},
{
"epoch": 6.83,
"grad_norm": 0.09972861409187317,
"learning_rate": 1.4433073625213242e-05,
"loss": 0.0008,
"step": 3595
},
{
"epoch": 6.84,
"grad_norm": 0.42135730385780334,
"learning_rate": 1.4417285429524753e-05,
"loss": 0.0016,
"step": 3596
},
{
"epoch": 6.84,
"grad_norm": 0.268408864736557,
"learning_rate": 1.4401503142128101e-05,
"loss": 0.0013,
"step": 3597
},
{
"epoch": 6.84,
"grad_norm": 0.6318941116333008,
"learning_rate": 1.4385726769007233e-05,
"loss": 0.0076,
"step": 3598
},
{
"epoch": 6.84,
"grad_norm": 0.30788454413414,
"learning_rate": 1.4369956316143907e-05,
"loss": 0.0021,
"step": 3599
},
{
"epoch": 6.84,
"grad_norm": 0.7555286288261414,
"learning_rate": 1.435419178951755e-05,
"loss": 0.0022,
"step": 3600
},
{
"epoch": 6.84,
"eval_blimp_filtered_avg": 0.7185074626865672,
"eval_blimp_filtered_std": 0.004969150642461204,
"step": 3600
},
{
"epoch": 6.84,
"eval_blimp_supplement_avg": 0.771551724137931,
"eval_blimp_supplement_std": 0.018125444435109727,
"step": 3600
},
{
"epoch": 6.84,
"eval_vqa_filtered_avg": 0.32,
"eval_vqa_filtered_std": 0.046882617226215034,
"step": 3600
},
{
"epoch": 6.84,
"eval_winoground_filtered_avg": 0.49,
"eval_winoground_filtered_std": 0.05024183937956912,
"step": 3600
},
{
"epoch": 6.85,
"grad_norm": 0.46336594223976135,
"learning_rate": 1.4338433195105417e-05,
"loss": 0.0035,
"step": 3601
},
{
"epoch": 6.85,
"grad_norm": 0.24471013247966766,
"learning_rate": 1.4322680538882483e-05,
"loss": 0.0015,
"step": 3602
},
{
"epoch": 6.85,
"grad_norm": 0.16866247355937958,
"learning_rate": 1.4306933826821466e-05,
"loss": 0.0011,
"step": 3603
},
{
"epoch": 6.85,
"grad_norm": 0.20353670418262482,
"learning_rate": 1.4291193064892848e-05,
"loss": 0.0018,
"step": 3604
},
{
"epoch": 6.85,
"grad_norm": 0.7487688064575195,
"learning_rate": 1.4275458259064838e-05,
"loss": 0.0056,
"step": 3605
},
{
"epoch": 6.86,
"grad_norm": 0.1938810646533966,
"learning_rate": 1.42597294153034e-05,
"loss": 0.0012,
"step": 3606
},
{
"epoch": 6.86,
"grad_norm": 0.2295682430267334,
"learning_rate": 1.4244006539572228e-05,
"loss": 0.0015,
"step": 3607
},
{
"epoch": 6.86,
"grad_norm": 0.0883665606379509,
"learning_rate": 1.4228289637832757e-05,
"loss": 0.0003,
"step": 3608
},
{
"epoch": 6.86,
"grad_norm": 0.34072139859199524,
"learning_rate": 1.4212578716044144e-05,
"loss": 0.0054,
"step": 3609
},
{
"epoch": 6.86,
"grad_norm": 0.09849987924098969,
"learning_rate": 1.419687378016333e-05,
"loss": 0.0009,
"step": 3610
},
{
"epoch": 6.87,
"grad_norm": 0.5105400681495667,
"learning_rate": 1.4181174836144899e-05,
"loss": 0.0038,
"step": 3611
},
{
"epoch": 6.87,
"grad_norm": 0.4277925193309784,
"learning_rate": 1.416548188994125e-05,
"loss": 0.0044,
"step": 3612
},
{
"epoch": 6.87,
"grad_norm": 0.2523181736469269,
"learning_rate": 1.4149794947502452e-05,
"loss": 0.0025,
"step": 3613
},
{
"epoch": 6.87,
"grad_norm": 0.15429171919822693,
"learning_rate": 1.4134114014776323e-05,
"loss": 0.0058,
"step": 3614
},
{
"epoch": 6.87,
"grad_norm": 0.12030776590108871,
"learning_rate": 1.4118439097708388e-05,
"loss": 0.0006,
"step": 3615
},
{
"epoch": 6.87,
"grad_norm": 0.43818801641464233,
"learning_rate": 1.4102770202241904e-05,
"loss": 0.0056,
"step": 3616
},
{
"epoch": 6.88,
"grad_norm": 0.33388832211494446,
"learning_rate": 1.4087107334317831e-05,
"loss": 0.0013,
"step": 3617
},
{
"epoch": 6.88,
"grad_norm": 0.376590371131897,
"learning_rate": 1.4071450499874851e-05,
"loss": 0.0022,
"step": 3618
},
{
"epoch": 6.88,
"grad_norm": 0.15228979289531708,
"learning_rate": 1.4055799704849386e-05,
"loss": 0.0039,
"step": 3619
},
{
"epoch": 6.88,
"grad_norm": 0.8668777942657471,
"learning_rate": 1.4040154955175494e-05,
"loss": 0.0145,
"step": 3620
},
{
"epoch": 6.88,
"grad_norm": 0.5097842216491699,
"learning_rate": 1.4024516256785026e-05,
"loss": 0.0019,
"step": 3621
},
{
"epoch": 6.89,
"grad_norm": 0.1299673169851303,
"learning_rate": 1.4008883615607489e-05,
"loss": 0.0022,
"step": 3622
},
{
"epoch": 6.89,
"grad_norm": 0.1287999451160431,
"learning_rate": 1.3993257037570103e-05,
"loss": 0.0015,
"step": 3623
},
{
"epoch": 6.89,
"grad_norm": 0.6111364364624023,
"learning_rate": 1.3977636528597794e-05,
"loss": 0.0014,
"step": 3624
},
{
"epoch": 6.89,
"grad_norm": 0.15051616728305817,
"learning_rate": 1.396202209461318e-05,
"loss": 0.0017,
"step": 3625
},
{
"epoch": 6.89,
"grad_norm": 1.2869657278060913,
"learning_rate": 1.394641374153658e-05,
"loss": 0.0289,
"step": 3626
},
{
"epoch": 6.9,
"grad_norm": 0.17569074034690857,
"learning_rate": 1.3930811475285998e-05,
"loss": 0.0013,
"step": 3627
},
{
"epoch": 6.9,
"grad_norm": 0.6146298050880432,
"learning_rate": 1.391521530177717e-05,
"loss": 0.0135,
"step": 3628
},
{
"epoch": 6.9,
"grad_norm": 0.27327123284339905,
"learning_rate": 1.3899625226923449e-05,
"loss": 0.0042,
"step": 3629
},
{
"epoch": 6.9,
"grad_norm": 0.31641021370887756,
"learning_rate": 1.388404125663596e-05,
"loss": 0.0023,
"step": 3630
},
{
"epoch": 6.9,
"grad_norm": 0.7004881501197815,
"learning_rate": 1.3868463396823425e-05,
"loss": 0.0048,
"step": 3631
},
{
"epoch": 6.9,
"grad_norm": 0.40467655658721924,
"learning_rate": 1.3852891653392335e-05,
"loss": 0.0038,
"step": 3632
},
{
"epoch": 6.91,
"grad_norm": 0.43152886629104614,
"learning_rate": 1.3837326032246802e-05,
"loss": 0.0037,
"step": 3633
},
{
"epoch": 6.91,
"grad_norm": 0.23595495522022247,
"learning_rate": 1.3821766539288644e-05,
"loss": 0.0037,
"step": 3634
},
{
"epoch": 6.91,
"grad_norm": 0.6540746092796326,
"learning_rate": 1.3806213180417344e-05,
"loss": 0.0063,
"step": 3635
},
{
"epoch": 6.91,
"grad_norm": 0.06413160264492035,
"learning_rate": 1.3790665961530063e-05,
"loss": 0.0004,
"step": 3636
},
{
"epoch": 6.91,
"grad_norm": 0.19548776745796204,
"learning_rate": 1.377512488852164e-05,
"loss": 0.0007,
"step": 3637
},
{
"epoch": 6.92,
"grad_norm": 0.0924106165766716,
"learning_rate": 1.3759589967284559e-05,
"loss": 0.0015,
"step": 3638
},
{
"epoch": 6.92,
"grad_norm": 0.09927769005298615,
"learning_rate": 1.3744061203709033e-05,
"loss": 0.0007,
"step": 3639
},
{
"epoch": 6.92,
"grad_norm": 0.19640189409255981,
"learning_rate": 1.3728538603682844e-05,
"loss": 0.002,
"step": 3640
},
{
"epoch": 6.92,
"grad_norm": 0.09748388826847076,
"learning_rate": 1.3713022173091531e-05,
"loss": 0.0019,
"step": 3641
},
{
"epoch": 6.92,
"grad_norm": 0.24196313321590424,
"learning_rate": 1.3697511917818243e-05,
"loss": 0.0016,
"step": 3642
},
{
"epoch": 6.93,
"grad_norm": 0.2343958020210266,
"learning_rate": 1.3682007843743794e-05,
"loss": 0.0014,
"step": 3643
},
{
"epoch": 6.93,
"grad_norm": 0.5504863858222961,
"learning_rate": 1.3666509956746652e-05,
"loss": 0.0033,
"step": 3644
},
{
"epoch": 6.93,
"grad_norm": 0.5008193254470825,
"learning_rate": 1.3651018262702975e-05,
"loss": 0.0032,
"step": 3645
},
{
"epoch": 6.93,
"grad_norm": 0.462004154920578,
"learning_rate": 1.3635532767486502e-05,
"loss": 0.0051,
"step": 3646
},
{
"epoch": 6.93,
"grad_norm": 0.3701416850090027,
"learning_rate": 1.3620053476968696e-05,
"loss": 0.0033,
"step": 3647
},
{
"epoch": 6.94,
"grad_norm": 0.31406447291374207,
"learning_rate": 1.3604580397018624e-05,
"loss": 0.0033,
"step": 3648
},
{
"epoch": 6.94,
"grad_norm": 0.24769949913024902,
"learning_rate": 1.3589113533503008e-05,
"loss": 0.005,
"step": 3649
},
{
"epoch": 6.94,
"grad_norm": 0.17918239533901215,
"learning_rate": 1.3573652892286215e-05,
"loss": 0.0011,
"step": 3650
},
{
"epoch": 6.94,
"grad_norm": 0.44212087988853455,
"learning_rate": 1.3558198479230257e-05,
"loss": 0.0047,
"step": 3651
},
{
"epoch": 6.94,
"grad_norm": 0.19054286181926727,
"learning_rate": 1.3542750300194772e-05,
"loss": 0.0022,
"step": 3652
},
{
"epoch": 6.94,
"grad_norm": 0.3542223572731018,
"learning_rate": 1.352730836103703e-05,
"loss": 0.0074,
"step": 3653
},
{
"epoch": 6.95,
"grad_norm": 0.11912594735622406,
"learning_rate": 1.3511872667611987e-05,
"loss": 0.0006,
"step": 3654
},
{
"epoch": 6.95,
"grad_norm": 0.6287583708763123,
"learning_rate": 1.349644322577214e-05,
"loss": 0.0032,
"step": 3655
},
{
"epoch": 6.95,
"grad_norm": 0.07936245948076248,
"learning_rate": 1.3481020041367717e-05,
"loss": 0.0014,
"step": 3656
},
{
"epoch": 6.95,
"grad_norm": 0.6659007668495178,
"learning_rate": 1.3465603120246474e-05,
"loss": 0.007,
"step": 3657
},
{
"epoch": 6.95,
"grad_norm": 0.740180492401123,
"learning_rate": 1.3450192468253876e-05,
"loss": 0.0057,
"step": 3658
},
{
"epoch": 6.96,
"grad_norm": 0.11990676820278168,
"learning_rate": 1.3434788091232963e-05,
"loss": 0.0022,
"step": 3659
},
{
"epoch": 6.96,
"grad_norm": 0.5384098291397095,
"learning_rate": 1.341938999502441e-05,
"loss": 0.0061,
"step": 3660
},
{
"epoch": 6.96,
"grad_norm": 0.23714251816272736,
"learning_rate": 1.3403998185466512e-05,
"loss": 0.0007,
"step": 3661
},
{
"epoch": 6.96,
"grad_norm": 0.4207341969013214,
"learning_rate": 1.3388612668395175e-05,
"loss": 0.0036,
"step": 3662
},
{
"epoch": 6.96,
"grad_norm": 0.5733606815338135,
"learning_rate": 1.3373233449643923e-05,
"loss": 0.0029,
"step": 3663
},
{
"epoch": 6.97,
"grad_norm": 0.5953826308250427,
"learning_rate": 1.335786053504388e-05,
"loss": 0.0037,
"step": 3664
},
{
"epoch": 6.97,
"grad_norm": 0.24752850830554962,
"learning_rate": 1.3342493930423824e-05,
"loss": 0.0026,
"step": 3665
},
{
"epoch": 6.97,
"grad_norm": 0.2917594611644745,
"learning_rate": 1.3327133641610066e-05,
"loss": 0.0061,
"step": 3666
},
{
"epoch": 6.97,
"grad_norm": 0.28315451741218567,
"learning_rate": 1.331177967442659e-05,
"loss": 0.0033,
"step": 3667
},
{
"epoch": 6.97,
"grad_norm": 0.154571071267128,
"learning_rate": 1.329643203469495e-05,
"loss": 0.001,
"step": 3668
},
{
"epoch": 6.98,
"grad_norm": 0.262869268655777,
"learning_rate": 1.3281090728234308e-05,
"loss": 0.0019,
"step": 3669
},
{
"epoch": 6.98,
"grad_norm": 0.500807523727417,
"learning_rate": 1.3265755760861424e-05,
"loss": 0.0009,
"step": 3670
},
{
"epoch": 6.98,
"grad_norm": 0.17312072217464447,
"learning_rate": 1.3250427138390653e-05,
"loss": 0.0009,
"step": 3671
},
{
"epoch": 6.98,
"grad_norm": 0.5070456266403198,
"learning_rate": 1.3235104866633948e-05,
"loss": 0.0028,
"step": 3672
},
{
"epoch": 6.98,
"grad_norm": 0.18190422654151917,
"learning_rate": 1.3219788951400843e-05,
"loss": 0.0019,
"step": 3673
},
{
"epoch": 6.98,
"grad_norm": 0.47402602434158325,
"learning_rate": 1.3204479398498496e-05,
"loss": 0.0049,
"step": 3674
},
{
"epoch": 6.99,
"grad_norm": 0.3839764893054962,
"learning_rate": 1.3189176213731593e-05,
"loss": 0.0044,
"step": 3675
},
{
"epoch": 6.99,
"grad_norm": 0.2484479397535324,
"learning_rate": 1.317387940290247e-05,
"loss": 0.0013,
"step": 3676
},
{
"epoch": 6.99,
"grad_norm": 0.2561711072921753,
"learning_rate": 1.3158588971811002e-05,
"loss": 0.0018,
"step": 3677
},
{
"epoch": 6.99,
"grad_norm": 0.18050727248191833,
"learning_rate": 1.3143304926254664e-05,
"loss": 0.0019,
"step": 3678
},
{
"epoch": 6.99,
"grad_norm": 0.06422854959964752,
"learning_rate": 1.3128027272028506e-05,
"loss": 0.0005,
"step": 3679
},
{
"epoch": 7.0,
"grad_norm": 0.90765380859375,
"learning_rate": 1.3112756014925148e-05,
"loss": 0.0054,
"step": 3680
},
{
"epoch": 7.0,
"grad_norm": 0.22836054861545563,
"learning_rate": 1.3097491160734797e-05,
"loss": 0.0017,
"step": 3681
},
{
"epoch": 7.0,
"grad_norm": 0.3712337017059326,
"learning_rate": 1.3082232715245226e-05,
"loss": 0.0031,
"step": 3682
},
{
"epoch": 7.0,
"grad_norm": 0.2048632651567459,
"learning_rate": 1.306698068424177e-05,
"loss": 0.0041,
"step": 3683
},
{
"epoch": 7.0,
"grad_norm": 0.15447811782360077,
"learning_rate": 1.3051735073507342e-05,
"loss": 0.0008,
"step": 3684
},
{
"epoch": 7.01,
"grad_norm": 0.15409047901630402,
"learning_rate": 1.3036495888822438e-05,
"loss": 0.0008,
"step": 3685
},
{
"epoch": 7.01,
"grad_norm": 0.33175987005233765,
"learning_rate": 1.3021263135965062e-05,
"loss": 0.0022,
"step": 3686
},
{
"epoch": 7.01,
"grad_norm": 0.06664341688156128,
"learning_rate": 1.3006036820710847e-05,
"loss": 0.0007,
"step": 3687
},
{
"epoch": 7.01,
"grad_norm": 0.14851140975952148,
"learning_rate": 1.2990816948832945e-05,
"loss": 0.0013,
"step": 3688
},
{
"epoch": 7.01,
"grad_norm": 0.06507047265768051,
"learning_rate": 1.2975603526102069e-05,
"loss": 0.0005,
"step": 3689
},
{
"epoch": 7.02,
"grad_norm": 0.22226427495479584,
"learning_rate": 1.2960396558286484e-05,
"loss": 0.0026,
"step": 3690
},
{
"epoch": 7.02,
"grad_norm": 0.04268237575888634,
"learning_rate": 1.2945196051152044e-05,
"loss": 0.0003,
"step": 3691
},
{
"epoch": 7.02,
"grad_norm": 0.11892061680555344,
"learning_rate": 1.2930002010462082e-05,
"loss": 0.001,
"step": 3692
},
{
"epoch": 7.02,
"grad_norm": 0.20415839552879333,
"learning_rate": 1.2914814441977559e-05,
"loss": 0.0019,
"step": 3693
},
{
"epoch": 7.02,
"grad_norm": 0.3049502372741699,
"learning_rate": 1.2899633351456926e-05,
"loss": 0.0025,
"step": 3694
},
{
"epoch": 7.02,
"grad_norm": 0.022404780611395836,
"learning_rate": 1.2884458744656202e-05,
"loss": 0.0003,
"step": 3695
},
{
"epoch": 7.03,
"grad_norm": 0.037146471440792084,
"learning_rate": 1.2869290627328938e-05,
"loss": 0.0004,
"step": 3696
},
{
"epoch": 7.03,
"grad_norm": 0.07159695029258728,
"learning_rate": 1.2854129005226231e-05,
"loss": 0.0005,
"step": 3697
},
{
"epoch": 7.03,
"grad_norm": 0.02220475673675537,
"learning_rate": 1.2838973884096715e-05,
"loss": 0.0002,
"step": 3698
},
{
"epoch": 7.03,
"grad_norm": 0.0895579531788826,
"learning_rate": 1.2823825269686541e-05,
"loss": 0.0004,
"step": 3699
},
{
"epoch": 7.03,
"grad_norm": 0.29885226488113403,
"learning_rate": 1.2808683167739444e-05,
"loss": 0.0056,
"step": 3700
},
{
"epoch": 7.03,
"eval_blimp_filtered_avg": 0.718955223880597,
"eval_blimp_filtered_std": 0.004975282675003165,
"step": 3700
},
{
"epoch": 7.03,
"eval_blimp_supplement_avg": 0.7758620689655172,
"eval_blimp_supplement_std": 0.01816175577138046,
"step": 3700
},
{
"epoch": 7.03,
"eval_vqa_filtered_avg": 0.32,
"eval_vqa_filtered_std": 0.046882617226215034,
"step": 3700
},
{
"epoch": 7.03,
"eval_winoground_filtered_avg": 0.52,
"eval_winoground_filtered_std": 0.05021167315686779,
"step": 3700
},
{
"epoch": 7.04,
"grad_norm": 0.38750842213630676,
"learning_rate": 1.2793547583996614e-05,
"loss": 0.0021,
"step": 3701
},
{
"epoch": 7.04,
"grad_norm": 0.44247809052467346,
"learning_rate": 1.2778418524196835e-05,
"loss": 0.0103,
"step": 3702
},
{
"epoch": 7.04,
"grad_norm": 0.5820782780647278,
"learning_rate": 1.2763295994076386e-05,
"loss": 0.0011,
"step": 3703
},
{
"epoch": 7.04,
"grad_norm": 0.23626424372196198,
"learning_rate": 1.2748179999369079e-05,
"loss": 0.0011,
"step": 3704
},
{
"epoch": 7.04,
"grad_norm": 0.08954320847988129,
"learning_rate": 1.2733070545806244e-05,
"loss": 0.0006,
"step": 3705
},
{
"epoch": 7.05,
"grad_norm": 0.1999017596244812,
"learning_rate": 1.2717967639116732e-05,
"loss": 0.0012,
"step": 3706
},
{
"epoch": 7.05,
"grad_norm": 0.2580361068248749,
"learning_rate": 1.2702871285026914e-05,
"loss": 0.0027,
"step": 3707
},
{
"epoch": 7.05,
"grad_norm": 0.22566673159599304,
"learning_rate": 1.2687781489260673e-05,
"loss": 0.0012,
"step": 3708
},
{
"epoch": 7.05,
"grad_norm": 0.04092521220445633,
"learning_rate": 1.267269825753941e-05,
"loss": 0.0004,
"step": 3709
},
{
"epoch": 7.05,
"grad_norm": 0.3995146155357361,
"learning_rate": 1.2657621595582018e-05,
"loss": 0.0016,
"step": 3710
},
{
"epoch": 7.06,
"grad_norm": 0.1520598828792572,
"learning_rate": 1.2642551509104954e-05,
"loss": 0.0021,
"step": 3711
},
{
"epoch": 7.06,
"grad_norm": 0.35466355085372925,
"learning_rate": 1.26274880038221e-05,
"loss": 0.0033,
"step": 3712
},
{
"epoch": 7.06,
"grad_norm": 0.20527786016464233,
"learning_rate": 1.2612431085444914e-05,
"loss": 0.0021,
"step": 3713
},
{
"epoch": 7.06,
"grad_norm": 0.2204934060573578,
"learning_rate": 1.2597380759682325e-05,
"loss": 0.002,
"step": 3714
},
{
"epoch": 7.06,
"grad_norm": 0.4650256633758545,
"learning_rate": 1.2582337032240757e-05,
"loss": 0.0054,
"step": 3715
},
{
"epoch": 7.06,
"grad_norm": 0.17817610502243042,
"learning_rate": 1.2567299908824146e-05,
"loss": 0.0016,
"step": 3716
},
{
"epoch": 7.07,
"grad_norm": 0.42798149585723877,
"learning_rate": 1.2552269395133925e-05,
"loss": 0.0031,
"step": 3717
},
{
"epoch": 7.07,
"grad_norm": 0.30149728059768677,
"learning_rate": 1.2537245496869002e-05,
"loss": 0.0088,
"step": 3718
},
{
"epoch": 7.07,
"grad_norm": 0.06808645278215408,
"learning_rate": 1.2522228219725791e-05,
"loss": 0.0005,
"step": 3719
},
{
"epoch": 7.07,
"grad_norm": 0.22406305372714996,
"learning_rate": 1.250721756939822e-05,
"loss": 0.0005,
"step": 3720
},
{
"epoch": 7.07,
"grad_norm": 0.04361759498715401,
"learning_rate": 1.2492213551577634e-05,
"loss": 0.0007,
"step": 3721
},
{
"epoch": 7.08,
"grad_norm": 0.1956644505262375,
"learning_rate": 1.247721617195295e-05,
"loss": 0.0029,
"step": 3722
},
{
"epoch": 7.08,
"grad_norm": 0.2625774145126343,
"learning_rate": 1.2462225436210507e-05,
"loss": 0.0028,
"step": 3723
},
{
"epoch": 7.08,
"grad_norm": 0.6972435116767883,
"learning_rate": 1.2447241350034147e-05,
"loss": 0.0028,
"step": 3724
},
{
"epoch": 7.08,
"grad_norm": 0.25095081329345703,
"learning_rate": 1.2432263919105187e-05,
"loss": 0.0026,
"step": 3725
},
{
"epoch": 7.08,
"grad_norm": 0.14239782094955444,
"learning_rate": 1.2417293149102422e-05,
"loss": 0.0006,
"step": 3726
},
{
"epoch": 7.09,
"grad_norm": 0.3643536865711212,
"learning_rate": 1.240232904570212e-05,
"loss": 0.0052,
"step": 3727
},
{
"epoch": 7.09,
"grad_norm": 0.11282506585121155,
"learning_rate": 1.2387371614578021e-05,
"loss": 0.0022,
"step": 3728
},
{
"epoch": 7.09,
"grad_norm": 0.3162233233451843,
"learning_rate": 1.2372420861401342e-05,
"loss": 0.0018,
"step": 3729
},
{
"epoch": 7.09,
"grad_norm": 0.21538382768630981,
"learning_rate": 1.2357476791840747e-05,
"loss": 0.0011,
"step": 3730
},
{
"epoch": 7.09,
"grad_norm": 0.08606944233179092,
"learning_rate": 1.2342539411562414e-05,
"loss": 0.0011,
"step": 3731
},
{
"epoch": 7.1,
"grad_norm": 0.12400852888822556,
"learning_rate": 1.2327608726229907e-05,
"loss": 0.0005,
"step": 3732
},
{
"epoch": 7.1,
"grad_norm": 0.28813502192497253,
"learning_rate": 1.2312684741504334e-05,
"loss": 0.0036,
"step": 3733
},
{
"epoch": 7.1,
"grad_norm": 0.07817381620407104,
"learning_rate": 1.229776746304421e-05,
"loss": 0.0009,
"step": 3734
},
{
"epoch": 7.1,
"grad_norm": 0.05238693207502365,
"learning_rate": 1.2282856896505523e-05,
"loss": 0.0004,
"step": 3735
},
{
"epoch": 7.1,
"grad_norm": 0.24943718314170837,
"learning_rate": 1.2267953047541706e-05,
"loss": 0.0026,
"step": 3736
},
{
"epoch": 7.1,
"grad_norm": 0.23178398609161377,
"learning_rate": 1.2253055921803684e-05,
"loss": 0.0009,
"step": 3737
},
{
"epoch": 7.11,
"grad_norm": 0.12075436115264893,
"learning_rate": 1.2238165524939767e-05,
"loss": 0.001,
"step": 3738
},
{
"epoch": 7.11,
"grad_norm": 0.09166933596134186,
"learning_rate": 1.2223281862595773e-05,
"loss": 0.0008,
"step": 3739
},
{
"epoch": 7.11,
"grad_norm": 0.16486713290214539,
"learning_rate": 1.2208404940414937e-05,
"loss": 0.002,
"step": 3740
},
{
"epoch": 7.11,
"grad_norm": 0.14066269993782043,
"learning_rate": 1.2193534764037944e-05,
"loss": 0.0011,
"step": 3741
},
{
"epoch": 7.11,
"grad_norm": 0.17790164053440094,
"learning_rate": 1.2178671339102923e-05,
"loss": 0.0015,
"step": 3742
},
{
"epoch": 7.12,
"grad_norm": 0.20579181611537933,
"learning_rate": 1.2163814671245444e-05,
"loss": 0.0022,
"step": 3743
},
{
"epoch": 7.12,
"grad_norm": 0.09432464838027954,
"learning_rate": 1.2148964766098512e-05,
"loss": 0.0003,
"step": 3744
},
{
"epoch": 7.12,
"grad_norm": 0.24268285930156708,
"learning_rate": 1.2134121629292554e-05,
"loss": 0.0015,
"step": 3745
},
{
"epoch": 7.12,
"grad_norm": 0.032425545156002045,
"learning_rate": 1.2119285266455486e-05,
"loss": 0.0003,
"step": 3746
},
{
"epoch": 7.12,
"grad_norm": 0.028813472017645836,
"learning_rate": 1.2104455683212566e-05,
"loss": 0.0002,
"step": 3747
},
{
"epoch": 7.13,
"grad_norm": 0.1043398454785347,
"learning_rate": 1.2089632885186565e-05,
"loss": 0.001,
"step": 3748
},
{
"epoch": 7.13,
"grad_norm": 0.020669279620051384,
"learning_rate": 1.2074816877997638e-05,
"loss": 0.0003,
"step": 3749
},
{
"epoch": 7.13,
"grad_norm": 0.3790847957134247,
"learning_rate": 1.2060007667263374e-05,
"loss": 0.0018,
"step": 3750
},
{
"epoch": 7.13,
"grad_norm": 0.08919855207204819,
"learning_rate": 1.2045205258598788e-05,
"loss": 0.0012,
"step": 3751
},
{
"epoch": 7.13,
"grad_norm": 0.2859713137149811,
"learning_rate": 1.2030409657616308e-05,
"loss": 0.0025,
"step": 3752
},
{
"epoch": 7.13,
"grad_norm": 0.0973004549741745,
"learning_rate": 1.201562086992579e-05,
"loss": 0.0006,
"step": 3753
},
{
"epoch": 7.14,
"grad_norm": 0.3528778553009033,
"learning_rate": 1.2000838901134503e-05,
"loss": 0.0028,
"step": 3754
},
{
"epoch": 7.14,
"grad_norm": 0.1071939468383789,
"learning_rate": 1.1986063756847131e-05,
"loss": 0.0007,
"step": 3755
},
{
"epoch": 7.14,
"grad_norm": 0.09322035312652588,
"learning_rate": 1.1971295442665755e-05,
"loss": 0.0005,
"step": 3756
},
{
"epoch": 7.14,
"grad_norm": 0.07125116139650345,
"learning_rate": 1.195653396418992e-05,
"loss": 0.0005,
"step": 3757
},
{
"epoch": 7.14,
"grad_norm": 0.06330625712871552,
"learning_rate": 1.1941779327016494e-05,
"loss": 0.0007,
"step": 3758
},
{
"epoch": 7.15,
"grad_norm": 0.07898931950330734,
"learning_rate": 1.1927031536739831e-05,
"loss": 0.0004,
"step": 3759
},
{
"epoch": 7.15,
"grad_norm": 0.02535741962492466,
"learning_rate": 1.1912290598951646e-05,
"loss": 0.0003,
"step": 3760
},
{
"epoch": 7.15,
"grad_norm": 0.3211941421031952,
"learning_rate": 1.1897556519241064e-05,
"loss": 0.0025,
"step": 3761
},
{
"epoch": 7.15,
"grad_norm": 0.17697042226791382,
"learning_rate": 1.1882829303194607e-05,
"loss": 0.002,
"step": 3762
},
{
"epoch": 7.15,
"grad_norm": 0.07358191907405853,
"learning_rate": 1.1868108956396208e-05,
"loss": 0.0017,
"step": 3763
},
{
"epoch": 7.16,
"grad_norm": 0.7348214983940125,
"learning_rate": 1.1853395484427177e-05,
"loss": 0.0099,
"step": 3764
},
{
"epoch": 7.16,
"grad_norm": 0.21505609154701233,
"learning_rate": 1.1838688892866221e-05,
"loss": 0.0029,
"step": 3765
},
{
"epoch": 7.16,
"grad_norm": 0.042086247354745865,
"learning_rate": 1.1823989187289477e-05,
"loss": 0.0004,
"step": 3766
},
{
"epoch": 7.16,
"grad_norm": 0.017496705055236816,
"learning_rate": 1.180929637327039e-05,
"loss": 0.0001,
"step": 3767
},
{
"epoch": 7.16,
"grad_norm": 0.3806425929069519,
"learning_rate": 1.179461045637988e-05,
"loss": 0.0042,
"step": 3768
},
{
"epoch": 7.17,
"grad_norm": 0.33081817626953125,
"learning_rate": 1.1779931442186193e-05,
"loss": 0.0013,
"step": 3769
},
{
"epoch": 7.17,
"grad_norm": 0.050627700984478,
"learning_rate": 1.1765259336254984e-05,
"loss": 0.0002,
"step": 3770
},
{
"epoch": 7.17,
"grad_norm": 0.048394400626420975,
"learning_rate": 1.1750594144149274e-05,
"loss": 0.0002,
"step": 3771
},
{
"epoch": 7.17,
"grad_norm": 0.07594076544046402,
"learning_rate": 1.1735935871429476e-05,
"loss": 0.0004,
"step": 3772
},
{
"epoch": 7.17,
"grad_norm": 0.10494375228881836,
"learning_rate": 1.172128452365337e-05,
"loss": 0.0003,
"step": 3773
},
{
"epoch": 7.17,
"grad_norm": 0.14415504038333893,
"learning_rate": 1.1706640106376105e-05,
"loss": 0.0009,
"step": 3774
},
{
"epoch": 7.18,
"grad_norm": 0.5666473507881165,
"learning_rate": 1.1692002625150239e-05,
"loss": 0.0014,
"step": 3775
},
{
"epoch": 7.18,
"grad_norm": 0.026823218911886215,
"learning_rate": 1.1677372085525638e-05,
"loss": 0.0002,
"step": 3776
},
{
"epoch": 7.18,
"grad_norm": 0.023652467876672745,
"learning_rate": 1.1662748493049605e-05,
"loss": 0.0002,
"step": 3777
},
{
"epoch": 7.18,
"grad_norm": 0.15022513270378113,
"learning_rate": 1.1648131853266735e-05,
"loss": 0.0014,
"step": 3778
},
{
"epoch": 7.18,
"grad_norm": 0.10312525182962418,
"learning_rate": 1.1633522171719055e-05,
"loss": 0.0009,
"step": 3779
},
{
"epoch": 7.19,
"grad_norm": 0.7911930680274963,
"learning_rate": 1.161891945394592e-05,
"loss": 0.0062,
"step": 3780
},
{
"epoch": 7.19,
"grad_norm": 0.3949212431907654,
"learning_rate": 1.1604323705484043e-05,
"loss": 0.002,
"step": 3781
},
{
"epoch": 7.19,
"grad_norm": 0.2346293181180954,
"learning_rate": 1.1589734931867496e-05,
"loss": 0.0008,
"step": 3782
},
{
"epoch": 7.19,
"grad_norm": 0.1307692527770996,
"learning_rate": 1.157515313862774e-05,
"loss": 0.0017,
"step": 3783
},
{
"epoch": 7.19,
"grad_norm": 0.4388043284416199,
"learning_rate": 1.1560578331293517e-05,
"loss": 0.0015,
"step": 3784
},
{
"epoch": 7.2,
"grad_norm": 0.3536868691444397,
"learning_rate": 1.1546010515390998e-05,
"loss": 0.0055,
"step": 3785
},
{
"epoch": 7.2,
"grad_norm": 0.33000433444976807,
"learning_rate": 1.1531449696443659e-05,
"loss": 0.0047,
"step": 3786
},
{
"epoch": 7.2,
"grad_norm": 0.15843449532985687,
"learning_rate": 1.1516895879972332e-05,
"loss": 0.0028,
"step": 3787
},
{
"epoch": 7.2,
"grad_norm": 0.14149197936058044,
"learning_rate": 1.1502349071495192e-05,
"loss": 0.0009,
"step": 3788
},
{
"epoch": 7.2,
"grad_norm": 0.10609809309244156,
"learning_rate": 1.1487809276527768e-05,
"loss": 0.002,
"step": 3789
},
{
"epoch": 7.21,
"grad_norm": 0.37543240189552307,
"learning_rate": 1.1473276500582916e-05,
"loss": 0.0023,
"step": 3790
},
{
"epoch": 7.21,
"grad_norm": 0.03538886457681656,
"learning_rate": 1.1458750749170824e-05,
"loss": 0.0003,
"step": 3791
},
{
"epoch": 7.21,
"grad_norm": 0.3260718882083893,
"learning_rate": 1.1444232027799066e-05,
"loss": 0.0036,
"step": 3792
},
{
"epoch": 7.21,
"grad_norm": 0.4806782901287079,
"learning_rate": 1.1429720341972471e-05,
"loss": 0.0028,
"step": 3793
},
{
"epoch": 7.21,
"grad_norm": 0.2239394634962082,
"learning_rate": 1.1415215697193273e-05,
"loss": 0.0008,
"step": 3794
},
{
"epoch": 7.21,
"grad_norm": 0.43650028109550476,
"learning_rate": 1.1400718098960995e-05,
"loss": 0.0014,
"step": 3795
},
{
"epoch": 7.22,
"grad_norm": 0.04457372426986694,
"learning_rate": 1.13862275527725e-05,
"loss": 0.0003,
"step": 3796
},
{
"epoch": 7.22,
"grad_norm": 0.29227593541145325,
"learning_rate": 1.1371744064121977e-05,
"loss": 0.0012,
"step": 3797
},
{
"epoch": 7.22,
"grad_norm": 0.5221527814865112,
"learning_rate": 1.1357267638500937e-05,
"loss": 0.0015,
"step": 3798
},
{
"epoch": 7.22,
"grad_norm": 0.22518222033977509,
"learning_rate": 1.1342798281398222e-05,
"loss": 0.0029,
"step": 3799
},
{
"epoch": 7.22,
"grad_norm": 0.7775631546974182,
"learning_rate": 1.1328335998299964e-05,
"loss": 0.0032,
"step": 3800
},
{
"epoch": 7.22,
"eval_blimp_filtered_avg": 0.7188059701492537,
"eval_blimp_filtered_std": 0.00497168583117591,
"step": 3800
},
{
"epoch": 7.22,
"eval_blimp_supplement_avg": 0.7844827586206896,
"eval_blimp_supplement_std": 0.018065765087361258,
"step": 3800
},
{
"epoch": 7.22,
"eval_vqa_filtered_avg": 0.34,
"eval_vqa_filtered_std": 0.04760952285695235,
"step": 3800
},
{
"epoch": 7.22,
"eval_winoground_filtered_avg": 0.5,
"eval_winoground_filtered_std": 0.050251890762960605,
"step": 3800
},
{
"epoch": 7.23,
"grad_norm": 0.29893484711647034,
"learning_rate": 1.1313880794689681e-05,
"loss": 0.0013,
"step": 3801
},
{
"epoch": 7.23,
"grad_norm": 0.15125711262226105,
"learning_rate": 1.1299432676048112e-05,
"loss": 0.0006,
"step": 3802
},
{
"epoch": 7.23,
"grad_norm": 0.11963362246751785,
"learning_rate": 1.1284991647853399e-05,
"loss": 0.0004,
"step": 3803
},
{
"epoch": 7.23,
"grad_norm": 0.6380123496055603,
"learning_rate": 1.1270557715580919e-05,
"loss": 0.0036,
"step": 3804
},
{
"epoch": 7.23,
"grad_norm": 0.28375884890556335,
"learning_rate": 1.1256130884703422e-05,
"loss": 0.0013,
"step": 3805
},
{
"epoch": 7.24,
"grad_norm": 0.1797834187746048,
"learning_rate": 1.1241711160690933e-05,
"loss": 0.0018,
"step": 3806
},
{
"epoch": 7.24,
"grad_norm": 0.13901770114898682,
"learning_rate": 1.1227298549010783e-05,
"loss": 0.0033,
"step": 3807
},
{
"epoch": 7.24,
"grad_norm": 0.0519222617149353,
"learning_rate": 1.1212893055127616e-05,
"loss": 0.0006,
"step": 3808
},
{
"epoch": 7.24,
"grad_norm": 0.05914094299077988,
"learning_rate": 1.1198494684503369e-05,
"loss": 0.0004,
"step": 3809
},
{
"epoch": 7.24,
"grad_norm": 0.0927082970738411,
"learning_rate": 1.1184103442597288e-05,
"loss": 0.0003,
"step": 3810
},
{
"epoch": 7.25,
"grad_norm": 0.2479933351278305,
"learning_rate": 1.1169719334865891e-05,
"loss": 0.0011,
"step": 3811
},
{
"epoch": 7.25,
"grad_norm": 0.22383713722229004,
"learning_rate": 1.1155342366763047e-05,
"loss": 0.0021,
"step": 3812
},
{
"epoch": 7.25,
"grad_norm": 0.19257281720638275,
"learning_rate": 1.1140972543739836e-05,
"loss": 0.0014,
"step": 3813
},
{
"epoch": 7.25,
"grad_norm": 0.24553710222244263,
"learning_rate": 1.1126609871244706e-05,
"loss": 0.0058,
"step": 3814
},
{
"epoch": 7.25,
"grad_norm": 0.8123114705085754,
"learning_rate": 1.1112254354723352e-05,
"loss": 0.0028,
"step": 3815
},
{
"epoch": 7.25,
"grad_norm": 0.22939364612102509,
"learning_rate": 1.1097905999618765e-05,
"loss": 0.0004,
"step": 3816
},
{
"epoch": 7.26,
"grad_norm": 0.8987175226211548,
"learning_rate": 1.1083564811371222e-05,
"loss": 0.007,
"step": 3817
},
{
"epoch": 7.26,
"grad_norm": 0.10716027766466141,
"learning_rate": 1.106923079541828e-05,
"loss": 0.0016,
"step": 3818
},
{
"epoch": 7.26,
"grad_norm": 0.07960526645183563,
"learning_rate": 1.1054903957194779e-05,
"loss": 0.0009,
"step": 3819
},
{
"epoch": 7.26,
"grad_norm": 0.37620437145233154,
"learning_rate": 1.1040584302132827e-05,
"loss": 0.003,
"step": 3820
},
{
"epoch": 7.26,
"grad_norm": 0.19653983414173126,
"learning_rate": 1.1026271835661852e-05,
"loss": 0.0009,
"step": 3821
},
{
"epoch": 7.27,
"grad_norm": 0.03111845813691616,
"learning_rate": 1.1011966563208474e-05,
"loss": 0.0002,
"step": 3822
},
{
"epoch": 7.27,
"grad_norm": 0.11719050258398056,
"learning_rate": 1.099766849019667e-05,
"loss": 0.0007,
"step": 3823
},
{
"epoch": 7.27,
"grad_norm": 0.28856122493743896,
"learning_rate": 1.0983377622047641e-05,
"loss": 0.0031,
"step": 3824
},
{
"epoch": 7.27,
"grad_norm": 0.2670513093471527,
"learning_rate": 1.0969093964179864e-05,
"loss": 0.0007,
"step": 3825
},
{
"epoch": 7.27,
"grad_norm": 0.20214910805225372,
"learning_rate": 1.0954817522009086e-05,
"loss": 0.0016,
"step": 3826
},
{
"epoch": 7.28,
"grad_norm": 0.3790026307106018,
"learning_rate": 1.0940548300948319e-05,
"loss": 0.0017,
"step": 3827
},
{
"epoch": 7.28,
"grad_norm": 0.164930060505867,
"learning_rate": 1.0926286306407822e-05,
"loss": 0.0024,
"step": 3828
},
{
"epoch": 7.28,
"grad_norm": 0.05598176643252373,
"learning_rate": 1.0912031543795157e-05,
"loss": 0.0006,
"step": 3829
},
{
"epoch": 7.28,
"grad_norm": 0.14985334873199463,
"learning_rate": 1.0897784018515075e-05,
"loss": 0.0011,
"step": 3830
},
{
"epoch": 7.28,
"grad_norm": 0.040562909096479416,
"learning_rate": 1.0883543735969653e-05,
"loss": 0.0003,
"step": 3831
},
{
"epoch": 7.29,
"grad_norm": 0.194222092628479,
"learning_rate": 1.0869310701558181e-05,
"loss": 0.0005,
"step": 3832
},
{
"epoch": 7.29,
"grad_norm": 0.1385013908147812,
"learning_rate": 1.0855084920677204e-05,
"loss": 0.0013,
"step": 3833
},
{
"epoch": 7.29,
"grad_norm": 0.6989863514900208,
"learning_rate": 1.084086639872053e-05,
"loss": 0.0027,
"step": 3834
},
{
"epoch": 7.29,
"grad_norm": 0.2491830289363861,
"learning_rate": 1.0826655141079203e-05,
"loss": 0.0006,
"step": 3835
},
{
"epoch": 7.29,
"grad_norm": 0.3277353048324585,
"learning_rate": 1.0812451153141524e-05,
"loss": 0.0035,
"step": 3836
},
{
"epoch": 7.29,
"grad_norm": 0.20804928243160248,
"learning_rate": 1.0798254440293019e-05,
"loss": 0.0015,
"step": 3837
},
{
"epoch": 7.3,
"grad_norm": 0.21407996118068695,
"learning_rate": 1.0784065007916492e-05,
"loss": 0.0011,
"step": 3838
},
{
"epoch": 7.3,
"grad_norm": 0.03192751109600067,
"learning_rate": 1.076988286139193e-05,
"loss": 0.0002,
"step": 3839
},
{
"epoch": 7.3,
"grad_norm": 0.24103425443172455,
"learning_rate": 1.0755708006096616e-05,
"loss": 0.0033,
"step": 3840
},
{
"epoch": 7.3,
"grad_norm": 0.8435620069503784,
"learning_rate": 1.0741540447405032e-05,
"loss": 0.0034,
"step": 3841
},
{
"epoch": 7.3,
"grad_norm": 0.10904035717248917,
"learning_rate": 1.072738019068891e-05,
"loss": 0.0012,
"step": 3842
},
{
"epoch": 7.31,
"grad_norm": 0.041802823543548584,
"learning_rate": 1.0713227241317196e-05,
"loss": 0.0002,
"step": 3843
},
{
"epoch": 7.31,
"grad_norm": 0.07222025096416473,
"learning_rate": 1.0699081604656088e-05,
"loss": 0.0005,
"step": 3844
},
{
"epoch": 7.31,
"grad_norm": 0.20888544619083405,
"learning_rate": 1.0684943286068994e-05,
"loss": 0.0012,
"step": 3845
},
{
"epoch": 7.31,
"grad_norm": 0.11883539706468582,
"learning_rate": 1.0670812290916545e-05,
"loss": 0.0005,
"step": 3846
},
{
"epoch": 7.31,
"grad_norm": 0.09083852916955948,
"learning_rate": 1.0656688624556633e-05,
"loss": 0.001,
"step": 3847
},
{
"epoch": 7.32,
"grad_norm": 0.20709002017974854,
"learning_rate": 1.0642572292344301e-05,
"loss": 0.0006,
"step": 3848
},
{
"epoch": 7.32,
"grad_norm": 0.3662835955619812,
"learning_rate": 1.0628463299631885e-05,
"loss": 0.0017,
"step": 3849
},
{
"epoch": 7.32,
"grad_norm": 0.26601678133010864,
"learning_rate": 1.0614361651768893e-05,
"loss": 0.0049,
"step": 3850
},
{
"epoch": 7.32,
"grad_norm": 0.3469623625278473,
"learning_rate": 1.060026735410206e-05,
"loss": 0.0015,
"step": 3851
},
{
"epoch": 7.32,
"grad_norm": 0.44047239422798157,
"learning_rate": 1.0586180411975335e-05,
"loss": 0.0024,
"step": 3852
},
{
"epoch": 7.33,
"grad_norm": 0.10921729356050491,
"learning_rate": 1.0572100830729877e-05,
"loss": 0.0006,
"step": 3853
},
{
"epoch": 7.33,
"grad_norm": 0.17980293929576874,
"learning_rate": 1.0558028615704056e-05,
"loss": 0.0031,
"step": 3854
},
{
"epoch": 7.33,
"grad_norm": 0.22856387495994568,
"learning_rate": 1.0543963772233448e-05,
"loss": 0.0018,
"step": 3855
},
{
"epoch": 7.33,
"grad_norm": 0.45484474301338196,
"learning_rate": 1.0529906305650833e-05,
"loss": 0.0016,
"step": 3856
},
{
"epoch": 7.33,
"grad_norm": 0.3045208752155304,
"learning_rate": 1.0515856221286184e-05,
"loss": 0.0012,
"step": 3857
},
{
"epoch": 7.33,
"grad_norm": 0.4190689027309418,
"learning_rate": 1.0501813524466715e-05,
"loss": 0.0033,
"step": 3858
},
{
"epoch": 7.34,
"grad_norm": 0.44263750314712524,
"learning_rate": 1.0487778220516769e-05,
"loss": 0.0076,
"step": 3859
},
{
"epoch": 7.34,
"grad_norm": 0.5154221057891846,
"learning_rate": 1.0473750314757964e-05,
"loss": 0.0025,
"step": 3860
},
{
"epoch": 7.34,
"grad_norm": 0.1813124418258667,
"learning_rate": 1.0459729812509057e-05,
"loss": 0.0003,
"step": 3861
},
{
"epoch": 7.34,
"grad_norm": 0.05971493571996689,
"learning_rate": 1.044571671908602e-05,
"loss": 0.0008,
"step": 3862
},
{
"epoch": 7.34,
"grad_norm": 0.2584426999092102,
"learning_rate": 1.0431711039802011e-05,
"loss": 0.0013,
"step": 3863
},
{
"epoch": 7.35,
"grad_norm": 0.05983912944793701,
"learning_rate": 1.0417712779967377e-05,
"loss": 0.0004,
"step": 3864
},
{
"epoch": 7.35,
"grad_norm": 0.3702884912490845,
"learning_rate": 1.0403721944889656e-05,
"loss": 0.0031,
"step": 3865
},
{
"epoch": 7.35,
"grad_norm": 0.22895483672618866,
"learning_rate": 1.038973853987356e-05,
"loss": 0.0007,
"step": 3866
},
{
"epoch": 7.35,
"grad_norm": 0.02468941919505596,
"learning_rate": 1.0375762570221016e-05,
"loss": 0.0002,
"step": 3867
},
{
"epoch": 7.35,
"grad_norm": 0.5275208353996277,
"learning_rate": 1.0361794041231072e-05,
"loss": 0.0037,
"step": 3868
},
{
"epoch": 7.36,
"grad_norm": 0.029183542355895042,
"learning_rate": 1.0347832958200022e-05,
"loss": 0.0002,
"step": 3869
},
{
"epoch": 7.36,
"grad_norm": 0.2583842873573303,
"learning_rate": 1.0333879326421285e-05,
"loss": 0.0084,
"step": 3870
},
{
"epoch": 7.36,
"grad_norm": 0.4344351291656494,
"learning_rate": 1.0319933151185486e-05,
"loss": 0.0027,
"step": 3871
},
{
"epoch": 7.36,
"grad_norm": 0.06245533749461174,
"learning_rate": 1.03059944377804e-05,
"loss": 0.0005,
"step": 3872
},
{
"epoch": 7.36,
"grad_norm": 0.5742080211639404,
"learning_rate": 1.029206319149101e-05,
"loss": 0.0019,
"step": 3873
},
{
"epoch": 7.37,
"grad_norm": 0.03434763103723526,
"learning_rate": 1.0278139417599402e-05,
"loss": 0.0001,
"step": 3874
},
{
"epoch": 7.37,
"grad_norm": 1.1970933675765991,
"learning_rate": 1.0264223121384917e-05,
"loss": 0.014,
"step": 3875
},
{
"epoch": 7.37,
"grad_norm": 0.13805250823497772,
"learning_rate": 1.0250314308123968e-05,
"loss": 0.0009,
"step": 3876
},
{
"epoch": 7.37,
"grad_norm": 0.19729770720005035,
"learning_rate": 1.0236412983090205e-05,
"loss": 0.0007,
"step": 3877
},
{
"epoch": 7.37,
"grad_norm": 0.07012030482292175,
"learning_rate": 1.0222519151554402e-05,
"loss": 0.0006,
"step": 3878
},
{
"epoch": 7.37,
"grad_norm": 0.08937878906726837,
"learning_rate": 1.0208632818784495e-05,
"loss": 0.0015,
"step": 3879
},
{
"epoch": 7.38,
"grad_norm": 0.5024104118347168,
"learning_rate": 1.019475399004559e-05,
"loss": 0.0024,
"step": 3880
},
{
"epoch": 7.38,
"grad_norm": 0.44746169447898865,
"learning_rate": 1.018088267059993e-05,
"loss": 0.0013,
"step": 3881
},
{
"epoch": 7.38,
"grad_norm": 0.672454833984375,
"learning_rate": 1.0167018865706922e-05,
"loss": 0.0024,
"step": 3882
},
{
"epoch": 7.38,
"grad_norm": 0.05334677919745445,
"learning_rate": 1.0153162580623112e-05,
"loss": 0.0008,
"step": 3883
},
{
"epoch": 7.38,
"grad_norm": 0.08294852823019028,
"learning_rate": 1.0139313820602238e-05,
"loss": 0.0007,
"step": 3884
},
{
"epoch": 7.39,
"grad_norm": 0.03401881456375122,
"learning_rate": 1.0125472590895109e-05,
"loss": 0.0004,
"step": 3885
},
{
"epoch": 7.39,
"grad_norm": 0.626732587814331,
"learning_rate": 1.0111638896749748e-05,
"loss": 0.0071,
"step": 3886
},
{
"epoch": 7.39,
"grad_norm": 0.2696470618247986,
"learning_rate": 1.0097812743411285e-05,
"loss": 0.0028,
"step": 3887
},
{
"epoch": 7.39,
"grad_norm": 0.04598533362150192,
"learning_rate": 1.0083994136122002e-05,
"loss": 0.0003,
"step": 3888
},
{
"epoch": 7.39,
"grad_norm": 0.20212337374687195,
"learning_rate": 1.007018308012132e-05,
"loss": 0.0018,
"step": 3889
},
{
"epoch": 7.4,
"grad_norm": 0.04220814257860184,
"learning_rate": 1.0056379580645786e-05,
"loss": 0.0005,
"step": 3890
},
{
"epoch": 7.4,
"grad_norm": 0.3461586833000183,
"learning_rate": 1.00425836429291e-05,
"loss": 0.0007,
"step": 3891
},
{
"epoch": 7.4,
"grad_norm": 0.5738538503646851,
"learning_rate": 1.0028795272202067e-05,
"loss": 0.0098,
"step": 3892
},
{
"epoch": 7.4,
"grad_norm": 0.03577322140336037,
"learning_rate": 1.0015014473692677e-05,
"loss": 0.0002,
"step": 3893
},
{
"epoch": 7.4,
"grad_norm": 0.04260101169347763,
"learning_rate": 1.0001241252625971e-05,
"loss": 0.0002,
"step": 3894
},
{
"epoch": 7.4,
"grad_norm": 0.03839372098445892,
"learning_rate": 9.987475614224187e-06,
"loss": 0.0002,
"step": 3895
},
{
"epoch": 7.41,
"grad_norm": 0.3242708444595337,
"learning_rate": 9.973717563706654e-06,
"loss": 0.001,
"step": 3896
},
{
"epoch": 7.41,
"grad_norm": 0.19412927329540253,
"learning_rate": 9.95996710628983e-06,
"loss": 0.0009,
"step": 3897
},
{
"epoch": 7.41,
"grad_norm": 0.5900408625602722,
"learning_rate": 9.94622424718729e-06,
"loss": 0.0021,
"step": 3898
},
{
"epoch": 7.41,
"grad_norm": 0.07700662314891815,
"learning_rate": 9.932488991609736e-06,
"loss": 0.0011,
"step": 3899
},
{
"epoch": 7.41,
"grad_norm": 0.1705576777458191,
"learning_rate": 9.91876134476498e-06,
"loss": 0.0025,
"step": 3900
},
{
"epoch": 7.41,
"eval_blimp_filtered_avg": 0.7185074626865672,
"eval_blimp_filtered_std": 0.004982491153796648,
"step": 3900
},
{
"epoch": 7.41,
"eval_blimp_supplement_avg": 0.7780172413793104,
"eval_blimp_supplement_std": 0.018106005392130545,
"step": 3900
},
{
"epoch": 7.41,
"eval_vqa_filtered_avg": 0.33,
"eval_vqa_filtered_std": 0.04725815626252604,
"step": 3900
},
{
"epoch": 7.41,
"eval_winoground_filtered_avg": 0.5,
"eval_winoground_filtered_std": 0.050251890762960605,
"step": 3900
},
{
"epoch": 7.42,
"grad_norm": 0.48269471526145935,
"learning_rate": 9.905041311857956e-06,
"loss": 0.0033,
"step": 3901
},
{
"epoch": 7.42,
"grad_norm": 0.06087570637464523,
"learning_rate": 9.891328898090705e-06,
"loss": 0.001,
"step": 3902
},
{
"epoch": 7.42,
"grad_norm": 0.4798135459423065,
"learning_rate": 9.877624108662371e-06,
"loss": 0.0009,
"step": 3903
},
{
"epoch": 7.42,
"grad_norm": 0.054261885583400726,
"learning_rate": 9.863926948769251e-06,
"loss": 0.0005,
"step": 3904
},
{
"epoch": 7.42,
"grad_norm": 0.4724716246128082,
"learning_rate": 9.850237423604677e-06,
"loss": 0.0025,
"step": 3905
},
{
"epoch": 7.43,
"grad_norm": 0.05672892928123474,
"learning_rate": 9.836555538359147e-06,
"loss": 0.0003,
"step": 3906
},
{
"epoch": 7.43,
"grad_norm": 0.10598628222942352,
"learning_rate": 9.822881298220238e-06,
"loss": 0.0006,
"step": 3907
},
{
"epoch": 7.43,
"grad_norm": 0.4372518062591553,
"learning_rate": 9.80921470837263e-06,
"loss": 0.0038,
"step": 3908
},
{
"epoch": 7.43,
"grad_norm": 0.15734049677848816,
"learning_rate": 9.7955557739981e-06,
"loss": 0.0007,
"step": 3909
},
{
"epoch": 7.43,
"grad_norm": 0.33495476841926575,
"learning_rate": 9.781904500275527e-06,
"loss": 0.0017,
"step": 3910
},
{
"epoch": 7.44,
"grad_norm": 0.015100725926458836,
"learning_rate": 9.768260892380885e-06,
"loss": 0.0001,
"step": 3911
},
{
"epoch": 7.44,
"grad_norm": 0.13718558847904205,
"learning_rate": 9.754624955487233e-06,
"loss": 0.0009,
"step": 3912
},
{
"epoch": 7.44,
"grad_norm": 0.21440330147743225,
"learning_rate": 9.740996694764751e-06,
"loss": 0.0023,
"step": 3913
},
{
"epoch": 7.44,
"grad_norm": 0.027186062186956406,
"learning_rate": 9.727376115380652e-06,
"loss": 0.0003,
"step": 3914
},
{
"epoch": 7.44,
"grad_norm": 0.05432227998971939,
"learning_rate": 9.7137632224993e-06,
"loss": 0.0004,
"step": 3915
},
{
"epoch": 7.44,
"grad_norm": 0.1651182919740677,
"learning_rate": 9.700158021282107e-06,
"loss": 0.0006,
"step": 3916
},
{
"epoch": 7.45,
"grad_norm": 0.2806062698364258,
"learning_rate": 9.686560516887577e-06,
"loss": 0.001,
"step": 3917
},
{
"epoch": 7.45,
"grad_norm": 0.18145093321800232,
"learning_rate": 9.672970714471281e-06,
"loss": 0.0007,
"step": 3918
},
{
"epoch": 7.45,
"grad_norm": 0.1783367097377777,
"learning_rate": 9.659388619185921e-06,
"loss": 0.001,
"step": 3919
},
{
"epoch": 7.45,
"grad_norm": 0.6045911312103271,
"learning_rate": 9.645814236181197e-06,
"loss": 0.0048,
"step": 3920
},
{
"epoch": 7.45,
"grad_norm": 0.25310537219047546,
"learning_rate": 9.63224757060396e-06,
"loss": 0.0048,
"step": 3921
},
{
"epoch": 7.46,
"grad_norm": 0.31694209575653076,
"learning_rate": 9.618688627598092e-06,
"loss": 0.0006,
"step": 3922
},
{
"epoch": 7.46,
"grad_norm": 0.10906882584095001,
"learning_rate": 9.605137412304556e-06,
"loss": 0.0011,
"step": 3923
},
{
"epoch": 7.46,
"grad_norm": 0.6736571192741394,
"learning_rate": 9.591593929861405e-06,
"loss": 0.0024,
"step": 3924
},
{
"epoch": 7.46,
"grad_norm": 0.1853550523519516,
"learning_rate": 9.578058185403704e-06,
"loss": 0.0012,
"step": 3925
},
{
"epoch": 7.46,
"grad_norm": 0.6082493662834167,
"learning_rate": 9.564530184063654e-06,
"loss": 0.0015,
"step": 3926
},
{
"epoch": 7.47,
"grad_norm": 0.06418062746524811,
"learning_rate": 9.55100993097048e-06,
"loss": 0.0003,
"step": 3927
},
{
"epoch": 7.47,
"grad_norm": 0.17114675045013428,
"learning_rate": 9.537497431250479e-06,
"loss": 0.0017,
"step": 3928
},
{
"epoch": 7.47,
"grad_norm": 0.1492406278848648,
"learning_rate": 9.523992690026993e-06,
"loss": 0.0013,
"step": 3929
},
{
"epoch": 7.47,
"grad_norm": 0.13354071974754333,
"learning_rate": 9.51049571242047e-06,
"loss": 0.0015,
"step": 3930
},
{
"epoch": 7.47,
"grad_norm": 0.08157804608345032,
"learning_rate": 9.497006503548336e-06,
"loss": 0.0005,
"step": 3931
},
{
"epoch": 7.48,
"grad_norm": 0.10767817497253418,
"learning_rate": 9.48352506852515e-06,
"loss": 0.001,
"step": 3932
},
{
"epoch": 7.48,
"grad_norm": 0.1292194426059723,
"learning_rate": 9.47005141246248e-06,
"loss": 0.0006,
"step": 3933
},
{
"epoch": 7.48,
"grad_norm": 0.056460894644260406,
"learning_rate": 9.45658554046895e-06,
"loss": 0.0007,
"step": 3934
},
{
"epoch": 7.48,
"grad_norm": 0.021156983450055122,
"learning_rate": 9.44312745765024e-06,
"loss": 0.0001,
"step": 3935
},
{
"epoch": 7.48,
"grad_norm": 0.22647006809711456,
"learning_rate": 9.429677169109072e-06,
"loss": 0.0029,
"step": 3936
},
{
"epoch": 7.48,
"grad_norm": 0.13638705015182495,
"learning_rate": 9.41623467994521e-06,
"loss": 0.0027,
"step": 3937
},
{
"epoch": 7.49,
"grad_norm": 0.19083188474178314,
"learning_rate": 9.402799995255461e-06,
"loss": 0.0045,
"step": 3938
},
{
"epoch": 7.49,
"grad_norm": 0.5341401696205139,
"learning_rate": 9.389373120133705e-06,
"loss": 0.0025,
"step": 3939
},
{
"epoch": 7.49,
"grad_norm": 0.012964884750545025,
"learning_rate": 9.375954059670797e-06,
"loss": 0.0001,
"step": 3940
},
{
"epoch": 7.49,
"grad_norm": 0.039807166904211044,
"learning_rate": 9.362542818954683e-06,
"loss": 0.0004,
"step": 3941
},
{
"epoch": 7.49,
"grad_norm": 0.6944048404693604,
"learning_rate": 9.349139403070326e-06,
"loss": 0.0016,
"step": 3942
},
{
"epoch": 7.5,
"grad_norm": 0.3766193091869354,
"learning_rate": 9.335743817099718e-06,
"loss": 0.0032,
"step": 3943
},
{
"epoch": 7.5,
"grad_norm": 0.06944786012172699,
"learning_rate": 9.32235606612189e-06,
"loss": 0.0004,
"step": 3944
},
{
"epoch": 7.5,
"grad_norm": 0.2732830345630646,
"learning_rate": 9.308976155212893e-06,
"loss": 0.0012,
"step": 3945
},
{
"epoch": 7.5,
"grad_norm": 0.5199292898178101,
"learning_rate": 9.295604089445812e-06,
"loss": 0.0009,
"step": 3946
},
{
"epoch": 7.5,
"grad_norm": 0.45663923025131226,
"learning_rate": 9.282239873890752e-06,
"loss": 0.0013,
"step": 3947
},
{
"epoch": 7.51,
"grad_norm": 0.33895257115364075,
"learning_rate": 9.268883513614867e-06,
"loss": 0.0029,
"step": 3948
},
{
"epoch": 7.51,
"grad_norm": 0.19757308065891266,
"learning_rate": 9.25553501368228e-06,
"loss": 0.002,
"step": 3949
},
{
"epoch": 7.51,
"grad_norm": 0.5578463077545166,
"learning_rate": 9.2421943791542e-06,
"loss": 0.0039,
"step": 3950
},
{
"epoch": 7.51,
"grad_norm": 0.12307825684547424,
"learning_rate": 9.22886161508878e-06,
"loss": 0.0005,
"step": 3951
},
{
"epoch": 7.51,
"grad_norm": 0.12373200058937073,
"learning_rate": 9.215536726541259e-06,
"loss": 0.0007,
"step": 3952
},
{
"epoch": 7.52,
"grad_norm": 0.21534214913845062,
"learning_rate": 9.20221971856385e-06,
"loss": 0.0011,
"step": 3953
},
{
"epoch": 7.52,
"grad_norm": 0.1034073755145073,
"learning_rate": 9.18891059620578e-06,
"loss": 0.0005,
"step": 3954
},
{
"epoch": 7.52,
"grad_norm": 1.099151611328125,
"learning_rate": 9.1756093645133e-06,
"loss": 0.0026,
"step": 3955
},
{
"epoch": 7.52,
"grad_norm": 0.30171456933021545,
"learning_rate": 9.162316028529663e-06,
"loss": 0.0032,
"step": 3956
},
{
"epoch": 7.52,
"grad_norm": 0.03533285856246948,
"learning_rate": 9.149030593295123e-06,
"loss": 0.0003,
"step": 3957
},
{
"epoch": 7.52,
"grad_norm": 0.50489741563797,
"learning_rate": 9.13575306384693e-06,
"loss": 0.002,
"step": 3958
},
{
"epoch": 7.53,
"grad_norm": 0.18265804648399353,
"learning_rate": 9.122483445219389e-06,
"loss": 0.0032,
"step": 3959
},
{
"epoch": 7.53,
"grad_norm": 0.06352107226848602,
"learning_rate": 9.10922174244372e-06,
"loss": 0.0004,
"step": 3960
},
{
"epoch": 7.53,
"grad_norm": 0.1388656497001648,
"learning_rate": 9.095967960548215e-06,
"loss": 0.0019,
"step": 3961
},
{
"epoch": 7.53,
"grad_norm": 0.589281439781189,
"learning_rate": 9.082722104558127e-06,
"loss": 0.0029,
"step": 3962
},
{
"epoch": 7.53,
"grad_norm": 0.22214657068252563,
"learning_rate": 9.069484179495717e-06,
"loss": 0.0065,
"step": 3963
},
{
"epoch": 7.54,
"grad_norm": 0.055167682468891144,
"learning_rate": 9.05625419038022e-06,
"loss": 0.0008,
"step": 3964
},
{
"epoch": 7.54,
"grad_norm": 0.057885196059942245,
"learning_rate": 9.043032142227903e-06,
"loss": 0.0003,
"step": 3965
},
{
"epoch": 7.54,
"grad_norm": 0.10997253656387329,
"learning_rate": 9.029818040051958e-06,
"loss": 0.0004,
"step": 3966
},
{
"epoch": 7.54,
"grad_norm": 0.032723620533943176,
"learning_rate": 9.016611888862639e-06,
"loss": 0.0002,
"step": 3967
},
{
"epoch": 7.54,
"grad_norm": 0.16148625314235687,
"learning_rate": 9.003413693667124e-06,
"loss": 0.0019,
"step": 3968
},
{
"epoch": 7.55,
"grad_norm": 0.110807865858078,
"learning_rate": 8.990223459469604e-06,
"loss": 0.001,
"step": 3969
},
{
"epoch": 7.55,
"grad_norm": 0.2213355153799057,
"learning_rate": 8.977041191271249e-06,
"loss": 0.0011,
"step": 3970
},
{
"epoch": 7.55,
"grad_norm": 0.2399233877658844,
"learning_rate": 8.9638668940702e-06,
"loss": 0.0016,
"step": 3971
},
{
"epoch": 7.55,
"grad_norm": 0.41127002239227295,
"learning_rate": 8.950700572861586e-06,
"loss": 0.0046,
"step": 3972
},
{
"epoch": 7.55,
"grad_norm": 0.03968670964241028,
"learning_rate": 8.937542232637506e-06,
"loss": 0.0004,
"step": 3973
},
{
"epoch": 7.56,
"grad_norm": 0.7982978224754333,
"learning_rate": 8.92439187838703e-06,
"loss": 0.0079,
"step": 3974
},
{
"epoch": 7.56,
"grad_norm": 0.1991325169801712,
"learning_rate": 8.911249515096196e-06,
"loss": 0.0005,
"step": 3975
},
{
"epoch": 7.56,
"grad_norm": 0.10407639294862747,
"learning_rate": 8.898115147748053e-06,
"loss": 0.0015,
"step": 3976
},
{
"epoch": 7.56,
"grad_norm": 0.5840522646903992,
"learning_rate": 8.884988781322541e-06,
"loss": 0.0148,
"step": 3977
},
{
"epoch": 7.56,
"grad_norm": 0.12891338765621185,
"learning_rate": 8.871870420796644e-06,
"loss": 0.0005,
"step": 3978
},
{
"epoch": 7.56,
"grad_norm": 0.16564643383026123,
"learning_rate": 8.858760071144264e-06,
"loss": 0.0027,
"step": 3979
},
{
"epoch": 7.57,
"grad_norm": 0.6031332612037659,
"learning_rate": 8.845657737336285e-06,
"loss": 0.007,
"step": 3980
},
{
"epoch": 7.57,
"grad_norm": 0.31511518359184265,
"learning_rate": 8.832563424340536e-06,
"loss": 0.0046,
"step": 3981
},
{
"epoch": 7.57,
"grad_norm": 0.45697250962257385,
"learning_rate": 8.819477137121825e-06,
"loss": 0.0083,
"step": 3982
},
{
"epoch": 7.57,
"grad_norm": 0.21940094232559204,
"learning_rate": 8.806398880641899e-06,
"loss": 0.0021,
"step": 3983
},
{
"epoch": 7.57,
"grad_norm": 0.06017604470252991,
"learning_rate": 8.793328659859462e-06,
"loss": 0.0003,
"step": 3984
},
{
"epoch": 7.58,
"grad_norm": 0.26314905285835266,
"learning_rate": 8.780266479730201e-06,
"loss": 0.0074,
"step": 3985
},
{
"epoch": 7.58,
"grad_norm": 0.026405276730656624,
"learning_rate": 8.767212345206698e-06,
"loss": 0.0003,
"step": 3986
},
{
"epoch": 7.58,
"grad_norm": 0.08451911807060242,
"learning_rate": 8.754166261238543e-06,
"loss": 0.0004,
"step": 3987
},
{
"epoch": 7.58,
"grad_norm": 0.09489631652832031,
"learning_rate": 8.741128232772235e-06,
"loss": 0.0006,
"step": 3988
},
{
"epoch": 7.58,
"grad_norm": 0.5388056039810181,
"learning_rate": 8.72809826475123e-06,
"loss": 0.0038,
"step": 3989
},
{
"epoch": 7.59,
"grad_norm": 0.08892013877630234,
"learning_rate": 8.715076362115936e-06,
"loss": 0.0016,
"step": 3990
},
{
"epoch": 7.59,
"grad_norm": 0.030492683872580528,
"learning_rate": 8.70206252980369e-06,
"loss": 0.0003,
"step": 3991
},
{
"epoch": 7.59,
"grad_norm": 0.17382140457630157,
"learning_rate": 8.689056772748775e-06,
"loss": 0.0009,
"step": 3992
},
{
"epoch": 7.59,
"grad_norm": 0.9870418310165405,
"learning_rate": 8.676059095882401e-06,
"loss": 0.0017,
"step": 3993
},
{
"epoch": 7.59,
"grad_norm": 0.5269887447357178,
"learning_rate": 8.663069504132758e-06,
"loss": 0.0017,
"step": 3994
},
{
"epoch": 7.6,
"grad_norm": 0.11195862293243408,
"learning_rate": 8.6500880024249e-06,
"loss": 0.0008,
"step": 3995
},
{
"epoch": 7.6,
"grad_norm": 0.3082568049430847,
"learning_rate": 8.637114595680876e-06,
"loss": 0.001,
"step": 3996
},
{
"epoch": 7.6,
"grad_norm": 0.06805071979761124,
"learning_rate": 8.624149288819629e-06,
"loss": 0.0012,
"step": 3997
},
{
"epoch": 7.6,
"grad_norm": 0.22789660096168518,
"learning_rate": 8.611192086757051e-06,
"loss": 0.0015,
"step": 3998
},
{
"epoch": 7.6,
"grad_norm": 0.20034611225128174,
"learning_rate": 8.59824299440595e-06,
"loss": 0.0019,
"step": 3999
},
{
"epoch": 7.6,
"grad_norm": 0.12724469602108002,
"learning_rate": 8.585302016676059e-06,
"loss": 0.0008,
"step": 4000
},
{
"epoch": 7.6,
"eval_blimp_filtered_avg": 0.718955223880597,
"eval_blimp_filtered_std": 0.00497686536702203,
"step": 4000
},
{
"epoch": 7.6,
"eval_blimp_supplement_avg": 0.7758620689655172,
"eval_blimp_supplement_std": 0.01811248739120767,
"step": 4000
},
{
"epoch": 7.6,
"eval_vqa_filtered_avg": 0.31,
"eval_vqa_filtered_std": 0.04648231987117316,
"step": 4000
},
{
"epoch": 7.6,
"eval_winoground_filtered_avg": 0.5,
"eval_winoground_filtered_std": 0.050251890762960605,
"step": 4000
},
{
"epoch": 7.61,
"grad_norm": 0.05688934028148651,
"learning_rate": 8.572369158474042e-06,
"loss": 0.0004,
"step": 4001
},
{
"epoch": 7.61,
"grad_norm": 0.46273767948150635,
"learning_rate": 8.559444424703472e-06,
"loss": 0.0063,
"step": 4002
},
{
"epoch": 7.61,
"grad_norm": 0.09548939764499664,
"learning_rate": 8.54652782026486e-06,
"loss": 0.0005,
"step": 4003
},
{
"epoch": 7.61,
"grad_norm": 0.10774649679660797,
"learning_rate": 8.533619350055608e-06,
"loss": 0.001,
"step": 4004
},
{
"epoch": 7.61,
"grad_norm": 0.25860968232154846,
"learning_rate": 8.520719018970079e-06,
"loss": 0.0035,
"step": 4005
},
{
"epoch": 7.62,
"grad_norm": 0.03977625072002411,
"learning_rate": 8.507826831899483e-06,
"loss": 0.0002,
"step": 4006
},
{
"epoch": 7.62,
"grad_norm": 0.1274503916501999,
"learning_rate": 8.49494279373201e-06,
"loss": 0.0008,
"step": 4007
},
{
"epoch": 7.62,
"grad_norm": 0.20307974517345428,
"learning_rate": 8.482066909352724e-06,
"loss": 0.0007,
"step": 4008
},
{
"epoch": 7.62,
"grad_norm": 0.1084689125418663,
"learning_rate": 8.4691991836436e-06,
"loss": 0.0009,
"step": 4009
},
{
"epoch": 7.62,
"grad_norm": 0.020024720579385757,
"learning_rate": 8.456339621483507e-06,
"loss": 0.0002,
"step": 4010
},
{
"epoch": 7.63,
"grad_norm": 0.10501285642385483,
"learning_rate": 8.443488227748276e-06,
"loss": 0.0012,
"step": 4011
},
{
"epoch": 7.63,
"grad_norm": 0.03906962275505066,
"learning_rate": 8.430645007310557e-06,
"loss": 0.0002,
"step": 4012
},
{
"epoch": 7.63,
"grad_norm": 0.35857272148132324,
"learning_rate": 8.41780996503997e-06,
"loss": 0.0023,
"step": 4013
},
{
"epoch": 7.63,
"grad_norm": 0.9417741894721985,
"learning_rate": 8.404983105803002e-06,
"loss": 0.0035,
"step": 4014
},
{
"epoch": 7.63,
"grad_norm": 0.26643791794776917,
"learning_rate": 8.392164434463045e-06,
"loss": 0.0008,
"step": 4015
},
{
"epoch": 7.63,
"grad_norm": 0.09323861449956894,
"learning_rate": 8.379353955880379e-06,
"loss": 0.0006,
"step": 4016
},
{
"epoch": 7.64,
"grad_norm": 0.19513775408267975,
"learning_rate": 8.366551674912191e-06,
"loss": 0.0027,
"step": 4017
},
{
"epoch": 7.64,
"grad_norm": 0.2567938566207886,
"learning_rate": 8.353757596412546e-06,
"loss": 0.001,
"step": 4018
},
{
"epoch": 7.64,
"grad_norm": 0.1127931997179985,
"learning_rate": 8.3409717252324e-06,
"loss": 0.0006,
"step": 4019
},
{
"epoch": 7.64,
"grad_norm": 0.09068554639816284,
"learning_rate": 8.328194066219633e-06,
"loss": 0.0003,
"step": 4020
},
{
"epoch": 7.64,
"grad_norm": 0.1304280161857605,
"learning_rate": 8.315424624218937e-06,
"loss": 0.0004,
"step": 4021
},
{
"epoch": 7.65,
"grad_norm": 0.13967347145080566,
"learning_rate": 8.302663404071975e-06,
"loss": 0.0008,
"step": 4022
},
{
"epoch": 7.65,
"grad_norm": 0.06981847435235977,
"learning_rate": 8.289910410617215e-06,
"loss": 0.0007,
"step": 4023
},
{
"epoch": 7.65,
"grad_norm": 0.23649661242961884,
"learning_rate": 8.277165648690059e-06,
"loss": 0.0013,
"step": 4024
},
{
"epoch": 7.65,
"grad_norm": 0.04708298295736313,
"learning_rate": 8.264429123122772e-06,
"loss": 0.0003,
"step": 4025
},
{
"epoch": 7.65,
"grad_norm": 0.1221221536397934,
"learning_rate": 8.251700838744485e-06,
"loss": 0.0002,
"step": 4026
},
{
"epoch": 7.66,
"grad_norm": 0.08329764753580093,
"learning_rate": 8.238980800381217e-06,
"loss": 0.0015,
"step": 4027
},
{
"epoch": 7.66,
"grad_norm": 0.20287173986434937,
"learning_rate": 8.226269012855861e-06,
"loss": 0.0013,
"step": 4028
},
{
"epoch": 7.66,
"grad_norm": 0.5095852613449097,
"learning_rate": 8.21356548098817e-06,
"loss": 0.002,
"step": 4029
},
{
"epoch": 7.66,
"grad_norm": 0.1451786309480667,
"learning_rate": 8.20087020959477e-06,
"loss": 0.0007,
"step": 4030
},
{
"epoch": 7.66,
"grad_norm": 0.08433229476213455,
"learning_rate": 8.188183203489184e-06,
"loss": 0.0003,
"step": 4031
},
{
"epoch": 7.67,
"grad_norm": 0.2256261557340622,
"learning_rate": 8.175504467481737e-06,
"loss": 0.0008,
"step": 4032
},
{
"epoch": 7.67,
"grad_norm": 0.09659327566623688,
"learning_rate": 8.162834006379695e-06,
"loss": 0.0007,
"step": 4033
},
{
"epoch": 7.67,
"grad_norm": 0.0701504796743393,
"learning_rate": 8.15017182498713e-06,
"loss": 0.0005,
"step": 4034
},
{
"epoch": 7.67,
"grad_norm": 0.05150916054844856,
"learning_rate": 8.137517928104998e-06,
"loss": 0.0005,
"step": 4035
},
{
"epoch": 7.67,
"grad_norm": 1.1541883945465088,
"learning_rate": 8.12487232053111e-06,
"loss": 0.0064,
"step": 4036
},
{
"epoch": 7.67,
"grad_norm": 0.08933593332767487,
"learning_rate": 8.112235007060132e-06,
"loss": 0.0011,
"step": 4037
},
{
"epoch": 7.68,
"grad_norm": 0.1480831801891327,
"learning_rate": 8.099605992483591e-06,
"loss": 0.002,
"step": 4038
},
{
"epoch": 7.68,
"grad_norm": 0.1149938777089119,
"learning_rate": 8.086985281589852e-06,
"loss": 0.0008,
"step": 4039
},
{
"epoch": 7.68,
"grad_norm": 0.2511685788631439,
"learning_rate": 8.074372879164167e-06,
"loss": 0.0008,
"step": 4040
},
{
"epoch": 7.68,
"grad_norm": 0.07072022557258606,
"learning_rate": 8.061768789988582e-06,
"loss": 0.0003,
"step": 4041
},
{
"epoch": 7.68,
"grad_norm": 0.024631841108202934,
"learning_rate": 8.049173018842047e-06,
"loss": 0.0001,
"step": 4042
},
{
"epoch": 7.69,
"grad_norm": 0.06437274068593979,
"learning_rate": 8.03658557050033e-06,
"loss": 0.0004,
"step": 4043
},
{
"epoch": 7.69,
"grad_norm": 0.32501158118247986,
"learning_rate": 8.024006449736041e-06,
"loss": 0.0011,
"step": 4044
},
{
"epoch": 7.69,
"grad_norm": 0.33693039417266846,
"learning_rate": 8.011435661318634e-06,
"loss": 0.0008,
"step": 4045
},
{
"epoch": 7.69,
"grad_norm": 0.18374291062355042,
"learning_rate": 7.998873210014435e-06,
"loss": 0.0007,
"step": 4046
},
{
"epoch": 7.69,
"grad_norm": 0.2313518226146698,
"learning_rate": 7.98631910058655e-06,
"loss": 0.0022,
"step": 4047
},
{
"epoch": 7.7,
"grad_norm": 0.10124185681343079,
"learning_rate": 7.973773337794977e-06,
"loss": 0.0006,
"step": 4048
},
{
"epoch": 7.7,
"grad_norm": 0.3494603633880615,
"learning_rate": 7.961235926396512e-06,
"loss": 0.001,
"step": 4049
},
{
"epoch": 7.7,
"grad_norm": 0.3213108479976654,
"learning_rate": 7.9487068711448e-06,
"loss": 0.0027,
"step": 4050
},
{
"epoch": 7.7,
"grad_norm": 0.23749741911888123,
"learning_rate": 7.93618617679034e-06,
"loss": 0.001,
"step": 4051
},
{
"epoch": 7.7,
"grad_norm": 0.3156917095184326,
"learning_rate": 7.923673848080405e-06,
"loss": 0.0009,
"step": 4052
},
{
"epoch": 7.71,
"grad_norm": 0.06433945149183273,
"learning_rate": 7.911169889759153e-06,
"loss": 0.0008,
"step": 4053
},
{
"epoch": 7.71,
"grad_norm": 0.09628526866436005,
"learning_rate": 7.89867430656754e-06,
"loss": 0.0012,
"step": 4054
},
{
"epoch": 7.71,
"grad_norm": 0.5612062811851501,
"learning_rate": 7.886187103243344e-06,
"loss": 0.003,
"step": 4055
},
{
"epoch": 7.71,
"grad_norm": 0.13382011651992798,
"learning_rate": 7.87370828452117e-06,
"loss": 0.0009,
"step": 4056
},
{
"epoch": 7.71,
"grad_norm": 0.08450856059789658,
"learning_rate": 7.861237855132474e-06,
"loss": 0.0007,
"step": 4057
},
{
"epoch": 7.71,
"grad_norm": 0.3193015456199646,
"learning_rate": 7.848775819805465e-06,
"loss": 0.0009,
"step": 4058
},
{
"epoch": 7.72,
"grad_norm": 0.09931007772684097,
"learning_rate": 7.83632218326523e-06,
"loss": 0.0009,
"step": 4059
},
{
"epoch": 7.72,
"grad_norm": 0.26327890157699585,
"learning_rate": 7.823876950233652e-06,
"loss": 0.0006,
"step": 4060
},
{
"epoch": 7.72,
"grad_norm": 0.16994498670101166,
"learning_rate": 7.811440125429417e-06,
"loss": 0.0014,
"step": 4061
},
{
"epoch": 7.72,
"grad_norm": 0.4773654043674469,
"learning_rate": 7.799011713568035e-06,
"loss": 0.0008,
"step": 4062
},
{
"epoch": 7.72,
"grad_norm": 0.1689019650220871,
"learning_rate": 7.786591719361824e-06,
"loss": 0.0003,
"step": 4063
},
{
"epoch": 7.73,
"grad_norm": 0.07232107222080231,
"learning_rate": 7.774180147519905e-06,
"loss": 0.0007,
"step": 4064
},
{
"epoch": 7.73,
"grad_norm": 0.0660555511713028,
"learning_rate": 7.761777002748203e-06,
"loss": 0.0004,
"step": 4065
},
{
"epoch": 7.73,
"grad_norm": 0.1710626631975174,
"learning_rate": 7.74938228974948e-06,
"loss": 0.0015,
"step": 4066
},
{
"epoch": 7.73,
"grad_norm": 0.4248620569705963,
"learning_rate": 7.736996013223243e-06,
"loss": 0.0045,
"step": 4067
},
{
"epoch": 7.73,
"grad_norm": 0.008610131219029427,
"learning_rate": 7.724618177865851e-06,
"loss": 0.0001,
"step": 4068
},
{
"epoch": 7.74,
"grad_norm": 0.42559507489204407,
"learning_rate": 7.712248788370446e-06,
"loss": 0.0019,
"step": 4069
},
{
"epoch": 7.74,
"grad_norm": 0.3335561454296112,
"learning_rate": 7.69988784942696e-06,
"loss": 0.0013,
"step": 4070
},
{
"epoch": 7.74,
"grad_norm": 0.14192043244838715,
"learning_rate": 7.687535365722128e-06,
"loss": 0.0012,
"step": 4071
},
{
"epoch": 7.74,
"grad_norm": 0.32461896538734436,
"learning_rate": 7.675191341939474e-06,
"loss": 0.0009,
"step": 4072
},
{
"epoch": 7.74,
"grad_norm": 0.13669994473457336,
"learning_rate": 7.662855782759326e-06,
"loss": 0.0019,
"step": 4073
},
{
"epoch": 7.75,
"grad_norm": 0.17524461448192596,
"learning_rate": 7.650528692858789e-06,
"loss": 0.001,
"step": 4074
},
{
"epoch": 7.75,
"grad_norm": 0.24207572638988495,
"learning_rate": 7.638210076911765e-06,
"loss": 0.0073,
"step": 4075
},
{
"epoch": 7.75,
"grad_norm": 0.03728511929512024,
"learning_rate": 7.625899939588929e-06,
"loss": 0.0004,
"step": 4076
},
{
"epoch": 7.75,
"grad_norm": 0.10254962742328644,
"learning_rate": 7.613598285557787e-06,
"loss": 0.0004,
"step": 4077
},
{
"epoch": 7.75,
"grad_norm": 0.02493688277900219,
"learning_rate": 7.601305119482549e-06,
"loss": 0.0003,
"step": 4078
},
{
"epoch": 7.75,
"grad_norm": 0.3288654386997223,
"learning_rate": 7.589020446024287e-06,
"loss": 0.0012,
"step": 4079
},
{
"epoch": 7.76,
"grad_norm": 0.10999511927366257,
"learning_rate": 7.576744269840803e-06,
"loss": 0.0009,
"step": 4080
},
{
"epoch": 7.76,
"grad_norm": 0.3886186480522156,
"learning_rate": 7.564476595586699e-06,
"loss": 0.0014,
"step": 4081
},
{
"epoch": 7.76,
"grad_norm": 0.43756353855133057,
"learning_rate": 7.552217427913338e-06,
"loss": 0.0014,
"step": 4082
},
{
"epoch": 7.76,
"grad_norm": 0.3307347297668457,
"learning_rate": 7.539966771468876e-06,
"loss": 0.0008,
"step": 4083
},
{
"epoch": 7.76,
"grad_norm": 0.262184202671051,
"learning_rate": 7.52772463089823e-06,
"loss": 0.0009,
"step": 4084
},
{
"epoch": 7.77,
"grad_norm": 0.34264788031578064,
"learning_rate": 7.515491010843076e-06,
"loss": 0.0018,
"step": 4085
},
{
"epoch": 7.77,
"grad_norm": 0.24883995950222015,
"learning_rate": 7.503265915941906e-06,
"loss": 0.0009,
"step": 4086
},
{
"epoch": 7.77,
"grad_norm": 0.011627244763076305,
"learning_rate": 7.491049350829911e-06,
"loss": 0.0001,
"step": 4087
},
{
"epoch": 7.77,
"grad_norm": 0.21853184700012207,
"learning_rate": 7.4788413201391085e-06,
"loss": 0.0006,
"step": 4088
},
{
"epoch": 7.77,
"grad_norm": 0.22365134954452515,
"learning_rate": 7.46664182849825e-06,
"loss": 0.0028,
"step": 4089
},
{
"epoch": 7.78,
"grad_norm": 0.37186142802238464,
"learning_rate": 7.454450880532855e-06,
"loss": 0.0009,
"step": 4090
},
{
"epoch": 7.78,
"grad_norm": 0.17585837841033936,
"learning_rate": 7.442268480865207e-06,
"loss": 0.0006,
"step": 4091
},
{
"epoch": 7.78,
"grad_norm": 0.17626003921031952,
"learning_rate": 7.430094634114336e-06,
"loss": 0.0021,
"step": 4092
},
{
"epoch": 7.78,
"grad_norm": 0.06164146959781647,
"learning_rate": 7.4179293448960485e-06,
"loss": 0.0006,
"step": 4093
},
{
"epoch": 7.78,
"grad_norm": 0.5001264214515686,
"learning_rate": 7.405772617822884e-06,
"loss": 0.0037,
"step": 4094
},
{
"epoch": 7.79,
"grad_norm": 0.331493616104126,
"learning_rate": 7.393624457504174e-06,
"loss": 0.001,
"step": 4095
},
{
"epoch": 7.79,
"grad_norm": 0.06986657530069351,
"learning_rate": 7.38148486854594e-06,
"loss": 0.0007,
"step": 4096
},
{
"epoch": 7.79,
"grad_norm": 0.5475260019302368,
"learning_rate": 7.3693538555510255e-06,
"loss": 0.0031,
"step": 4097
},
{
"epoch": 7.79,
"grad_norm": 0.16792064905166626,
"learning_rate": 7.357231423118951e-06,
"loss": 0.0009,
"step": 4098
},
{
"epoch": 7.79,
"grad_norm": 0.2036399394273758,
"learning_rate": 7.345117575846046e-06,
"loss": 0.0019,
"step": 4099
},
{
"epoch": 7.79,
"grad_norm": 0.2594846785068512,
"learning_rate": 7.333012318325349e-06,
"loss": 0.0008,
"step": 4100
},
{
"epoch": 7.79,
"eval_blimp_filtered_avg": 0.7171641791044776,
"eval_blimp_filtered_std": 0.0049850872655252615,
"step": 4100
},
{
"epoch": 7.79,
"eval_blimp_supplement_avg": 0.7823275862068966,
"eval_blimp_supplement_std": 0.01794720092061663,
"step": 4100
},
{
"epoch": 7.79,
"eval_vqa_filtered_avg": 0.32,
"eval_vqa_filtered_std": 0.046882617226215034,
"step": 4100
},
{
"epoch": 7.79,
"eval_winoground_filtered_avg": 0.48,
"eval_winoground_filtered_std": 0.05021167315686779,
"step": 4100
},
{
"epoch": 7.8,
"grad_norm": 0.120963454246521,
"learning_rate": 7.320915655146648e-06,
"loss": 0.0014,
"step": 4101
},
{
"epoch": 7.8,
"grad_norm": 0.08267587423324585,
"learning_rate": 7.308827590896467e-06,
"loss": 0.0005,
"step": 4102
},
{
"epoch": 7.8,
"grad_norm": 0.14349953830242157,
"learning_rate": 7.296748130158106e-06,
"loss": 0.0004,
"step": 4103
},
{
"epoch": 7.8,
"grad_norm": 0.05127566307783127,
"learning_rate": 7.284677277511538e-06,
"loss": 0.0004,
"step": 4104
},
{
"epoch": 7.8,
"grad_norm": 0.10821498185396194,
"learning_rate": 7.272615037533534e-06,
"loss": 0.0021,
"step": 4105
},
{
"epoch": 7.81,
"grad_norm": 0.09036865830421448,
"learning_rate": 7.2605614147975675e-06,
"loss": 0.0008,
"step": 4106
},
{
"epoch": 7.81,
"grad_norm": 0.07076483964920044,
"learning_rate": 7.248516413873848e-06,
"loss": 0.0005,
"step": 4107
},
{
"epoch": 7.81,
"grad_norm": 0.056843724101781845,
"learning_rate": 7.23648003932932e-06,
"loss": 0.0004,
"step": 4108
},
{
"epoch": 7.81,
"grad_norm": 0.3755875825881958,
"learning_rate": 7.224452295727662e-06,
"loss": 0.0058,
"step": 4109
},
{
"epoch": 7.81,
"grad_norm": 0.11845730990171432,
"learning_rate": 7.212433187629266e-06,
"loss": 0.0007,
"step": 4110
},
{
"epoch": 7.82,
"grad_norm": 0.22822058200836182,
"learning_rate": 7.200422719591254e-06,
"loss": 0.0015,
"step": 4111
},
{
"epoch": 7.82,
"grad_norm": 0.0794849842786789,
"learning_rate": 7.188420896167503e-06,
"loss": 0.0002,
"step": 4112
},
{
"epoch": 7.82,
"grad_norm": 0.7242807745933533,
"learning_rate": 7.1764277219085475e-06,
"loss": 0.0073,
"step": 4113
},
{
"epoch": 7.82,
"grad_norm": 0.04256763681769371,
"learning_rate": 7.164443201361711e-06,
"loss": 0.0004,
"step": 4114
},
{
"epoch": 7.82,
"grad_norm": 0.18121100962162018,
"learning_rate": 7.152467339070997e-06,
"loss": 0.0007,
"step": 4115
},
{
"epoch": 7.83,
"grad_norm": 0.3838452696800232,
"learning_rate": 7.140500139577133e-06,
"loss": 0.0013,
"step": 4116
},
{
"epoch": 7.83,
"grad_norm": 0.0714602917432785,
"learning_rate": 7.128541607417562e-06,
"loss": 0.0002,
"step": 4117
},
{
"epoch": 7.83,
"grad_norm": 0.3244604766368866,
"learning_rate": 7.1165917471264455e-06,
"loss": 0.0021,
"step": 4118
},
{
"epoch": 7.83,
"grad_norm": 0.18848006427288055,
"learning_rate": 7.104650563234654e-06,
"loss": 0.0007,
"step": 4119
},
{
"epoch": 7.83,
"grad_norm": 0.44399628043174744,
"learning_rate": 7.092718060269763e-06,
"loss": 0.0016,
"step": 4120
},
{
"epoch": 7.83,
"grad_norm": 0.03560929000377655,
"learning_rate": 7.08079424275607e-06,
"loss": 0.0003,
"step": 4121
},
{
"epoch": 7.84,
"grad_norm": 0.7429314255714417,
"learning_rate": 7.068879115214559e-06,
"loss": 0.0148,
"step": 4122
},
{
"epoch": 7.84,
"grad_norm": 0.015057491138577461,
"learning_rate": 7.056972682162953e-06,
"loss": 0.0001,
"step": 4123
},
{
"epoch": 7.84,
"grad_norm": 0.06351376324892044,
"learning_rate": 7.045074948115631e-06,
"loss": 0.0003,
"step": 4124
},
{
"epoch": 7.84,
"grad_norm": 0.10154907405376434,
"learning_rate": 7.0331859175837195e-06,
"loss": 0.0012,
"step": 4125
},
{
"epoch": 7.84,
"grad_norm": 0.34010088443756104,
"learning_rate": 7.021305595075015e-06,
"loss": 0.0011,
"step": 4126
},
{
"epoch": 7.85,
"grad_norm": 0.4375177025794983,
"learning_rate": 7.009433985094026e-06,
"loss": 0.0018,
"step": 4127
},
{
"epoch": 7.85,
"grad_norm": 0.19274337589740753,
"learning_rate": 6.997571092141955e-06,
"loss": 0.0019,
"step": 4128
},
{
"epoch": 7.85,
"grad_norm": 0.21760840713977814,
"learning_rate": 6.985716920716693e-06,
"loss": 0.0007,
"step": 4129
},
{
"epoch": 7.85,
"grad_norm": 0.29339998960494995,
"learning_rate": 6.973871475312835e-06,
"loss": 0.0066,
"step": 4130
},
{
"epoch": 7.85,
"grad_norm": 0.05779773369431496,
"learning_rate": 6.962034760421647e-06,
"loss": 0.0004,
"step": 4131
},
{
"epoch": 7.86,
"grad_norm": 0.49420464038848877,
"learning_rate": 6.9502067805311324e-06,
"loss": 0.0044,
"step": 4132
},
{
"epoch": 7.86,
"grad_norm": 0.7091224789619446,
"learning_rate": 6.938387540125912e-06,
"loss": 0.0128,
"step": 4133
},
{
"epoch": 7.86,
"grad_norm": 0.08009470254182816,
"learning_rate": 6.926577043687357e-06,
"loss": 0.001,
"step": 4134
},
{
"epoch": 7.86,
"grad_norm": 0.3092297911643982,
"learning_rate": 6.914775295693486e-06,
"loss": 0.0013,
"step": 4135
},
{
"epoch": 7.86,
"grad_norm": 0.7008885145187378,
"learning_rate": 6.9029823006190155e-06,
"loss": 0.0017,
"step": 4136
},
{
"epoch": 7.87,
"grad_norm": 0.3131137788295746,
"learning_rate": 6.89119806293534e-06,
"loss": 0.0007,
"step": 4137
},
{
"epoch": 7.87,
"grad_norm": 0.033942148089408875,
"learning_rate": 6.87942258711053e-06,
"loss": 0.0002,
"step": 4138
},
{
"epoch": 7.87,
"grad_norm": 0.0576227605342865,
"learning_rate": 6.867655877609337e-06,
"loss": 0.0002,
"step": 4139
},
{
"epoch": 7.87,
"grad_norm": 0.3918400704860687,
"learning_rate": 6.855897938893185e-06,
"loss": 0.0008,
"step": 4140
},
{
"epoch": 7.87,
"grad_norm": 0.14942248165607452,
"learning_rate": 6.844148775420196e-06,
"loss": 0.0008,
"step": 4141
},
{
"epoch": 7.87,
"grad_norm": 0.07059932500123978,
"learning_rate": 6.832408391645112e-06,
"loss": 0.0003,
"step": 4142
},
{
"epoch": 7.88,
"grad_norm": 0.06054045632481575,
"learning_rate": 6.820676792019406e-06,
"loss": 0.0003,
"step": 4143
},
{
"epoch": 7.88,
"grad_norm": 0.08074217289686203,
"learning_rate": 6.808953980991186e-06,
"loss": 0.0003,
"step": 4144
},
{
"epoch": 7.88,
"grad_norm": 0.5490571856498718,
"learning_rate": 6.797239963005234e-06,
"loss": 0.001,
"step": 4145
},
{
"epoch": 7.88,
"grad_norm": 0.4995369017124176,
"learning_rate": 6.785534742503e-06,
"loss": 0.0166,
"step": 4146
},
{
"epoch": 7.88,
"grad_norm": 0.2981852889060974,
"learning_rate": 6.773838323922595e-06,
"loss": 0.0016,
"step": 4147
},
{
"epoch": 7.89,
"grad_norm": 0.44925040006637573,
"learning_rate": 6.762150711698789e-06,
"loss": 0.0011,
"step": 4148
},
{
"epoch": 7.89,
"grad_norm": 0.4283827543258667,
"learning_rate": 6.750471910263046e-06,
"loss": 0.0036,
"step": 4149
},
{
"epoch": 7.89,
"grad_norm": 0.1738491952419281,
"learning_rate": 6.738801924043424e-06,
"loss": 0.0006,
"step": 4150
},
{
"epoch": 7.89,
"grad_norm": 0.04501824826002121,
"learning_rate": 6.727140757464709e-06,
"loss": 0.0004,
"step": 4151
},
{
"epoch": 7.89,
"grad_norm": 0.0502665638923645,
"learning_rate": 6.715488414948302e-06,
"loss": 0.0002,
"step": 4152
},
{
"epoch": 7.9,
"grad_norm": 0.024165892973542213,
"learning_rate": 6.703844900912261e-06,
"loss": 0.0002,
"step": 4153
},
{
"epoch": 7.9,
"grad_norm": 0.126878023147583,
"learning_rate": 6.692210219771313e-06,
"loss": 0.0007,
"step": 4154
},
{
"epoch": 7.9,
"grad_norm": 0.09076157957315445,
"learning_rate": 6.680584375936818e-06,
"loss": 0.0006,
"step": 4155
},
{
"epoch": 7.9,
"grad_norm": 0.0892454981803894,
"learning_rate": 6.6689673738167966e-06,
"loss": 0.0011,
"step": 4156
},
{
"epoch": 7.9,
"grad_norm": 0.21645459532737732,
"learning_rate": 6.657359217815906e-06,
"loss": 0.0011,
"step": 4157
},
{
"epoch": 7.9,
"grad_norm": 0.15177641808986664,
"learning_rate": 6.645759912335482e-06,
"loss": 0.0007,
"step": 4158
},
{
"epoch": 7.91,
"grad_norm": 0.0453759990632534,
"learning_rate": 6.634169461773446e-06,
"loss": 0.0005,
"step": 4159
},
{
"epoch": 7.91,
"grad_norm": 0.06536399573087692,
"learning_rate": 6.622587870524424e-06,
"loss": 0.0005,
"step": 4160
},
{
"epoch": 7.91,
"grad_norm": 0.059902504086494446,
"learning_rate": 6.611015142979642e-06,
"loss": 0.0005,
"step": 4161
},
{
"epoch": 7.91,
"grad_norm": 0.10877792537212372,
"learning_rate": 6.599451283526977e-06,
"loss": 0.0006,
"step": 4162
},
{
"epoch": 7.91,
"grad_norm": 0.05095542222261429,
"learning_rate": 6.587896296550949e-06,
"loss": 0.0004,
"step": 4163
},
{
"epoch": 7.92,
"grad_norm": 0.9031468033790588,
"learning_rate": 6.57635018643271e-06,
"loss": 0.0065,
"step": 4164
},
{
"epoch": 7.92,
"grad_norm": 0.07147204875946045,
"learning_rate": 6.564812957550045e-06,
"loss": 0.0005,
"step": 4165
},
{
"epoch": 7.92,
"grad_norm": 0.04257633537054062,
"learning_rate": 6.553284614277361e-06,
"loss": 0.0003,
"step": 4166
},
{
"epoch": 7.92,
"grad_norm": 0.017738549038767815,
"learning_rate": 6.541765160985734e-06,
"loss": 0.0002,
"step": 4167
},
{
"epoch": 7.92,
"grad_norm": 0.04792720079421997,
"learning_rate": 6.530254602042812e-06,
"loss": 0.0003,
"step": 4168
},
{
"epoch": 7.93,
"grad_norm": 0.32888808846473694,
"learning_rate": 6.51875294181293e-06,
"loss": 0.0006,
"step": 4169
},
{
"epoch": 7.93,
"grad_norm": 0.2585296928882599,
"learning_rate": 6.5072601846569926e-06,
"loss": 0.0011,
"step": 4170
},
{
"epoch": 7.93,
"grad_norm": 0.10473588109016418,
"learning_rate": 6.495776334932577e-06,
"loss": 0.0007,
"step": 4171
},
{
"epoch": 7.93,
"grad_norm": 0.04327563941478729,
"learning_rate": 6.484301396993858e-06,
"loss": 0.0004,
"step": 4172
},
{
"epoch": 7.93,
"grad_norm": 0.4245152473449707,
"learning_rate": 6.472835375191635e-06,
"loss": 0.0021,
"step": 4173
},
{
"epoch": 7.94,
"grad_norm": 0.02012992650270462,
"learning_rate": 6.4613782738733265e-06,
"loss": 0.0004,
"step": 4174
},
{
"epoch": 7.94,
"grad_norm": 0.4454987049102783,
"learning_rate": 6.449930097382971e-06,
"loss": 0.0017,
"step": 4175
},
{
"epoch": 7.94,
"grad_norm": 0.04466366022825241,
"learning_rate": 6.438490850061221e-06,
"loss": 0.0004,
"step": 4176
},
{
"epoch": 7.94,
"grad_norm": 0.20599092543125153,
"learning_rate": 6.427060536245339e-06,
"loss": 0.0014,
"step": 4177
},
{
"epoch": 7.94,
"grad_norm": 0.5876622796058655,
"learning_rate": 6.4156391602692275e-06,
"loss": 0.0032,
"step": 4178
},
{
"epoch": 7.94,
"grad_norm": 0.18972648680210114,
"learning_rate": 6.404226726463348e-06,
"loss": 0.0007,
"step": 4179
},
{
"epoch": 7.95,
"grad_norm": 0.01247115433216095,
"learning_rate": 6.392823239154825e-06,
"loss": 0.0001,
"step": 4180
},
{
"epoch": 7.95,
"grad_norm": 0.19547367095947266,
"learning_rate": 6.381428702667363e-06,
"loss": 0.0018,
"step": 4181
},
{
"epoch": 7.95,
"grad_norm": 0.05993138253688812,
"learning_rate": 6.370043121321275e-06,
"loss": 0.0005,
"step": 4182
},
{
"epoch": 7.95,
"grad_norm": 0.2719241976737976,
"learning_rate": 6.35866649943348e-06,
"loss": 0.0006,
"step": 4183
},
{
"epoch": 7.95,
"grad_norm": 0.030581068247556686,
"learning_rate": 6.347298841317506e-06,
"loss": 0.0002,
"step": 4184
},
{
"epoch": 7.96,
"grad_norm": 0.10038358718156815,
"learning_rate": 6.3359401512834725e-06,
"loss": 0.0016,
"step": 4185
},
{
"epoch": 7.96,
"grad_norm": 0.38063931465148926,
"learning_rate": 6.324590433638101e-06,
"loss": 0.0009,
"step": 4186
},
{
"epoch": 7.96,
"grad_norm": 0.17972616851329803,
"learning_rate": 6.313249692684732e-06,
"loss": 0.0005,
"step": 4187
},
{
"epoch": 7.96,
"grad_norm": 0.04446972906589508,
"learning_rate": 6.301917932723258e-06,
"loss": 0.0003,
"step": 4188
},
{
"epoch": 7.96,
"grad_norm": 0.10818032175302505,
"learning_rate": 6.290595158050213e-06,
"loss": 0.0008,
"step": 4189
},
{
"epoch": 7.97,
"grad_norm": 0.27519017457962036,
"learning_rate": 6.279281372958694e-06,
"loss": 0.001,
"step": 4190
},
{
"epoch": 7.97,
"grad_norm": 0.04331502690911293,
"learning_rate": 6.267976581738406e-06,
"loss": 0.0003,
"step": 4191
},
{
"epoch": 7.97,
"grad_norm": 0.306720107793808,
"learning_rate": 6.256680788675624e-06,
"loss": 0.0015,
"step": 4192
},
{
"epoch": 7.97,
"grad_norm": 0.46065255999565125,
"learning_rate": 6.245393998053248e-06,
"loss": 0.0021,
"step": 4193
},
{
"epoch": 7.97,
"grad_norm": 0.017609836533665657,
"learning_rate": 6.2341162141507125e-06,
"loss": 0.0002,
"step": 4194
},
{
"epoch": 7.98,
"grad_norm": 0.1307341754436493,
"learning_rate": 6.2228474412440924e-06,
"loss": 0.001,
"step": 4195
},
{
"epoch": 7.98,
"grad_norm": 0.8371809720993042,
"learning_rate": 6.211587683605995e-06,
"loss": 0.0032,
"step": 4196
},
{
"epoch": 7.98,
"grad_norm": 0.1644481122493744,
"learning_rate": 6.2003369455056535e-06,
"loss": 0.0004,
"step": 4197
},
{
"epoch": 7.98,
"grad_norm": 0.11752185970544815,
"learning_rate": 6.18909523120885e-06,
"loss": 0.0007,
"step": 4198
},
{
"epoch": 7.98,
"grad_norm": 0.13133354485034943,
"learning_rate": 6.177862544977965e-06,
"loss": 0.0019,
"step": 4199
},
{
"epoch": 7.98,
"grad_norm": 0.13598887622356415,
"learning_rate": 6.16663889107194e-06,
"loss": 0.0007,
"step": 4200
},
{
"epoch": 7.98,
"eval_blimp_filtered_avg": 0.716865671641791,
"eval_blimp_filtered_std": 0.004985538625219207,
"step": 4200
},
{
"epoch": 7.98,
"eval_blimp_supplement_avg": 0.7737068965517241,
"eval_blimp_supplement_std": 0.018116375477416412,
"step": 4200
},
{
"epoch": 7.98,
"eval_vqa_filtered_avg": 0.34,
"eval_vqa_filtered_std": 0.04760952285695235,
"step": 4200
},
{
"epoch": 7.98,
"eval_winoground_filtered_avg": 0.49,
"eval_winoground_filtered_std": 0.05024183937956912,
"step": 4200
},
{
"epoch": 7.99,
"grad_norm": 0.08354678004980087,
"learning_rate": 6.155424273746304e-06,
"loss": 0.0008,
"step": 4201
},
{
"epoch": 7.99,
"grad_norm": 0.06116069108247757,
"learning_rate": 6.14421869725315e-06,
"loss": 0.0003,
"step": 4202
},
{
"epoch": 7.99,
"grad_norm": 0.18562187254428864,
"learning_rate": 6.133022165841146e-06,
"loss": 0.0004,
"step": 4203
},
{
"epoch": 7.99,
"grad_norm": 0.40135806798934937,
"learning_rate": 6.121834683755559e-06,
"loss": 0.0014,
"step": 4204
},
{
"epoch": 7.99,
"grad_norm": 0.04748351871967316,
"learning_rate": 6.110656255238155e-06,
"loss": 0.0003,
"step": 4205
},
{
"epoch": 8.0,
"grad_norm": 0.08310114592313766,
"learning_rate": 6.099486884527347e-06,
"loss": 0.0003,
"step": 4206
},
{
"epoch": 8.0,
"grad_norm": 0.11161719262599945,
"learning_rate": 6.088326575858062e-06,
"loss": 0.0013,
"step": 4207
},
{
"epoch": 8.0,
"grad_norm": 0.14086154103279114,
"learning_rate": 6.077175333461809e-06,
"loss": 0.0006,
"step": 4208
},
{
"epoch": 8.0,
"grad_norm": 0.062327951192855835,
"learning_rate": 6.066033161566661e-06,
"loss": 0.0002,
"step": 4209
},
{
"epoch": 8.0,
"grad_norm": 0.04026307538151741,
"learning_rate": 6.054900064397243e-06,
"loss": 0.0003,
"step": 4210
},
{
"epoch": 8.01,
"grad_norm": 0.0691170021891594,
"learning_rate": 6.043776046174742e-06,
"loss": 0.0007,
"step": 4211
},
{
"epoch": 8.01,
"grad_norm": 0.3420390486717224,
"learning_rate": 6.032661111116906e-06,
"loss": 0.0015,
"step": 4212
},
{
"epoch": 8.01,
"grad_norm": 0.033876050263643265,
"learning_rate": 6.0215552634380525e-06,
"loss": 0.0002,
"step": 4213
},
{
"epoch": 8.01,
"grad_norm": 0.2442551553249359,
"learning_rate": 6.010458507349013e-06,
"loss": 0.0015,
"step": 4214
},
{
"epoch": 8.01,
"grad_norm": 0.2074851095676422,
"learning_rate": 5.999370847057222e-06,
"loss": 0.0009,
"step": 4215
},
{
"epoch": 8.02,
"grad_norm": 0.0441262386739254,
"learning_rate": 5.988292286766627e-06,
"loss": 0.0002,
"step": 4216
},
{
"epoch": 8.02,
"grad_norm": 0.4127471148967743,
"learning_rate": 5.977222830677744e-06,
"loss": 0.0007,
"step": 4217
},
{
"epoch": 8.02,
"grad_norm": 0.04880307614803314,
"learning_rate": 5.9661624829876334e-06,
"loss": 0.0003,
"step": 4218
},
{
"epoch": 8.02,
"grad_norm": 0.033505674451589584,
"learning_rate": 5.955111247889897e-06,
"loss": 0.0002,
"step": 4219
},
{
"epoch": 8.02,
"grad_norm": 0.2133377492427826,
"learning_rate": 5.944069129574686e-06,
"loss": 0.0011,
"step": 4220
},
{
"epoch": 8.02,
"grad_norm": 0.0844859704375267,
"learning_rate": 5.933036132228699e-06,
"loss": 0.0003,
"step": 4221
},
{
"epoch": 8.03,
"grad_norm": 0.14875498414039612,
"learning_rate": 5.922012260035168e-06,
"loss": 0.0014,
"step": 4222
},
{
"epoch": 8.03,
"grad_norm": 0.03871171921491623,
"learning_rate": 5.91099751717386e-06,
"loss": 0.0003,
"step": 4223
},
{
"epoch": 8.03,
"grad_norm": 0.06902189552783966,
"learning_rate": 5.899991907821116e-06,
"loss": 0.0007,
"step": 4224
},
{
"epoch": 8.03,
"grad_norm": 0.32611891627311707,
"learning_rate": 5.8889954361497536e-06,
"loss": 0.0009,
"step": 4225
},
{
"epoch": 8.03,
"grad_norm": 0.09312132745981216,
"learning_rate": 5.878008106329183e-06,
"loss": 0.0003,
"step": 4226
},
{
"epoch": 8.04,
"grad_norm": 0.06205948814749718,
"learning_rate": 5.867029922525321e-06,
"loss": 0.0004,
"step": 4227
},
{
"epoch": 8.04,
"grad_norm": 0.523986279964447,
"learning_rate": 5.856060888900613e-06,
"loss": 0.0012,
"step": 4228
},
{
"epoch": 8.04,
"grad_norm": 0.031348563730716705,
"learning_rate": 5.845101009614049e-06,
"loss": 0.0002,
"step": 4229
},
{
"epoch": 8.04,
"grad_norm": 0.03327319398522377,
"learning_rate": 5.834150288821142e-06,
"loss": 0.0002,
"step": 4230
},
{
"epoch": 8.04,
"grad_norm": 0.140588641166687,
"learning_rate": 5.823208730673924e-06,
"loss": 0.0004,
"step": 4231
},
{
"epoch": 8.05,
"grad_norm": 0.32775843143463135,
"learning_rate": 5.812276339320962e-06,
"loss": 0.0079,
"step": 4232
},
{
"epoch": 8.05,
"grad_norm": 0.12931513786315918,
"learning_rate": 5.801353118907365e-06,
"loss": 0.0018,
"step": 4233
},
{
"epoch": 8.05,
"grad_norm": 0.12185192108154297,
"learning_rate": 5.790439073574714e-06,
"loss": 0.0003,
"step": 4234
},
{
"epoch": 8.05,
"grad_norm": 0.02792348898947239,
"learning_rate": 5.77953420746117e-06,
"loss": 0.0002,
"step": 4235
},
{
"epoch": 8.05,
"grad_norm": 0.15111584961414337,
"learning_rate": 5.768638524701378e-06,
"loss": 0.0006,
"step": 4236
},
{
"epoch": 8.06,
"grad_norm": 0.012947790324687958,
"learning_rate": 5.75775202942651e-06,
"loss": 0.0001,
"step": 4237
},
{
"epoch": 8.06,
"grad_norm": 0.41861024498939514,
"learning_rate": 5.746874725764245e-06,
"loss": 0.0014,
"step": 4238
},
{
"epoch": 8.06,
"grad_norm": 0.09514033049345016,
"learning_rate": 5.736006617838811e-06,
"loss": 0.0007,
"step": 4239
},
{
"epoch": 8.06,
"grad_norm": 0.5159668922424316,
"learning_rate": 5.725147709770897e-06,
"loss": 0.0014,
"step": 4240
},
{
"epoch": 8.06,
"grad_norm": 0.019545137882232666,
"learning_rate": 5.714298005677748e-06,
"loss": 0.0002,
"step": 4241
},
{
"epoch": 8.06,
"grad_norm": 0.026386486366391182,
"learning_rate": 5.703457509673098e-06,
"loss": 0.0001,
"step": 4242
},
{
"epoch": 8.07,
"grad_norm": 0.014095514081418514,
"learning_rate": 5.692626225867198e-06,
"loss": 0.0001,
"step": 4243
},
{
"epoch": 8.07,
"grad_norm": 0.05163192003965378,
"learning_rate": 5.681804158366797e-06,
"loss": 0.0003,
"step": 4244
},
{
"epoch": 8.07,
"grad_norm": 0.01157307531684637,
"learning_rate": 5.670991311275162e-06,
"loss": 0.0001,
"step": 4245
},
{
"epoch": 8.07,
"grad_norm": 0.3206714391708374,
"learning_rate": 5.660187688692048e-06,
"loss": 0.0008,
"step": 4246
},
{
"epoch": 8.07,
"grad_norm": 0.05928007513284683,
"learning_rate": 5.64939329471373e-06,
"loss": 0.0005,
"step": 4247
},
{
"epoch": 8.08,
"grad_norm": 0.17311754822731018,
"learning_rate": 5.638608133432973e-06,
"loss": 0.0004,
"step": 4248
},
{
"epoch": 8.08,
"grad_norm": 0.21453316509723663,
"learning_rate": 5.627832208939037e-06,
"loss": 0.0009,
"step": 4249
},
{
"epoch": 8.08,
"grad_norm": 0.4050031900405884,
"learning_rate": 5.617065525317706e-06,
"loss": 0.0014,
"step": 4250
},
{
"epoch": 8.08,
"grad_norm": 0.24241989850997925,
"learning_rate": 5.606308086651216e-06,
"loss": 0.0006,
"step": 4251
},
{
"epoch": 8.08,
"grad_norm": 0.0463268905878067,
"learning_rate": 5.595559897018342e-06,
"loss": 0.0006,
"step": 4252
},
{
"epoch": 8.09,
"grad_norm": 0.04105130583047867,
"learning_rate": 5.5848209604943234e-06,
"loss": 0.0002,
"step": 4253
},
{
"epoch": 8.09,
"grad_norm": 0.2185387909412384,
"learning_rate": 5.574091281150907e-06,
"loss": 0.0005,
"step": 4254
},
{
"epoch": 8.09,
"grad_norm": 0.05630626156926155,
"learning_rate": 5.563370863056318e-06,
"loss": 0.0003,
"step": 4255
},
{
"epoch": 8.09,
"grad_norm": 0.03223489224910736,
"learning_rate": 5.55265971027528e-06,
"loss": 0.0003,
"step": 4256
},
{
"epoch": 8.09,
"grad_norm": 0.011499090120196342,
"learning_rate": 5.541957826868995e-06,
"loss": 0.0001,
"step": 4257
},
{
"epoch": 8.1,
"grad_norm": 0.05199449881911278,
"learning_rate": 5.531265216895149e-06,
"loss": 0.0004,
"step": 4258
},
{
"epoch": 8.1,
"grad_norm": 0.10791318863630295,
"learning_rate": 5.520581884407942e-06,
"loss": 0.0005,
"step": 4259
},
{
"epoch": 8.1,
"grad_norm": 0.17202763259410858,
"learning_rate": 5.509907833457998e-06,
"loss": 0.0008,
"step": 4260
},
{
"epoch": 8.1,
"grad_norm": 0.009432696737349033,
"learning_rate": 5.499243068092478e-06,
"loss": 0.0001,
"step": 4261
},
{
"epoch": 8.1,
"grad_norm": 0.2289062887430191,
"learning_rate": 5.488587592354998e-06,
"loss": 0.0007,
"step": 4262
},
{
"epoch": 8.1,
"grad_norm": 0.029807062819600105,
"learning_rate": 5.47794141028565e-06,
"loss": 0.0001,
"step": 4263
},
{
"epoch": 8.11,
"grad_norm": 0.03920219466090202,
"learning_rate": 5.467304525921007e-06,
"loss": 0.0002,
"step": 4264
},
{
"epoch": 8.11,
"grad_norm": 0.3621011972427368,
"learning_rate": 5.456676943294112e-06,
"loss": 0.0011,
"step": 4265
},
{
"epoch": 8.11,
"grad_norm": 0.2532673180103302,
"learning_rate": 5.446058666434493e-06,
"loss": 0.0011,
"step": 4266
},
{
"epoch": 8.11,
"grad_norm": 0.5887459516525269,
"learning_rate": 5.4354496993681346e-06,
"loss": 0.0013,
"step": 4267
},
{
"epoch": 8.11,
"grad_norm": 0.006665298715233803,
"learning_rate": 5.424850046117501e-06,
"loss": 0.0001,
"step": 4268
},
{
"epoch": 8.12,
"grad_norm": 0.008848692290484905,
"learning_rate": 5.414259710701515e-06,
"loss": 0.0001,
"step": 4269
},
{
"epoch": 8.12,
"grad_norm": 0.021025920286774635,
"learning_rate": 5.403678697135597e-06,
"loss": 0.0002,
"step": 4270
},
{
"epoch": 8.12,
"grad_norm": 0.029037991538643837,
"learning_rate": 5.3931070094315815e-06,
"loss": 0.0003,
"step": 4271
},
{
"epoch": 8.12,
"grad_norm": 0.04867730662226677,
"learning_rate": 5.382544651597813e-06,
"loss": 0.0005,
"step": 4272
},
{
"epoch": 8.12,
"grad_norm": 0.07016139477491379,
"learning_rate": 5.371991627639077e-06,
"loss": 0.0006,
"step": 4273
},
{
"epoch": 8.13,
"grad_norm": 0.20593419671058655,
"learning_rate": 5.361447941556626e-06,
"loss": 0.0005,
"step": 4274
},
{
"epoch": 8.13,
"grad_norm": 0.15504324436187744,
"learning_rate": 5.350913597348167e-06,
"loss": 0.0012,
"step": 4275
},
{
"epoch": 8.13,
"grad_norm": 0.18425977230072021,
"learning_rate": 5.340388599007867e-06,
"loss": 0.0019,
"step": 4276
},
{
"epoch": 8.13,
"grad_norm": 0.0976550281047821,
"learning_rate": 5.329872950526355e-06,
"loss": 0.0009,
"step": 4277
},
{
"epoch": 8.13,
"grad_norm": 0.023634426295757294,
"learning_rate": 5.319366655890696e-06,
"loss": 0.0002,
"step": 4278
},
{
"epoch": 8.13,
"grad_norm": 0.2677626311779022,
"learning_rate": 5.308869719084453e-06,
"loss": 0.0009,
"step": 4279
},
{
"epoch": 8.14,
"grad_norm": 0.013481765985488892,
"learning_rate": 5.298382144087577e-06,
"loss": 0.0001,
"step": 4280
},
{
"epoch": 8.14,
"grad_norm": 0.18845850229263306,
"learning_rate": 5.287903934876525e-06,
"loss": 0.0006,
"step": 4281
},
{
"epoch": 8.14,
"grad_norm": 0.04124622419476509,
"learning_rate": 5.277435095424172e-06,
"loss": 0.0003,
"step": 4282
},
{
"epoch": 8.14,
"grad_norm": 0.3568054735660553,
"learning_rate": 5.266975629699855e-06,
"loss": 0.0024,
"step": 4283
},
{
"epoch": 8.14,
"grad_norm": 0.011829476803541183,
"learning_rate": 5.2565255416693405e-06,
"loss": 0.0001,
"step": 4284
},
{
"epoch": 8.15,
"grad_norm": 0.023779241368174553,
"learning_rate": 5.246084835294871e-06,
"loss": 0.0001,
"step": 4285
},
{
"epoch": 8.15,
"grad_norm": 0.10504792630672455,
"learning_rate": 5.235653514535084e-06,
"loss": 0.0008,
"step": 4286
},
{
"epoch": 8.15,
"grad_norm": 0.1687890738248825,
"learning_rate": 5.225231583345104e-06,
"loss": 0.0015,
"step": 4287
},
{
"epoch": 8.15,
"grad_norm": 0.0338444709777832,
"learning_rate": 5.214819045676477e-06,
"loss": 0.0004,
"step": 4288
},
{
"epoch": 8.15,
"grad_norm": 0.7579665780067444,
"learning_rate": 5.204415905477178e-06,
"loss": 0.0057,
"step": 4289
},
{
"epoch": 8.16,
"grad_norm": 0.06252743303775787,
"learning_rate": 5.194022166691636e-06,
"loss": 0.0003,
"step": 4290
},
{
"epoch": 8.16,
"grad_norm": 0.08311913907527924,
"learning_rate": 5.183637833260703e-06,
"loss": 0.0013,
"step": 4291
},
{
"epoch": 8.16,
"grad_norm": 0.10357506573200226,
"learning_rate": 5.17326290912167e-06,
"loss": 0.0002,
"step": 4292
},
{
"epoch": 8.16,
"grad_norm": 0.4179705083370209,
"learning_rate": 5.1628973982082585e-06,
"loss": 0.0013,
"step": 4293
},
{
"epoch": 8.16,
"grad_norm": 0.6699661612510681,
"learning_rate": 5.152541304450625e-06,
"loss": 0.0015,
"step": 4294
},
{
"epoch": 8.17,
"grad_norm": 0.08450377732515335,
"learning_rate": 5.142194631775346e-06,
"loss": 0.0007,
"step": 4295
},
{
"epoch": 8.17,
"grad_norm": 0.17366290092468262,
"learning_rate": 5.131857384105451e-06,
"loss": 0.0005,
"step": 4296
},
{
"epoch": 8.17,
"grad_norm": 0.06300784647464752,
"learning_rate": 5.121529565360352e-06,
"loss": 0.0003,
"step": 4297
},
{
"epoch": 8.17,
"grad_norm": 0.3587765693664551,
"learning_rate": 5.111211179455929e-06,
"loss": 0.0006,
"step": 4298
},
{
"epoch": 8.17,
"grad_norm": 0.5342482328414917,
"learning_rate": 5.100902230304465e-06,
"loss": 0.0023,
"step": 4299
},
{
"epoch": 8.17,
"grad_norm": 0.025818495079874992,
"learning_rate": 5.090602721814664e-06,
"loss": 0.0003,
"step": 4300
},
{
"epoch": 8.17,
"eval_blimp_filtered_avg": 0.7173134328358209,
"eval_blimp_filtered_std": 0.004988629341142557,
"step": 4300
},
{
"epoch": 8.17,
"eval_blimp_supplement_avg": 0.7801724137931034,
"eval_blimp_supplement_std": 0.018091817922526954,
"step": 4300
},
{
"epoch": 8.17,
"eval_vqa_filtered_avg": 0.32,
"eval_vqa_filtered_std": 0.046882617226215034,
"step": 4300
},
{
"epoch": 8.17,
"eval_winoground_filtered_avg": 0.5,
"eval_winoground_filtered_std": 0.050251890762960605,
"step": 4300
},
{
"epoch": 8.18,
"grad_norm": 0.20564794540405273,
"learning_rate": 5.080312657891657e-06,
"loss": 0.0006,
"step": 4301
},
{
"epoch": 8.18,
"grad_norm": 0.1698586493730545,
"learning_rate": 5.070032042436992e-06,
"loss": 0.0005,
"step": 4302
},
{
"epoch": 8.18,
"grad_norm": 0.17516875267028809,
"learning_rate": 5.05976087934863e-06,
"loss": 0.0004,
"step": 4303
},
{
"epoch": 8.18,
"grad_norm": 0.6829932928085327,
"learning_rate": 5.049499172520952e-06,
"loss": 0.0025,
"step": 4304
},
{
"epoch": 8.18,
"grad_norm": 0.0844355896115303,
"learning_rate": 5.039246925844766e-06,
"loss": 0.001,
"step": 4305
},
{
"epoch": 8.19,
"grad_norm": 0.05232042819261551,
"learning_rate": 5.02900414320726e-06,
"loss": 0.0002,
"step": 4306
},
{
"epoch": 8.19,
"grad_norm": 0.019935037940740585,
"learning_rate": 5.018770828492073e-06,
"loss": 0.0002,
"step": 4307
},
{
"epoch": 8.19,
"grad_norm": 0.06795331090688705,
"learning_rate": 5.008546985579229e-06,
"loss": 0.0001,
"step": 4308
},
{
"epoch": 8.19,
"grad_norm": 0.09505223482847214,
"learning_rate": 4.998332618345166e-06,
"loss": 0.001,
"step": 4309
},
{
"epoch": 8.19,
"grad_norm": 0.15013574063777924,
"learning_rate": 4.988127730662734e-06,
"loss": 0.0021,
"step": 4310
},
{
"epoch": 8.2,
"grad_norm": 0.15763522684574127,
"learning_rate": 4.97793232640118e-06,
"loss": 0.0007,
"step": 4311
},
{
"epoch": 8.2,
"grad_norm": 0.022569961845874786,
"learning_rate": 4.967746409426166e-06,
"loss": 0.0001,
"step": 4312
},
{
"epoch": 8.2,
"grad_norm": 0.05333202704787254,
"learning_rate": 4.957569983599739e-06,
"loss": 0.0002,
"step": 4313
},
{
"epoch": 8.2,
"grad_norm": 0.01670350693166256,
"learning_rate": 4.947403052780386e-06,
"loss": 0.0001,
"step": 4314
},
{
"epoch": 8.2,
"grad_norm": 0.03749420866370201,
"learning_rate": 4.937245620822933e-06,
"loss": 0.0004,
"step": 4315
},
{
"epoch": 8.21,
"grad_norm": 0.06501667946577072,
"learning_rate": 4.927097691578677e-06,
"loss": 0.0005,
"step": 4316
},
{
"epoch": 8.21,
"grad_norm": 0.5726225972175598,
"learning_rate": 4.916959268895239e-06,
"loss": 0.0011,
"step": 4317
},
{
"epoch": 8.21,
"grad_norm": 0.034353092312812805,
"learning_rate": 4.906830356616691e-06,
"loss": 0.0002,
"step": 4318
},
{
"epoch": 8.21,
"grad_norm": 0.15825703740119934,
"learning_rate": 4.8967109585834755e-06,
"loss": 0.0003,
"step": 4319
},
{
"epoch": 8.21,
"grad_norm": 0.0723881870508194,
"learning_rate": 4.886601078632432e-06,
"loss": 0.0005,
"step": 4320
},
{
"epoch": 8.21,
"grad_norm": 0.013128970749676228,
"learning_rate": 4.8765007205967846e-06,
"loss": 0.0001,
"step": 4321
},
{
"epoch": 8.22,
"grad_norm": 0.0418151393532753,
"learning_rate": 4.866409888306157e-06,
"loss": 0.0002,
"step": 4322
},
{
"epoch": 8.22,
"grad_norm": 0.06738678365945816,
"learning_rate": 4.8563285855865545e-06,
"loss": 0.0002,
"step": 4323
},
{
"epoch": 8.22,
"grad_norm": 0.050115618854761124,
"learning_rate": 4.846256816260364e-06,
"loss": 0.0003,
"step": 4324
},
{
"epoch": 8.22,
"grad_norm": 0.46220970153808594,
"learning_rate": 4.8361945841463895e-06,
"loss": 0.0008,
"step": 4325
},
{
"epoch": 8.22,
"grad_norm": 0.011012556962668896,
"learning_rate": 4.82614189305976e-06,
"loss": 0.0001,
"step": 4326
},
{
"epoch": 8.23,
"grad_norm": 0.05582883581519127,
"learning_rate": 4.8160987468120495e-06,
"loss": 0.0001,
"step": 4327
},
{
"epoch": 8.23,
"grad_norm": 0.05129886791110039,
"learning_rate": 4.806065149211173e-06,
"loss": 0.0003,
"step": 4328
},
{
"epoch": 8.23,
"grad_norm": 0.044186487793922424,
"learning_rate": 4.796041104061439e-06,
"loss": 0.0007,
"step": 4329
},
{
"epoch": 8.23,
"grad_norm": 0.07311145216226578,
"learning_rate": 4.786026615163529e-06,
"loss": 0.001,
"step": 4330
},
{
"epoch": 8.23,
"grad_norm": 0.009052329696714878,
"learning_rate": 4.7760216863145175e-06,
"loss": 0.0001,
"step": 4331
},
{
"epoch": 8.24,
"grad_norm": 0.07166402786970139,
"learning_rate": 4.766026321307818e-06,
"loss": 0.0004,
"step": 4332
},
{
"epoch": 8.24,
"grad_norm": 0.02706899680197239,
"learning_rate": 4.756040523933261e-06,
"loss": 0.0001,
"step": 4333
},
{
"epoch": 8.24,
"grad_norm": 0.5649566054344177,
"learning_rate": 4.74606429797702e-06,
"loss": 0.0019,
"step": 4334
},
{
"epoch": 8.24,
"grad_norm": 0.18486475944519043,
"learning_rate": 4.736097647221653e-06,
"loss": 0.003,
"step": 4335
},
{
"epoch": 8.24,
"grad_norm": 0.12097765505313873,
"learning_rate": 4.726140575446083e-06,
"loss": 0.0008,
"step": 4336
},
{
"epoch": 8.25,
"grad_norm": 0.16778601706027985,
"learning_rate": 4.7161930864255955e-06,
"loss": 0.0011,
"step": 4337
},
{
"epoch": 8.25,
"grad_norm": 0.09574627131223679,
"learning_rate": 4.706255183931854e-06,
"loss": 0.0003,
"step": 4338
},
{
"epoch": 8.25,
"grad_norm": 0.039087746292352676,
"learning_rate": 4.69632687173287e-06,
"loss": 0.0002,
"step": 4339
},
{
"epoch": 8.25,
"grad_norm": 0.13794521987438202,
"learning_rate": 4.686408153593057e-06,
"loss": 0.0009,
"step": 4340
},
{
"epoch": 8.25,
"grad_norm": 0.022024864330887794,
"learning_rate": 4.676499033273133e-06,
"loss": 0.0001,
"step": 4341
},
{
"epoch": 8.25,
"grad_norm": 0.5170969367027283,
"learning_rate": 4.666599514530235e-06,
"loss": 0.0027,
"step": 4342
},
{
"epoch": 8.26,
"grad_norm": 0.04830276221036911,
"learning_rate": 4.656709601117809e-06,
"loss": 0.0001,
"step": 4343
},
{
"epoch": 8.26,
"grad_norm": 0.015258397907018661,
"learning_rate": 4.646829296785697e-06,
"loss": 0.0001,
"step": 4344
},
{
"epoch": 8.26,
"grad_norm": 0.08629477024078369,
"learning_rate": 4.63695860528008e-06,
"loss": 0.0004,
"step": 4345
},
{
"epoch": 8.26,
"grad_norm": 0.0736272931098938,
"learning_rate": 4.627097530343502e-06,
"loss": 0.0005,
"step": 4346
},
{
"epoch": 8.26,
"grad_norm": 0.041141003370285034,
"learning_rate": 4.617246075714849e-06,
"loss": 0.0002,
"step": 4347
},
{
"epoch": 8.27,
"grad_norm": 0.22111564874649048,
"learning_rate": 4.607404245129369e-06,
"loss": 0.0006,
"step": 4348
},
{
"epoch": 8.27,
"grad_norm": 0.0444243848323822,
"learning_rate": 4.59757204231866e-06,
"loss": 0.0002,
"step": 4349
},
{
"epoch": 8.27,
"grad_norm": 0.043807871639728546,
"learning_rate": 4.587749471010664e-06,
"loss": 0.0004,
"step": 4350
},
{
"epoch": 8.27,
"grad_norm": 0.006630151066929102,
"learning_rate": 4.577936534929689e-06,
"loss": 0.0001,
"step": 4351
},
{
"epoch": 8.27,
"grad_norm": 0.49490368366241455,
"learning_rate": 4.5681332377963555e-06,
"loss": 0.002,
"step": 4352
},
{
"epoch": 8.28,
"grad_norm": 0.19145682454109192,
"learning_rate": 4.5583395833276655e-06,
"loss": 0.0013,
"step": 4353
},
{
"epoch": 8.28,
"grad_norm": 0.15242141485214233,
"learning_rate": 4.5485555752369476e-06,
"loss": 0.0008,
"step": 4354
},
{
"epoch": 8.28,
"grad_norm": 0.0777655765414238,
"learning_rate": 4.538781217233873e-06,
"loss": 0.0006,
"step": 4355
},
{
"epoch": 8.28,
"grad_norm": 0.006433681584894657,
"learning_rate": 4.529016513024452e-06,
"loss": 0.0001,
"step": 4356
},
{
"epoch": 8.28,
"grad_norm": 0.08017545938491821,
"learning_rate": 4.519261466311044e-06,
"loss": 0.0004,
"step": 4357
},
{
"epoch": 8.29,
"grad_norm": 0.08373193442821503,
"learning_rate": 4.509516080792337e-06,
"loss": 0.0058,
"step": 4358
},
{
"epoch": 8.29,
"grad_norm": 0.06618062406778336,
"learning_rate": 4.4997803601633524e-06,
"loss": 0.0006,
"step": 4359
},
{
"epoch": 8.29,
"grad_norm": 0.0846109390258789,
"learning_rate": 4.490054308115483e-06,
"loss": 0.0002,
"step": 4360
},
{
"epoch": 8.29,
"grad_norm": 0.10554458200931549,
"learning_rate": 4.48033792833639e-06,
"loss": 0.0004,
"step": 4361
},
{
"epoch": 8.29,
"grad_norm": 0.02819911390542984,
"learning_rate": 4.470631224510132e-06,
"loss": 0.0002,
"step": 4362
},
{
"epoch": 8.29,
"grad_norm": 0.08534538000822067,
"learning_rate": 4.460934200317059e-06,
"loss": 0.0003,
"step": 4363
},
{
"epoch": 8.3,
"grad_norm": 0.13641533255577087,
"learning_rate": 4.45124685943387e-06,
"loss": 0.0012,
"step": 4364
},
{
"epoch": 8.3,
"grad_norm": 0.0861804410815239,
"learning_rate": 4.441569205533581e-06,
"loss": 0.0003,
"step": 4365
},
{
"epoch": 8.3,
"grad_norm": 0.33863842487335205,
"learning_rate": 4.431901242285542e-06,
"loss": 0.0012,
"step": 4366
},
{
"epoch": 8.3,
"grad_norm": 0.03126806020736694,
"learning_rate": 4.422242973355428e-06,
"loss": 0.0002,
"step": 4367
},
{
"epoch": 8.3,
"grad_norm": 0.1785004884004593,
"learning_rate": 4.412594402405233e-06,
"loss": 0.0004,
"step": 4368
},
{
"epoch": 8.31,
"grad_norm": 0.02725868672132492,
"learning_rate": 4.402955533093285e-06,
"loss": 0.0002,
"step": 4369
},
{
"epoch": 8.31,
"grad_norm": 0.08005505800247192,
"learning_rate": 4.393326369074212e-06,
"loss": 0.0005,
"step": 4370
},
{
"epoch": 8.31,
"grad_norm": 0.03599610552191734,
"learning_rate": 4.383706913999002e-06,
"loss": 0.0003,
"step": 4371
},
{
"epoch": 8.31,
"grad_norm": 0.04188176617026329,
"learning_rate": 4.374097171514906e-06,
"loss": 0.0004,
"step": 4372
},
{
"epoch": 8.31,
"grad_norm": 0.033030569553375244,
"learning_rate": 4.364497145265547e-06,
"loss": 0.0003,
"step": 4373
},
{
"epoch": 8.32,
"grad_norm": 0.06630969792604446,
"learning_rate": 4.354906838890833e-06,
"loss": 0.0008,
"step": 4374
},
{
"epoch": 8.32,
"grad_norm": 0.07822935283184052,
"learning_rate": 4.345326256026989e-06,
"loss": 0.0006,
"step": 4375
},
{
"epoch": 8.32,
"grad_norm": 0.043124835938215256,
"learning_rate": 4.335755400306559e-06,
"loss": 0.0003,
"step": 4376
},
{
"epoch": 8.32,
"grad_norm": 0.44735246896743774,
"learning_rate": 4.326194275358411e-06,
"loss": 0.0017,
"step": 4377
},
{
"epoch": 8.32,
"grad_norm": 0.290990948677063,
"learning_rate": 4.316642884807687e-06,
"loss": 0.001,
"step": 4378
},
{
"epoch": 8.33,
"grad_norm": 0.08410916477441788,
"learning_rate": 4.307101232275877e-06,
"loss": 0.0015,
"step": 4379
},
{
"epoch": 8.33,
"grad_norm": 0.20903731882572174,
"learning_rate": 4.297569321380765e-06,
"loss": 0.0012,
"step": 4380
},
{
"epoch": 8.33,
"grad_norm": 0.16458843648433685,
"learning_rate": 4.288047155736429e-06,
"loss": 0.002,
"step": 4381
},
{
"epoch": 8.33,
"grad_norm": 0.2672274112701416,
"learning_rate": 4.278534738953276e-06,
"loss": 0.0008,
"step": 4382
},
{
"epoch": 8.33,
"grad_norm": 0.13208775222301483,
"learning_rate": 4.269032074637989e-06,
"loss": 0.0004,
"step": 4383
},
{
"epoch": 8.33,
"grad_norm": 0.012308964505791664,
"learning_rate": 4.259539166393579e-06,
"loss": 0.0001,
"step": 4384
},
{
"epoch": 8.34,
"grad_norm": 0.206428661942482,
"learning_rate": 4.250056017819332e-06,
"loss": 0.0004,
"step": 4385
},
{
"epoch": 8.34,
"grad_norm": 0.03625408560037613,
"learning_rate": 4.2405826325108695e-06,
"loss": 0.0004,
"step": 4386
},
{
"epoch": 8.34,
"grad_norm": 0.021182911470532417,
"learning_rate": 4.231119014060067e-06,
"loss": 0.0001,
"step": 4387
},
{
"epoch": 8.34,
"grad_norm": 0.11311204731464386,
"learning_rate": 4.221665166055137e-06,
"loss": 0.0016,
"step": 4388
},
{
"epoch": 8.34,
"grad_norm": 0.04508427530527115,
"learning_rate": 4.212221092080566e-06,
"loss": 0.0002,
"step": 4389
},
{
"epoch": 8.35,
"grad_norm": 0.09324325621128082,
"learning_rate": 4.202786795717135e-06,
"loss": 0.0004,
"step": 4390
},
{
"epoch": 8.35,
"grad_norm": 0.06734511256217957,
"learning_rate": 4.193362280541921e-06,
"loss": 0.001,
"step": 4391
},
{
"epoch": 8.35,
"grad_norm": 0.021975286304950714,
"learning_rate": 4.1839475501282965e-06,
"loss": 0.0002,
"step": 4392
},
{
"epoch": 8.35,
"grad_norm": 0.09909654408693314,
"learning_rate": 4.174542608045915e-06,
"loss": 0.0004,
"step": 4393
},
{
"epoch": 8.35,
"grad_norm": 0.41877198219299316,
"learning_rate": 4.165147457860729e-06,
"loss": 0.0017,
"step": 4394
},
{
"epoch": 8.36,
"grad_norm": 0.10067819803953171,
"learning_rate": 4.155762103134974e-06,
"loss": 0.0004,
"step": 4395
},
{
"epoch": 8.36,
"grad_norm": 0.023712892085313797,
"learning_rate": 4.14638654742716e-06,
"loss": 0.0002,
"step": 4396
},
{
"epoch": 8.36,
"grad_norm": 0.07871938496828079,
"learning_rate": 4.137020794292116e-06,
"loss": 0.0012,
"step": 4397
},
{
"epoch": 8.36,
"grad_norm": 0.5025147795677185,
"learning_rate": 4.127664847280898e-06,
"loss": 0.0023,
"step": 4398
},
{
"epoch": 8.36,
"grad_norm": 0.010690111666917801,
"learning_rate": 4.1183187099409045e-06,
"loss": 0.0001,
"step": 4399
},
{
"epoch": 8.37,
"grad_norm": 0.10837773233652115,
"learning_rate": 4.108982385815776e-06,
"loss": 0.0004,
"step": 4400
},
{
"epoch": 8.37,
"eval_blimp_filtered_avg": 0.7170149253731344,
"eval_blimp_filtered_std": 0.00499489513509044,
"step": 4400
},
{
"epoch": 8.37,
"eval_blimp_supplement_avg": 0.7801724137931034,
"eval_blimp_supplement_std": 0.018094413033767815,
"step": 4400
},
{
"epoch": 8.37,
"eval_vqa_filtered_avg": 0.33,
"eval_vqa_filtered_std": 0.047258156262526045,
"step": 4400
},
{
"epoch": 8.37,
"eval_winoground_filtered_avg": 0.52,
"eval_winoground_filtered_std": 0.05021167315686779,
"step": 4400
},
{
"epoch": 8.37,
"grad_norm": 0.024507205933332443,
"learning_rate": 4.099655878445446e-06,
"loss": 0.0001,
"step": 4401
},
{
"epoch": 8.37,
"grad_norm": 0.025614093989133835,
"learning_rate": 4.090339191366121e-06,
"loss": 0.0002,
"step": 4402
},
{
"epoch": 8.37,
"grad_norm": 0.0246590506285429,
"learning_rate": 4.081032328110283e-06,
"loss": 0.0001,
"step": 4403
},
{
"epoch": 8.37,
"grad_norm": 0.007399612106382847,
"learning_rate": 4.071735292206696e-06,
"loss": 0.0001,
"step": 4404
},
{
"epoch": 8.37,
"grad_norm": 0.33453303575515747,
"learning_rate": 4.062448087180386e-06,
"loss": 0.0012,
"step": 4405
},
{
"epoch": 8.38,
"grad_norm": 0.03863902390003204,
"learning_rate": 4.053170716552681e-06,
"loss": 0.0005,
"step": 4406
},
{
"epoch": 8.38,
"grad_norm": 0.27126938104629517,
"learning_rate": 4.043903183841135e-06,
"loss": 0.0004,
"step": 4407
},
{
"epoch": 8.38,
"grad_norm": 0.24419312179088593,
"learning_rate": 4.034645492559609e-06,
"loss": 0.0009,
"step": 4408
},
{
"epoch": 8.38,
"grad_norm": 0.02669060416519642,
"learning_rate": 4.025397646218217e-06,
"loss": 0.0002,
"step": 4409
},
{
"epoch": 8.38,
"grad_norm": 0.399015337228775,
"learning_rate": 4.016159648323342e-06,
"loss": 0.0048,
"step": 4410
},
{
"epoch": 8.39,
"grad_norm": 0.017507625743746758,
"learning_rate": 4.0069315023776275e-06,
"loss": 0.0001,
"step": 4411
},
{
"epoch": 8.39,
"grad_norm": 0.10646995902061462,
"learning_rate": 3.9977132118800075e-06,
"loss": 0.0004,
"step": 4412
},
{
"epoch": 8.39,
"grad_norm": 0.053829167038202286,
"learning_rate": 3.988504780325636e-06,
"loss": 0.0004,
"step": 4413
},
{
"epoch": 8.39,
"grad_norm": 0.22759847342967987,
"learning_rate": 3.979306211205963e-06,
"loss": 0.0007,
"step": 4414
},
{
"epoch": 8.39,
"grad_norm": 0.558752179145813,
"learning_rate": 3.970117508008686e-06,
"loss": 0.0033,
"step": 4415
},
{
"epoch": 8.4,
"grad_norm": 0.08775019645690918,
"learning_rate": 3.960938674217758e-06,
"loss": 0.0009,
"step": 4416
},
{
"epoch": 8.4,
"grad_norm": 0.02771660126745701,
"learning_rate": 3.951769713313418e-06,
"loss": 0.0002,
"step": 4417
},
{
"epoch": 8.4,
"grad_norm": 0.10747211426496506,
"learning_rate": 3.942610628772112e-06,
"loss": 0.0005,
"step": 4418
},
{
"epoch": 8.4,
"grad_norm": 0.06290878355503082,
"learning_rate": 3.933461424066586e-06,
"loss": 0.0002,
"step": 4419
},
{
"epoch": 8.4,
"grad_norm": 0.023817501962184906,
"learning_rate": 3.92432210266582e-06,
"loss": 0.0002,
"step": 4420
},
{
"epoch": 8.4,
"grad_norm": 0.07082189619541168,
"learning_rate": 3.915192668035048e-06,
"loss": 0.0006,
"step": 4421
},
{
"epoch": 8.41,
"grad_norm": 0.13954636454582214,
"learning_rate": 3.90607312363575e-06,
"loss": 0.0007,
"step": 4422
},
{
"epoch": 8.41,
"grad_norm": 0.041713714599609375,
"learning_rate": 3.896963472925686e-06,
"loss": 0.0004,
"step": 4423
},
{
"epoch": 8.41,
"grad_norm": 0.033771149814128876,
"learning_rate": 3.88786371935881e-06,
"loss": 0.0001,
"step": 4424
},
{
"epoch": 8.41,
"grad_norm": 0.26919975876808167,
"learning_rate": 3.878773866385379e-06,
"loss": 0.0014,
"step": 4425
},
{
"epoch": 8.41,
"grad_norm": 0.14341183006763458,
"learning_rate": 3.869693917451863e-06,
"loss": 0.0005,
"step": 4426
},
{
"epoch": 8.42,
"grad_norm": 0.015739472582936287,
"learning_rate": 3.860623876000992e-06,
"loss": 0.0001,
"step": 4427
},
{
"epoch": 8.42,
"grad_norm": 0.02188861183822155,
"learning_rate": 3.851563745471727e-06,
"loss": 0.0002,
"step": 4428
},
{
"epoch": 8.42,
"grad_norm": 0.01673879288136959,
"learning_rate": 3.842513529299277e-06,
"loss": 0.0001,
"step": 4429
},
{
"epoch": 8.42,
"grad_norm": 0.1175539568066597,
"learning_rate": 3.8334732309151025e-06,
"loss": 0.0002,
"step": 4430
},
{
"epoch": 8.42,
"grad_norm": 0.12601719796657562,
"learning_rate": 3.824442853746879e-06,
"loss": 0.0003,
"step": 4431
},
{
"epoch": 8.43,
"grad_norm": 0.10789843648672104,
"learning_rate": 3.815422401218556e-06,
"loss": 0.0004,
"step": 4432
},
{
"epoch": 8.43,
"grad_norm": 0.20055601000785828,
"learning_rate": 3.806411876750273e-06,
"loss": 0.0008,
"step": 4433
},
{
"epoch": 8.43,
"grad_norm": 0.07943399995565414,
"learning_rate": 3.797411283758455e-06,
"loss": 0.0005,
"step": 4434
},
{
"epoch": 8.43,
"grad_norm": 0.17282743752002716,
"learning_rate": 3.7884206256557252e-06,
"loss": 0.0008,
"step": 4435
},
{
"epoch": 8.43,
"grad_norm": 0.07523642480373383,
"learning_rate": 3.7794399058509564e-06,
"loss": 0.0003,
"step": 4436
},
{
"epoch": 8.44,
"grad_norm": 0.005846839398145676,
"learning_rate": 3.770469127749252e-06,
"loss": 0.0001,
"step": 4437
},
{
"epoch": 8.44,
"grad_norm": 0.08751727640628815,
"learning_rate": 3.7615082947519375e-06,
"loss": 0.0005,
"step": 4438
},
{
"epoch": 8.44,
"grad_norm": 0.1628558188676834,
"learning_rate": 3.7525574102565753e-06,
"loss": 0.0008,
"step": 4439
},
{
"epoch": 8.44,
"grad_norm": 0.07722613215446472,
"learning_rate": 3.7436164776569537e-06,
"loss": 0.0002,
"step": 4440
},
{
"epoch": 8.44,
"grad_norm": 0.053356923162937164,
"learning_rate": 3.7346855003430915e-06,
"loss": 0.0003,
"step": 4441
},
{
"epoch": 8.44,
"grad_norm": 0.1179957464337349,
"learning_rate": 3.7257644817012194e-06,
"loss": 0.0008,
"step": 4442
},
{
"epoch": 8.45,
"grad_norm": 0.17323695123195648,
"learning_rate": 3.716853425113822e-06,
"loss": 0.0017,
"step": 4443
},
{
"epoch": 8.45,
"grad_norm": 0.04241637885570526,
"learning_rate": 3.70795233395956e-06,
"loss": 0.0004,
"step": 4444
},
{
"epoch": 8.45,
"grad_norm": 0.027324091643095016,
"learning_rate": 3.6990612116133635e-06,
"loss": 0.0003,
"step": 4445
},
{
"epoch": 8.45,
"grad_norm": 0.08928991854190826,
"learning_rate": 3.690180061446353e-06,
"loss": 0.0004,
"step": 4446
},
{
"epoch": 8.45,
"grad_norm": 0.11912596225738525,
"learning_rate": 3.6813088868258813e-06,
"loss": 0.0004,
"step": 4447
},
{
"epoch": 8.46,
"grad_norm": 0.07597754895687103,
"learning_rate": 3.6724476911155112e-06,
"loss": 0.0005,
"step": 4448
},
{
"epoch": 8.46,
"grad_norm": 0.023825405165553093,
"learning_rate": 3.6635964776750253e-06,
"loss": 0.0002,
"step": 4449
},
{
"epoch": 8.46,
"grad_norm": 0.04346529766917229,
"learning_rate": 3.6547552498604185e-06,
"loss": 0.0003,
"step": 4450
},
{
"epoch": 8.46,
"grad_norm": 0.26598677039146423,
"learning_rate": 3.6459240110239033e-06,
"loss": 0.0026,
"step": 4451
},
{
"epoch": 8.46,
"grad_norm": 0.18180295825004578,
"learning_rate": 3.6371027645139145e-06,
"loss": 0.0006,
"step": 4452
},
{
"epoch": 8.47,
"grad_norm": 0.24721390008926392,
"learning_rate": 3.6282915136750692e-06,
"loss": 0.0008,
"step": 4453
},
{
"epoch": 8.47,
"grad_norm": 0.27238011360168457,
"learning_rate": 3.6194902618482264e-06,
"loss": 0.001,
"step": 4454
},
{
"epoch": 8.47,
"grad_norm": 0.018393712118268013,
"learning_rate": 3.6106990123704353e-06,
"loss": 0.0002,
"step": 4455
},
{
"epoch": 8.47,
"grad_norm": 0.039847031235694885,
"learning_rate": 3.6019177685749603e-06,
"loss": 0.0003,
"step": 4456
},
{
"epoch": 8.47,
"grad_norm": 0.012090745382010937,
"learning_rate": 3.593146533791262e-06,
"loss": 0.0001,
"step": 4457
},
{
"epoch": 8.48,
"grad_norm": 0.1930002123117447,
"learning_rate": 3.584385311345029e-06,
"loss": 0.0003,
"step": 4458
},
{
"epoch": 8.48,
"grad_norm": 0.583185076713562,
"learning_rate": 3.575634104558124e-06,
"loss": 0.0029,
"step": 4459
},
{
"epoch": 8.48,
"grad_norm": 0.17912626266479492,
"learning_rate": 3.5668929167486274e-06,
"loss": 0.0016,
"step": 4460
},
{
"epoch": 8.48,
"grad_norm": 0.05676186829805374,
"learning_rate": 3.558161751230831e-06,
"loss": 0.0004,
"step": 4461
},
{
"epoch": 8.48,
"grad_norm": 0.021056978031992912,
"learning_rate": 3.549440611315199e-06,
"loss": 0.0001,
"step": 4462
},
{
"epoch": 8.48,
"grad_norm": 0.09392949193716049,
"learning_rate": 3.5407295003084306e-06,
"loss": 0.0015,
"step": 4463
},
{
"epoch": 8.49,
"grad_norm": 0.2580009698867798,
"learning_rate": 3.532028421513377e-06,
"loss": 0.0011,
"step": 4464
},
{
"epoch": 8.49,
"grad_norm": 0.2179614156484604,
"learning_rate": 3.5233373782291322e-06,
"loss": 0.0021,
"step": 4465
},
{
"epoch": 8.49,
"grad_norm": 0.03921540454030037,
"learning_rate": 3.5146563737509574e-06,
"loss": 0.0002,
"step": 4466
},
{
"epoch": 8.49,
"grad_norm": 0.014690748415887356,
"learning_rate": 3.505985411370316e-06,
"loss": 0.0002,
"step": 4467
},
{
"epoch": 8.49,
"grad_norm": 0.09834066033363342,
"learning_rate": 3.497324494374855e-06,
"loss": 0.0016,
"step": 4468
},
{
"epoch": 8.5,
"grad_norm": 0.01740516722202301,
"learning_rate": 3.488673626048434e-06,
"loss": 0.0001,
"step": 4469
},
{
"epoch": 8.5,
"grad_norm": 0.11252184212207794,
"learning_rate": 3.48003280967107e-06,
"loss": 0.0009,
"step": 4470
},
{
"epoch": 8.5,
"grad_norm": 0.0747825875878334,
"learning_rate": 3.4714020485190048e-06,
"loss": 0.0003,
"step": 4471
},
{
"epoch": 8.5,
"grad_norm": 0.05333748832345009,
"learning_rate": 3.4627813458646374e-06,
"loss": 0.0002,
"step": 4472
},
{
"epoch": 8.5,
"grad_norm": 0.005126305390149355,
"learning_rate": 3.4541707049765734e-06,
"loss": 0.0001,
"step": 4473
},
{
"epoch": 8.51,
"grad_norm": 0.006881397683173418,
"learning_rate": 3.4455701291195884e-06,
"loss": 0.0001,
"step": 4474
},
{
"epoch": 8.51,
"grad_norm": 0.010040461085736752,
"learning_rate": 3.4369796215546566e-06,
"loss": 0.0001,
"step": 4475
},
{
"epoch": 8.51,
"grad_norm": 0.026476487517356873,
"learning_rate": 3.428399185538923e-06,
"loss": 0.0002,
"step": 4476
},
{
"epoch": 8.51,
"grad_norm": 0.23119160532951355,
"learning_rate": 3.4198288243257115e-06,
"loss": 0.0006,
"step": 4477
},
{
"epoch": 8.51,
"grad_norm": 0.1395696997642517,
"learning_rate": 3.4112685411645526e-06,
"loss": 0.0013,
"step": 4478
},
{
"epoch": 8.52,
"grad_norm": 0.04958971589803696,
"learning_rate": 3.4027183393011107e-06,
"loss": 0.0005,
"step": 4479
},
{
"epoch": 8.52,
"grad_norm": 0.25696861743927,
"learning_rate": 3.394178221977272e-06,
"loss": 0.0006,
"step": 4480
},
{
"epoch": 8.52,
"grad_norm": 0.26771190762519836,
"learning_rate": 3.3856481924310734e-06,
"loss": 0.0005,
"step": 4481
},
{
"epoch": 8.52,
"grad_norm": 0.052304767072200775,
"learning_rate": 3.3771282538967306e-06,
"loss": 0.0002,
"step": 4482
},
{
"epoch": 8.52,
"grad_norm": 0.08709216862916946,
"learning_rate": 3.36861840960464e-06,
"loss": 0.0005,
"step": 4483
},
{
"epoch": 8.52,
"grad_norm": 0.034308455884456635,
"learning_rate": 3.360118662781364e-06,
"loss": 0.0003,
"step": 4484
},
{
"epoch": 8.53,
"grad_norm": 0.10619954019784927,
"learning_rate": 3.351629016649638e-06,
"loss": 0.0007,
"step": 4485
},
{
"epoch": 8.53,
"grad_norm": 0.13072478771209717,
"learning_rate": 3.3431494744283665e-06,
"loss": 0.0005,
"step": 4486
},
{
"epoch": 8.53,
"grad_norm": 0.3545732796192169,
"learning_rate": 3.334680039332645e-06,
"loss": 0.0011,
"step": 4487
},
{
"epoch": 8.53,
"grad_norm": 0.1802055388689041,
"learning_rate": 3.3262207145736878e-06,
"loss": 0.0008,
"step": 4488
},
{
"epoch": 8.53,
"grad_norm": 0.038682807236909866,
"learning_rate": 3.3177715033589297e-06,
"loss": 0.0002,
"step": 4489
},
{
"epoch": 8.54,
"grad_norm": 0.011632559821009636,
"learning_rate": 3.3093324088919274e-06,
"loss": 0.0001,
"step": 4490
},
{
"epoch": 8.54,
"grad_norm": 0.2825629413127899,
"learning_rate": 3.300903434372432e-06,
"loss": 0.0012,
"step": 4491
},
{
"epoch": 8.54,
"grad_norm": 0.04558581858873367,
"learning_rate": 3.2924845829963466e-06,
"loss": 0.0002,
"step": 4492
},
{
"epoch": 8.54,
"grad_norm": 0.26057547330856323,
"learning_rate": 3.2840758579557352e-06,
"loss": 0.001,
"step": 4493
},
{
"epoch": 8.54,
"grad_norm": 0.01179661974310875,
"learning_rate": 3.2756772624388166e-06,
"loss": 0.0001,
"step": 4494
},
{
"epoch": 8.55,
"grad_norm": 0.04299997165799141,
"learning_rate": 3.2672887996299805e-06,
"loss": 0.0004,
"step": 4495
},
{
"epoch": 8.55,
"grad_norm": 0.13504189252853394,
"learning_rate": 3.2589104727097697e-06,
"loss": 0.0005,
"step": 4496
},
{
"epoch": 8.55,
"grad_norm": 0.021305272355675697,
"learning_rate": 3.250542284854876e-06,
"loss": 0.0002,
"step": 4497
},
{
"epoch": 8.55,
"grad_norm": 0.6632449626922607,
"learning_rate": 3.242184239238172e-06,
"loss": 0.0021,
"step": 4498
},
{
"epoch": 8.55,
"grad_norm": 0.254896879196167,
"learning_rate": 3.233836339028645e-06,
"loss": 0.0005,
"step": 4499
},
{
"epoch": 8.56,
"grad_norm": 0.1352105587720871,
"learning_rate": 3.2254985873914744e-06,
"loss": 0.001,
"step": 4500
},
{
"epoch": 8.56,
"eval_blimp_filtered_avg": 0.7170149253731344,
"eval_blimp_filtered_std": 0.004996561684154108,
"step": 4500
},
{
"epoch": 8.56,
"eval_blimp_supplement_avg": 0.7801724137931034,
"eval_blimp_supplement_std": 0.018026818725829057,
"step": 4500
},
{
"epoch": 8.56,
"eval_vqa_filtered_avg": 0.32,
"eval_vqa_filtered_std": 0.046882617226215034,
"step": 4500
},
{
"epoch": 8.56,
"eval_winoground_filtered_avg": 0.51,
"eval_winoground_filtered_std": 0.05024183937956912,
"step": 4500
},
{
"epoch": 8.56,
"grad_norm": 0.02473735436797142,
"learning_rate": 3.217170987487971e-06,
"loss": 0.0002,
"step": 4501
},
{
"epoch": 8.56,
"grad_norm": 0.30320021510124207,
"learning_rate": 3.208853542475598e-06,
"loss": 0.0011,
"step": 4502
},
{
"epoch": 8.56,
"grad_norm": 0.058982595801353455,
"learning_rate": 3.200546255507968e-06,
"loss": 0.0006,
"step": 4503
},
{
"epoch": 8.56,
"grad_norm": 0.046631861478090286,
"learning_rate": 3.19224912973486e-06,
"loss": 0.0002,
"step": 4504
},
{
"epoch": 8.56,
"grad_norm": 0.42206522822380066,
"learning_rate": 3.1839621683021713e-06,
"loss": 0.0045,
"step": 4505
},
{
"epoch": 8.57,
"grad_norm": 0.3617534935474396,
"learning_rate": 3.1756853743519587e-06,
"loss": 0.0013,
"step": 4506
},
{
"epoch": 8.57,
"grad_norm": 0.0903761237859726,
"learning_rate": 3.1674187510224382e-06,
"loss": 0.0002,
"step": 4507
},
{
"epoch": 8.57,
"grad_norm": 0.7055791020393372,
"learning_rate": 3.1591623014479365e-06,
"loss": 0.0027,
"step": 4508
},
{
"epoch": 8.57,
"grad_norm": 0.12981247901916504,
"learning_rate": 3.1509160287589588e-06,
"loss": 0.0007,
"step": 4509
},
{
"epoch": 8.57,
"grad_norm": 0.03810209780931473,
"learning_rate": 3.1426799360821313e-06,
"loss": 0.0002,
"step": 4510
},
{
"epoch": 8.58,
"grad_norm": 0.2933924198150635,
"learning_rate": 3.134454026540222e-06,
"loss": 0.0005,
"step": 4511
},
{
"epoch": 8.58,
"grad_norm": 0.006710784509778023,
"learning_rate": 3.1262383032521434e-06,
"loss": 0.0001,
"step": 4512
},
{
"epoch": 8.58,
"grad_norm": 0.020868832245469093,
"learning_rate": 3.118032769332937e-06,
"loss": 0.0001,
"step": 4513
},
{
"epoch": 8.58,
"grad_norm": 0.06272855401039124,
"learning_rate": 3.109837427893787e-06,
"loss": 0.0003,
"step": 4514
},
{
"epoch": 8.58,
"grad_norm": 0.12084665894508362,
"learning_rate": 3.1016522820420245e-06,
"loss": 0.0005,
"step": 4515
},
{
"epoch": 8.59,
"grad_norm": 0.6454626321792603,
"learning_rate": 3.093477334881084e-06,
"loss": 0.0018,
"step": 4516
},
{
"epoch": 8.59,
"grad_norm": 0.19851770997047424,
"learning_rate": 3.0853125895105662e-06,
"loss": 0.001,
"step": 4517
},
{
"epoch": 8.59,
"grad_norm": 0.2768668830394745,
"learning_rate": 3.0771580490261874e-06,
"loss": 0.0008,
"step": 4518
},
{
"epoch": 8.59,
"grad_norm": 0.07741827517747879,
"learning_rate": 3.0690137165197953e-06,
"loss": 0.0004,
"step": 4519
},
{
"epoch": 8.59,
"grad_norm": 0.07333717495203018,
"learning_rate": 3.0608795950793687e-06,
"loss": 0.0004,
"step": 4520
},
{
"epoch": 8.6,
"grad_norm": 0.06678545475006104,
"learning_rate": 3.052755687789015e-06,
"loss": 0.0002,
"step": 4521
},
{
"epoch": 8.6,
"grad_norm": 0.2920079827308655,
"learning_rate": 3.0446419977289665e-06,
"loss": 0.0005,
"step": 4522
},
{
"epoch": 8.6,
"grad_norm": 0.5785551071166992,
"learning_rate": 3.036538527975582e-06,
"loss": 0.0013,
"step": 4523
},
{
"epoch": 8.6,
"grad_norm": 0.03338746353983879,
"learning_rate": 3.0284452816013607e-06,
"loss": 0.0002,
"step": 4524
},
{
"epoch": 8.6,
"grad_norm": 0.06680819392204285,
"learning_rate": 3.0203622616748914e-06,
"loss": 0.0006,
"step": 4525
},
{
"epoch": 8.6,
"grad_norm": 0.07284952700138092,
"learning_rate": 3.012289471260923e-06,
"loss": 0.0002,
"step": 4526
},
{
"epoch": 8.61,
"grad_norm": 0.08917441219091415,
"learning_rate": 3.0042269134202995e-06,
"loss": 0.0003,
"step": 4527
},
{
"epoch": 8.61,
"grad_norm": 0.013458757661283016,
"learning_rate": 2.996174591209999e-06,
"loss": 0.0001,
"step": 4528
},
{
"epoch": 8.61,
"grad_norm": 0.03767577186226845,
"learning_rate": 2.98813250768311e-06,
"loss": 0.0001,
"step": 4529
},
{
"epoch": 8.61,
"grad_norm": 0.04716615751385689,
"learning_rate": 2.980100665888846e-06,
"loss": 0.0004,
"step": 4530
},
{
"epoch": 8.61,
"grad_norm": 0.21663503348827362,
"learning_rate": 2.972079068872535e-06,
"loss": 0.0015,
"step": 4531
},
{
"epoch": 8.62,
"grad_norm": 0.021853242069482803,
"learning_rate": 2.9640677196756104e-06,
"loss": 0.0001,
"step": 4532
},
{
"epoch": 8.62,
"grad_norm": 0.02178061567246914,
"learning_rate": 2.956066621335649e-06,
"loss": 0.0002,
"step": 4533
},
{
"epoch": 8.62,
"grad_norm": 0.05219707638025284,
"learning_rate": 2.9480757768862976e-06,
"loss": 0.0003,
"step": 4534
},
{
"epoch": 8.62,
"grad_norm": 0.08680331707000732,
"learning_rate": 2.9400951893573592e-06,
"loss": 0.001,
"step": 4535
},
{
"epoch": 8.62,
"grad_norm": 0.04587429389357567,
"learning_rate": 2.9321248617747164e-06,
"loss": 0.0004,
"step": 4536
},
{
"epoch": 8.63,
"grad_norm": 0.015058638527989388,
"learning_rate": 2.924164797160376e-06,
"loss": 0.0001,
"step": 4537
},
{
"epoch": 8.63,
"grad_norm": 0.1961059272289276,
"learning_rate": 2.9162149985324527e-06,
"loss": 0.0015,
"step": 4538
},
{
"epoch": 8.63,
"grad_norm": 0.3084276020526886,
"learning_rate": 2.908275468905163e-06,
"loss": 0.0039,
"step": 4539
},
{
"epoch": 8.63,
"grad_norm": 0.1653963178396225,
"learning_rate": 2.9003462112888344e-06,
"loss": 0.0025,
"step": 4540
},
{
"epoch": 8.63,
"grad_norm": 0.00692038144916296,
"learning_rate": 2.8924272286899e-06,
"loss": 0.0001,
"step": 4541
},
{
"epoch": 8.63,
"grad_norm": 0.6074125170707703,
"learning_rate": 2.884518524110894e-06,
"loss": 0.0016,
"step": 4542
},
{
"epoch": 8.64,
"grad_norm": 0.02031126618385315,
"learning_rate": 2.876620100550451e-06,
"loss": 0.0001,
"step": 4543
},
{
"epoch": 8.64,
"grad_norm": 0.04634946212172508,
"learning_rate": 2.8687319610033292e-06,
"loss": 0.0003,
"step": 4544
},
{
"epoch": 8.64,
"grad_norm": 0.008682692423462868,
"learning_rate": 2.8608541084603425e-06,
"loss": 0.0,
"step": 4545
},
{
"epoch": 8.64,
"grad_norm": 0.01898583583533764,
"learning_rate": 2.852986545908458e-06,
"loss": 0.0002,
"step": 4546
},
{
"epoch": 8.64,
"grad_norm": 0.07521553337574005,
"learning_rate": 2.8451292763307003e-06,
"loss": 0.0004,
"step": 4547
},
{
"epoch": 8.65,
"grad_norm": 0.023751555010676384,
"learning_rate": 2.8372823027062123e-06,
"loss": 0.0001,
"step": 4548
},
{
"epoch": 8.65,
"grad_norm": 0.6449877619743347,
"learning_rate": 2.829445628010221e-06,
"loss": 0.0011,
"step": 4549
},
{
"epoch": 8.65,
"grad_norm": 0.10660184174776077,
"learning_rate": 2.8216192552140697e-06,
"loss": 0.0005,
"step": 4550
},
{
"epoch": 8.65,
"grad_norm": 0.008558154106140137,
"learning_rate": 2.813803187285168e-06,
"loss": 0.0001,
"step": 4551
},
{
"epoch": 8.65,
"grad_norm": 0.16370873153209686,
"learning_rate": 2.805997427187025e-06,
"loss": 0.0005,
"step": 4552
},
{
"epoch": 8.66,
"grad_norm": 0.07830721884965897,
"learning_rate": 2.798201977879267e-06,
"loss": 0.0006,
"step": 4553
},
{
"epoch": 8.66,
"grad_norm": 0.02952200174331665,
"learning_rate": 2.790416842317569e-06,
"loss": 0.0001,
"step": 4554
},
{
"epoch": 8.66,
"grad_norm": 0.16262687742710114,
"learning_rate": 2.7826420234537343e-06,
"loss": 0.0005,
"step": 4555
},
{
"epoch": 8.66,
"grad_norm": 0.10451652854681015,
"learning_rate": 2.774877524235637e-06,
"loss": 0.0012,
"step": 4556
},
{
"epoch": 8.66,
"grad_norm": 0.21987134218215942,
"learning_rate": 2.767123347607233e-06,
"loss": 0.0007,
"step": 4557
},
{
"epoch": 8.67,
"grad_norm": 0.09059207886457443,
"learning_rate": 2.7593794965085674e-06,
"loss": 0.0008,
"step": 4558
},
{
"epoch": 8.67,
"grad_norm": 0.025645047426223755,
"learning_rate": 2.751645973875794e-06,
"loss": 0.0001,
"step": 4559
},
{
"epoch": 8.67,
"grad_norm": 0.018971215933561325,
"learning_rate": 2.743922782641104e-06,
"loss": 0.0001,
"step": 4560
},
{
"epoch": 8.67,
"grad_norm": 0.21706421673297882,
"learning_rate": 2.736209925732822e-06,
"loss": 0.0004,
"step": 4561
},
{
"epoch": 8.67,
"grad_norm": 0.10453557223081589,
"learning_rate": 2.7285074060753057e-06,
"loss": 0.0021,
"step": 4562
},
{
"epoch": 8.67,
"grad_norm": 0.3545166254043579,
"learning_rate": 2.720815226589035e-06,
"loss": 0.0013,
"step": 4563
},
{
"epoch": 8.68,
"grad_norm": 0.026478098705410957,
"learning_rate": 2.7131333901905508e-06,
"loss": 0.0001,
"step": 4564
},
{
"epoch": 8.68,
"grad_norm": 0.061124518513679504,
"learning_rate": 2.7054618997924664e-06,
"loss": 0.0002,
"step": 4565
},
{
"epoch": 8.68,
"grad_norm": 0.2160719335079193,
"learning_rate": 2.6978007583034845e-06,
"loss": 0.0017,
"step": 4566
},
{
"epoch": 8.68,
"grad_norm": 0.031818147748708725,
"learning_rate": 2.690149968628376e-06,
"loss": 0.0001,
"step": 4567
},
{
"epoch": 8.68,
"grad_norm": 0.4018324911594391,
"learning_rate": 2.6825095336679896e-06,
"loss": 0.0015,
"step": 4568
},
{
"epoch": 8.69,
"grad_norm": 0.04355862736701965,
"learning_rate": 2.6748794563192446e-06,
"loss": 0.0002,
"step": 4569
},
{
"epoch": 8.69,
"grad_norm": 0.0923234149813652,
"learning_rate": 2.6672597394751508e-06,
"loss": 0.0002,
"step": 4570
},
{
"epoch": 8.69,
"grad_norm": 0.03620320186018944,
"learning_rate": 2.6596503860247566e-06,
"loss": 0.0002,
"step": 4571
},
{
"epoch": 8.69,
"grad_norm": 0.04653416946530342,
"learning_rate": 2.6520513988532113e-06,
"loss": 0.0002,
"step": 4572
},
{
"epoch": 8.69,
"grad_norm": 0.0436859130859375,
"learning_rate": 2.6444627808417232e-06,
"loss": 0.0002,
"step": 4573
},
{
"epoch": 8.7,
"grad_norm": 0.13702163100242615,
"learning_rate": 2.636884534867564e-06,
"loss": 0.0009,
"step": 4574
},
{
"epoch": 8.7,
"grad_norm": 0.17420293390750885,
"learning_rate": 2.629316663804079e-06,
"loss": 0.0005,
"step": 4575
},
{
"epoch": 8.7,
"grad_norm": 0.05962143465876579,
"learning_rate": 2.6217591705206758e-06,
"loss": 0.0006,
"step": 4576
},
{
"epoch": 8.7,
"grad_norm": 0.018487969413399696,
"learning_rate": 2.614212057882829e-06,
"loss": 0.0001,
"step": 4577
},
{
"epoch": 8.7,
"grad_norm": 0.10738100856542587,
"learning_rate": 2.6066753287520728e-06,
"loss": 0.0003,
"step": 4578
},
{
"epoch": 8.71,
"grad_norm": 0.014770083129405975,
"learning_rate": 2.5991489859860285e-06,
"loss": 0.0001,
"step": 4579
},
{
"epoch": 8.71,
"grad_norm": 0.016990412026643753,
"learning_rate": 2.5916330324383364e-06,
"loss": 0.0001,
"step": 4580
},
{
"epoch": 8.71,
"grad_norm": 0.17886172235012054,
"learning_rate": 2.5841274709587336e-06,
"loss": 0.0007,
"step": 4581
},
{
"epoch": 8.71,
"grad_norm": 0.10688559710979462,
"learning_rate": 2.576632304393004e-06,
"loss": 0.001,
"step": 4582
},
{
"epoch": 8.71,
"grad_norm": 0.18494728207588196,
"learning_rate": 2.5691475355829906e-06,
"loss": 0.0008,
"step": 4583
},
{
"epoch": 8.71,
"grad_norm": 0.01251134742051363,
"learning_rate": 2.5616731673665905e-06,
"loss": 0.0001,
"step": 4584
},
{
"epoch": 8.72,
"grad_norm": 0.057323407381772995,
"learning_rate": 2.5542092025777664e-06,
"loss": 0.0004,
"step": 4585
},
{
"epoch": 8.72,
"grad_norm": 0.602567732334137,
"learning_rate": 2.5467556440465245e-06,
"loss": 0.0019,
"step": 4586
},
{
"epoch": 8.72,
"grad_norm": 0.021107418462634087,
"learning_rate": 2.5393124945989376e-06,
"loss": 0.0002,
"step": 4587
},
{
"epoch": 8.72,
"grad_norm": 0.0848756805062294,
"learning_rate": 2.531879757057125e-06,
"loss": 0.0006,
"step": 4588
},
{
"epoch": 8.72,
"grad_norm": 0.00923674926161766,
"learning_rate": 2.524457434239252e-06,
"loss": 0.0001,
"step": 4589
},
{
"epoch": 8.73,
"grad_norm": 0.0838325172662735,
"learning_rate": 2.5170455289595585e-06,
"loss": 0.0005,
"step": 4590
},
{
"epoch": 8.73,
"grad_norm": 0.005768090486526489,
"learning_rate": 2.509644044028302e-06,
"loss": 0.0,
"step": 4591
},
{
"epoch": 8.73,
"grad_norm": 0.529915452003479,
"learning_rate": 2.502252982251818e-06,
"loss": 0.0062,
"step": 4592
},
{
"epoch": 8.73,
"grad_norm": 0.022304901853203773,
"learning_rate": 2.494872346432472e-06,
"loss": 0.0001,
"step": 4593
},
{
"epoch": 8.73,
"grad_norm": 0.02718343771994114,
"learning_rate": 2.4875021393686848e-06,
"loss": 0.0001,
"step": 4594
},
{
"epoch": 8.74,
"grad_norm": 0.02343880571424961,
"learning_rate": 2.4801423638549114e-06,
"loss": 0.0002,
"step": 4595
},
{
"epoch": 8.74,
"grad_norm": 0.6256465911865234,
"learning_rate": 2.472793022681676e-06,
"loss": 0.0029,
"step": 4596
},
{
"epoch": 8.74,
"grad_norm": 0.012249810621142387,
"learning_rate": 2.4654541186355205e-06,
"loss": 0.0001,
"step": 4597
},
{
"epoch": 8.74,
"grad_norm": 0.10077741742134094,
"learning_rate": 2.4581256544990383e-06,
"loss": 0.001,
"step": 4598
},
{
"epoch": 8.74,
"grad_norm": 0.006806628778576851,
"learning_rate": 2.45080763305088e-06,
"loss": 0.0001,
"step": 4599
},
{
"epoch": 8.75,
"grad_norm": 0.05678814649581909,
"learning_rate": 2.4435000570656998e-06,
"loss": 0.0005,
"step": 4600
},
{
"epoch": 8.75,
"eval_blimp_filtered_avg": 0.7170149253731344,
"eval_blimp_filtered_std": 0.0049906586967288806,
"step": 4600
},
{
"epoch": 8.75,
"eval_blimp_supplement_avg": 0.7780172413793104,
"eval_blimp_supplement_std": 0.018094331942175326,
"step": 4600
},
{
"epoch": 8.75,
"eval_vqa_filtered_avg": 0.33,
"eval_vqa_filtered_std": 0.047258156262526045,
"step": 4600
},
{
"epoch": 8.75,
"eval_winoground_filtered_avg": 0.52,
"eval_winoground_filtered_std": 0.05021167315686779,
"step": 4600
},
{
"epoch": 8.75,
"grad_norm": 0.09178295731544495,
"learning_rate": 2.4362029293142374e-06,
"loss": 0.0012,
"step": 4601
},
{
"epoch": 8.75,
"grad_norm": 0.05759723484516144,
"learning_rate": 2.428916252563237e-06,
"loss": 0.0003,
"step": 4602
},
{
"epoch": 8.75,
"grad_norm": 0.3394083082675934,
"learning_rate": 2.4216400295754946e-06,
"loss": 0.003,
"step": 4603
},
{
"epoch": 8.75,
"grad_norm": 0.014159739948809147,
"learning_rate": 2.414374263109831e-06,
"loss": 0.0001,
"step": 4604
},
{
"epoch": 8.75,
"grad_norm": 0.007393274921923876,
"learning_rate": 2.4071189559211314e-06,
"loss": 0.0001,
"step": 4605
},
{
"epoch": 8.76,
"grad_norm": 0.5170931816101074,
"learning_rate": 2.3998741107602697e-06,
"loss": 0.0016,
"step": 4606
},
{
"epoch": 8.76,
"grad_norm": 0.11731146275997162,
"learning_rate": 2.3926397303741943e-06,
"loss": 0.0004,
"step": 4607
},
{
"epoch": 8.76,
"grad_norm": 0.025873156264424324,
"learning_rate": 2.385415817505868e-06,
"loss": 0.0002,
"step": 4608
},
{
"epoch": 8.76,
"grad_norm": 0.11595389991998672,
"learning_rate": 2.3782023748942803e-06,
"loss": 0.0005,
"step": 4609
},
{
"epoch": 8.76,
"grad_norm": 0.009126797318458557,
"learning_rate": 2.37099940527446e-06,
"loss": 0.0001,
"step": 4610
},
{
"epoch": 8.77,
"grad_norm": 0.20593422651290894,
"learning_rate": 2.3638069113774653e-06,
"loss": 0.0013,
"step": 4611
},
{
"epoch": 8.77,
"grad_norm": 0.1504182368516922,
"learning_rate": 2.3566248959303715e-06,
"loss": 0.0014,
"step": 4612
},
{
"epoch": 8.77,
"grad_norm": 0.05412362515926361,
"learning_rate": 2.349453361656294e-06,
"loss": 0.0001,
"step": 4613
},
{
"epoch": 8.77,
"grad_norm": 0.36349114775657654,
"learning_rate": 2.3422923112743677e-06,
"loss": 0.0025,
"step": 4614
},
{
"epoch": 8.77,
"grad_norm": 0.12839463353157043,
"learning_rate": 2.335141747499747e-06,
"loss": 0.0002,
"step": 4615
},
{
"epoch": 8.78,
"grad_norm": 0.06529521942138672,
"learning_rate": 2.3280016730436337e-06,
"loss": 0.0002,
"step": 4616
},
{
"epoch": 8.78,
"grad_norm": 0.47023525834083557,
"learning_rate": 2.320872090613212e-06,
"loss": 0.0017,
"step": 4617
},
{
"epoch": 8.78,
"grad_norm": 0.016715897247195244,
"learning_rate": 2.3137530029117295e-06,
"loss": 0.0001,
"step": 4618
},
{
"epoch": 8.78,
"grad_norm": 0.13880613446235657,
"learning_rate": 2.30664441263843e-06,
"loss": 0.0005,
"step": 4619
},
{
"epoch": 8.78,
"grad_norm": 0.06809261441230774,
"learning_rate": 2.299546322488585e-06,
"loss": 0.0003,
"step": 4620
},
{
"epoch": 8.79,
"grad_norm": 0.07443056255578995,
"learning_rate": 2.292458735153481e-06,
"loss": 0.0003,
"step": 4621
},
{
"epoch": 8.79,
"grad_norm": 0.48374849557876587,
"learning_rate": 2.2853816533204288e-06,
"loss": 0.0009,
"step": 4622
},
{
"epoch": 8.79,
"grad_norm": 0.5177217125892639,
"learning_rate": 2.278315079672748e-06,
"loss": 0.0039,
"step": 4623
},
{
"epoch": 8.79,
"grad_norm": 0.15317504107952118,
"learning_rate": 2.271259016889776e-06,
"loss": 0.0023,
"step": 4624
},
{
"epoch": 8.79,
"grad_norm": 0.011444780975580215,
"learning_rate": 2.2642134676468796e-06,
"loss": 0.0001,
"step": 4625
},
{
"epoch": 8.79,
"grad_norm": 0.033059582114219666,
"learning_rate": 2.2571784346154047e-06,
"loss": 0.0003,
"step": 4626
},
{
"epoch": 8.8,
"grad_norm": 0.11914518475532532,
"learning_rate": 2.250153920462754e-06,
"loss": 0.0004,
"step": 4627
},
{
"epoch": 8.8,
"grad_norm": 0.7876043319702148,
"learning_rate": 2.243139927852308e-06,
"loss": 0.0045,
"step": 4628
},
{
"epoch": 8.8,
"grad_norm": 0.11784661561250687,
"learning_rate": 2.2361364594434686e-06,
"loss": 0.0003,
"step": 4629
},
{
"epoch": 8.8,
"grad_norm": 0.008431946858763695,
"learning_rate": 2.2291435178916552e-06,
"loss": 0.0001,
"step": 4630
},
{
"epoch": 8.8,
"grad_norm": 0.057966090738773346,
"learning_rate": 2.222161105848284e-06,
"loss": 0.0003,
"step": 4631
},
{
"epoch": 8.81,
"grad_norm": 0.15400613844394684,
"learning_rate": 2.2151892259607833e-06,
"loss": 0.0013,
"step": 4632
},
{
"epoch": 8.81,
"grad_norm": 0.040687792003154755,
"learning_rate": 2.208227880872585e-06,
"loss": 0.0003,
"step": 4633
},
{
"epoch": 8.81,
"grad_norm": 0.11467866599559784,
"learning_rate": 2.2012770732231493e-06,
"loss": 0.0005,
"step": 4634
},
{
"epoch": 8.81,
"grad_norm": 0.04431634768843651,
"learning_rate": 2.1943368056478962e-06,
"loss": 0.0004,
"step": 4635
},
{
"epoch": 8.81,
"grad_norm": 0.019636640325188637,
"learning_rate": 2.1874070807782952e-06,
"loss": 0.0002,
"step": 4636
},
{
"epoch": 8.82,
"grad_norm": 0.03395551070570946,
"learning_rate": 2.180487901241781e-06,
"loss": 0.0002,
"step": 4637
},
{
"epoch": 8.82,
"grad_norm": 0.10025951266288757,
"learning_rate": 2.173579269661822e-06,
"loss": 0.0005,
"step": 4638
},
{
"epoch": 8.82,
"grad_norm": 0.5412251353263855,
"learning_rate": 2.1666811886578688e-06,
"loss": 0.0006,
"step": 4639
},
{
"epoch": 8.82,
"grad_norm": 0.09923485666513443,
"learning_rate": 2.1597936608453727e-06,
"loss": 0.0002,
"step": 4640
},
{
"epoch": 8.82,
"grad_norm": 0.11386619508266449,
"learning_rate": 2.1529166888357853e-06,
"loss": 0.0003,
"step": 4641
},
{
"epoch": 8.83,
"grad_norm": 0.02950025536119938,
"learning_rate": 2.1460502752365672e-06,
"loss": 0.0001,
"step": 4642
},
{
"epoch": 8.83,
"grad_norm": 0.05076417326927185,
"learning_rate": 2.139194422651153e-06,
"loss": 0.0004,
"step": 4643
},
{
"epoch": 8.83,
"grad_norm": 0.5874290466308594,
"learning_rate": 2.132349133678989e-06,
"loss": 0.0037,
"step": 4644
},
{
"epoch": 8.83,
"grad_norm": 0.10737746208906174,
"learning_rate": 2.1255144109155234e-06,
"loss": 0.0004,
"step": 4645
},
{
"epoch": 8.83,
"grad_norm": 0.02840178832411766,
"learning_rate": 2.118690256952169e-06,
"loss": 0.0002,
"step": 4646
},
{
"epoch": 8.83,
"grad_norm": 0.05873400717973709,
"learning_rate": 2.111876674376367e-06,
"loss": 0.0003,
"step": 4647
},
{
"epoch": 8.84,
"grad_norm": 0.09925812482833862,
"learning_rate": 2.1050736657715242e-06,
"loss": 0.0006,
"step": 4648
},
{
"epoch": 8.84,
"grad_norm": 0.05025234445929527,
"learning_rate": 2.0982812337170497e-06,
"loss": 0.0001,
"step": 4649
},
{
"epoch": 8.84,
"grad_norm": 0.09930465370416641,
"learning_rate": 2.0914993807883375e-06,
"loss": 0.0005,
"step": 4650
},
{
"epoch": 8.84,
"grad_norm": 0.1430976390838623,
"learning_rate": 2.084728109556784e-06,
"loss": 0.0019,
"step": 4651
},
{
"epoch": 8.84,
"grad_norm": 0.0570414774119854,
"learning_rate": 2.077967422589745e-06,
"loss": 0.0003,
"step": 4652
},
{
"epoch": 8.85,
"grad_norm": 0.02111702226102352,
"learning_rate": 2.071217322450597e-06,
"loss": 0.0001,
"step": 4653
},
{
"epoch": 8.85,
"grad_norm": 0.1604692041873932,
"learning_rate": 2.0644778116986798e-06,
"loss": 0.0006,
"step": 4654
},
{
"epoch": 8.85,
"grad_norm": 0.2252551168203354,
"learning_rate": 2.0577488928893216e-06,
"loss": 0.002,
"step": 4655
},
{
"epoch": 8.85,
"grad_norm": 0.51157146692276,
"learning_rate": 2.0510305685738397e-06,
"loss": 0.001,
"step": 4656
},
{
"epoch": 8.85,
"grad_norm": 0.17236408591270447,
"learning_rate": 2.0443228412995352e-06,
"loss": 0.0012,
"step": 4657
},
{
"epoch": 8.86,
"grad_norm": 0.024810392409563065,
"learning_rate": 2.0376257136096843e-06,
"loss": 0.0001,
"step": 4658
},
{
"epoch": 8.86,
"grad_norm": 0.07398432493209839,
"learning_rate": 2.0309391880435457e-06,
"loss": 0.0015,
"step": 4659
},
{
"epoch": 8.86,
"grad_norm": 0.20366047322750092,
"learning_rate": 2.024263267136375e-06,
"loss": 0.0012,
"step": 4660
},
{
"epoch": 8.86,
"grad_norm": 0.019363315775990486,
"learning_rate": 2.0175979534193767e-06,
"loss": 0.0001,
"step": 4661
},
{
"epoch": 8.86,
"grad_norm": 0.00914311595261097,
"learning_rate": 2.01094324941976e-06,
"loss": 0.0001,
"step": 4662
},
{
"epoch": 8.87,
"grad_norm": 0.07300450652837753,
"learning_rate": 2.0042991576606893e-06,
"loss": 0.0002,
"step": 4663
},
{
"epoch": 8.87,
"grad_norm": 0.017182854935526848,
"learning_rate": 1.9976656806613336e-06,
"loss": 0.0001,
"step": 4664
},
{
"epoch": 8.87,
"grad_norm": 0.3499352037906647,
"learning_rate": 1.991042820936807e-06,
"loss": 0.0013,
"step": 4665
},
{
"epoch": 8.87,
"grad_norm": 0.10591781139373779,
"learning_rate": 1.9844305809982177e-06,
"loss": 0.0001,
"step": 4666
},
{
"epoch": 8.87,
"grad_norm": 0.30358022451400757,
"learning_rate": 1.9778289633526402e-06,
"loss": 0.0007,
"step": 4667
},
{
"epoch": 8.87,
"grad_norm": 0.07155616581439972,
"learning_rate": 1.9712379705031225e-06,
"loss": 0.0003,
"step": 4668
},
{
"epoch": 8.88,
"grad_norm": 0.010810419917106628,
"learning_rate": 1.9646576049486853e-06,
"loss": 0.0001,
"step": 4669
},
{
"epoch": 8.88,
"grad_norm": 0.09337535500526428,
"learning_rate": 1.958087869184313e-06,
"loss": 0.0021,
"step": 4670
},
{
"epoch": 8.88,
"grad_norm": 0.13956142961978912,
"learning_rate": 1.951528765700976e-06,
"loss": 0.0021,
"step": 4671
},
{
"epoch": 8.88,
"grad_norm": 0.263908714056015,
"learning_rate": 1.9449802969855924e-06,
"loss": 0.0006,
"step": 4672
},
{
"epoch": 8.88,
"grad_norm": 0.17581988871097565,
"learning_rate": 1.938442465521065e-06,
"loss": 0.0008,
"step": 4673
},
{
"epoch": 8.89,
"grad_norm": 0.02684350311756134,
"learning_rate": 1.9319152737862557e-06,
"loss": 0.0001,
"step": 4674
},
{
"epoch": 8.89,
"grad_norm": 0.03221539780497551,
"learning_rate": 1.92539872425599e-06,
"loss": 0.0002,
"step": 4675
},
{
"epoch": 8.89,
"grad_norm": 0.5744677782058716,
"learning_rate": 1.918892819401069e-06,
"loss": 0.0012,
"step": 4676
},
{
"epoch": 8.89,
"grad_norm": 0.08129557967185974,
"learning_rate": 1.9123975616882427e-06,
"loss": 0.0006,
"step": 4677
},
{
"epoch": 8.89,
"grad_norm": 0.026351241394877434,
"learning_rate": 1.9059129535802378e-06,
"loss": 0.0001,
"step": 4678
},
{
"epoch": 8.9,
"grad_norm": 0.05412183701992035,
"learning_rate": 1.8994389975357275e-06,
"loss": 0.0004,
"step": 4679
},
{
"epoch": 8.9,
"grad_norm": 0.10783185064792633,
"learning_rate": 1.8929756960093748e-06,
"loss": 0.0003,
"step": 4680
},
{
"epoch": 8.9,
"grad_norm": 0.6738510131835938,
"learning_rate": 1.8865230514517662e-06,
"loss": 0.003,
"step": 4681
},
{
"epoch": 8.9,
"grad_norm": 0.045805200934410095,
"learning_rate": 1.8800810663094746e-06,
"loss": 0.0003,
"step": 4682
},
{
"epoch": 8.9,
"grad_norm": 0.3390243351459503,
"learning_rate": 1.87364974302502e-06,
"loss": 0.0008,
"step": 4683
},
{
"epoch": 8.9,
"grad_norm": 0.03300372138619423,
"learning_rate": 1.8672290840368844e-06,
"loss": 0.0003,
"step": 4684
},
{
"epoch": 8.91,
"grad_norm": 0.053998783230781555,
"learning_rate": 1.8608190917795009e-06,
"loss": 0.0004,
"step": 4685
},
{
"epoch": 8.91,
"grad_norm": 0.034828126430511475,
"learning_rate": 1.8544197686832654e-06,
"loss": 0.0002,
"step": 4686
},
{
"epoch": 8.91,
"grad_norm": 0.019023120403289795,
"learning_rate": 1.8480311171745202e-06,
"loss": 0.0002,
"step": 4687
},
{
"epoch": 8.91,
"grad_norm": 0.046864528208971024,
"learning_rate": 1.8416531396755677e-06,
"loss": 0.0001,
"step": 4688
},
{
"epoch": 8.91,
"grad_norm": 0.007893942296504974,
"learning_rate": 1.83528583860466e-06,
"loss": 0.0001,
"step": 4689
},
{
"epoch": 8.92,
"grad_norm": 0.027218453586101532,
"learning_rate": 1.8289292163760029e-06,
"loss": 0.0001,
"step": 4690
},
{
"epoch": 8.92,
"grad_norm": 0.012962251901626587,
"learning_rate": 1.8225832753997584e-06,
"loss": 0.0001,
"step": 4691
},
{
"epoch": 8.92,
"grad_norm": 0.031790610402822495,
"learning_rate": 1.8162480180820218e-06,
"loss": 0.0002,
"step": 4692
},
{
"epoch": 8.92,
"grad_norm": 0.010785268619656563,
"learning_rate": 1.8099234468248616e-06,
"loss": 0.0001,
"step": 4693
},
{
"epoch": 8.92,
"grad_norm": 0.08944685757160187,
"learning_rate": 1.8036095640262728e-06,
"loss": 0.0008,
"step": 4694
},
{
"epoch": 8.93,
"grad_norm": 0.03840061277151108,
"learning_rate": 1.7973063720802108e-06,
"loss": 0.0002,
"step": 4695
},
{
"epoch": 8.93,
"grad_norm": 0.024621956050395966,
"learning_rate": 1.7910138733765668e-06,
"loss": 0.0002,
"step": 4696
},
{
"epoch": 8.93,
"grad_norm": 0.08167129009962082,
"learning_rate": 1.7847320703012027e-06,
"loss": 0.0002,
"step": 4697
},
{
"epoch": 8.93,
"grad_norm": 0.009565987624228,
"learning_rate": 1.778460965235883e-06,
"loss": 0.0001,
"step": 4698
},
{
"epoch": 8.93,
"grad_norm": 0.18731284141540527,
"learning_rate": 1.7722005605583525e-06,
"loss": 0.0005,
"step": 4699
},
{
"epoch": 8.94,
"grad_norm": 0.09767553955316544,
"learning_rate": 1.7659508586422858e-06,
"loss": 0.0018,
"step": 4700
},
{
"epoch": 8.94,
"eval_blimp_filtered_avg": 0.7176119402985075,
"eval_blimp_filtered_std": 0.004997237153821365,
"step": 4700
},
{
"epoch": 8.94,
"eval_blimp_supplement_avg": 0.7823275862068966,
"eval_blimp_supplement_std": 0.01801770012633094,
"step": 4700
},
{
"epoch": 8.94,
"eval_vqa_filtered_avg": 0.32,
"eval_vqa_filtered_std": 0.046882617226215034,
"step": 4700
},
{
"epoch": 8.94,
"eval_winoground_filtered_avg": 0.51,
"eval_winoground_filtered_std": 0.05024183937956912,
"step": 4700
},
{
"epoch": 8.94,
"grad_norm": 0.017770908772945404,
"learning_rate": 1.7597118618573005e-06,
"loss": 0.0001,
"step": 4701
},
{
"epoch": 8.94,
"grad_norm": 0.6925677061080933,
"learning_rate": 1.7534835725689513e-06,
"loss": 0.004,
"step": 4702
},
{
"epoch": 8.94,
"grad_norm": 0.073093481361866,
"learning_rate": 1.7472659931387357e-06,
"loss": 0.0013,
"step": 4703
},
{
"epoch": 8.94,
"grad_norm": 0.008850627578794956,
"learning_rate": 1.7410591259240916e-06,
"loss": 0.0001,
"step": 4704
},
{
"epoch": 8.94,
"grad_norm": 0.02766992338001728,
"learning_rate": 1.7348629732783928e-06,
"loss": 0.0002,
"step": 4705
},
{
"epoch": 8.95,
"grad_norm": 0.10123324394226074,
"learning_rate": 1.7286775375509668e-06,
"loss": 0.0002,
"step": 4706
},
{
"epoch": 8.95,
"grad_norm": 0.009397914633154869,
"learning_rate": 1.7225028210870409e-06,
"loss": 0.0001,
"step": 4707
},
{
"epoch": 8.95,
"grad_norm": 0.03876335173845291,
"learning_rate": 1.7163388262278157e-06,
"loss": 0.0003,
"step": 4708
},
{
"epoch": 8.95,
"grad_norm": 0.14825305342674255,
"learning_rate": 1.7101855553104085e-06,
"loss": 0.0006,
"step": 4709
},
{
"epoch": 8.95,
"grad_norm": 0.021532295271754265,
"learning_rate": 1.7040430106678728e-06,
"loss": 0.0001,
"step": 4710
},
{
"epoch": 8.96,
"grad_norm": 0.09348251670598984,
"learning_rate": 1.6979111946291958e-06,
"loss": 0.0004,
"step": 4711
},
{
"epoch": 8.96,
"grad_norm": 0.02957487665116787,
"learning_rate": 1.6917901095193011e-06,
"loss": 0.0001,
"step": 4712
},
{
"epoch": 8.96,
"grad_norm": 0.05971124395728111,
"learning_rate": 1.6856797576590322e-06,
"loss": 0.0007,
"step": 4713
},
{
"epoch": 8.96,
"grad_norm": 0.3444651663303375,
"learning_rate": 1.6795801413651756e-06,
"loss": 0.0019,
"step": 4714
},
{
"epoch": 8.96,
"grad_norm": 0.21642756462097168,
"learning_rate": 1.6734912629504417e-06,
"loss": 0.0006,
"step": 4715
},
{
"epoch": 8.97,
"grad_norm": 0.029698865488171577,
"learning_rate": 1.6674131247234636e-06,
"loss": 0.0002,
"step": 4716
},
{
"epoch": 8.97,
"grad_norm": 0.16166429221630096,
"learning_rate": 1.6613457289888245e-06,
"loss": 0.0024,
"step": 4717
},
{
"epoch": 8.97,
"grad_norm": 0.07993683964014053,
"learning_rate": 1.6552890780469977e-06,
"loss": 0.0003,
"step": 4718
},
{
"epoch": 8.97,
"grad_norm": 0.1011294275522232,
"learning_rate": 1.6492431741944192e-06,
"loss": 0.0004,
"step": 4719
},
{
"epoch": 8.97,
"grad_norm": 0.01885506696999073,
"learning_rate": 1.643208019723429e-06,
"loss": 0.0001,
"step": 4720
},
{
"epoch": 8.98,
"grad_norm": 0.08363159745931625,
"learning_rate": 1.6371836169222997e-06,
"loss": 0.0005,
"step": 4721
},
{
"epoch": 8.98,
"grad_norm": 0.10607343167066574,
"learning_rate": 1.6311699680752179e-06,
"loss": 0.0004,
"step": 4722
},
{
"epoch": 8.98,
"grad_norm": 0.009857023134827614,
"learning_rate": 1.625167075462306e-06,
"loss": 0.0001,
"step": 4723
},
{
"epoch": 8.98,
"grad_norm": 0.045258209109306335,
"learning_rate": 1.619174941359597e-06,
"loss": 0.0002,
"step": 4724
},
{
"epoch": 8.98,
"grad_norm": 0.18081235885620117,
"learning_rate": 1.6131935680390497e-06,
"loss": 0.0038,
"step": 4725
},
{
"epoch": 8.98,
"grad_norm": 0.07224899530410767,
"learning_rate": 1.607222957768547e-06,
"loss": 0.0002,
"step": 4726
},
{
"epoch": 8.99,
"grad_norm": 0.035417377948760986,
"learning_rate": 1.6012631128118782e-06,
"loss": 0.0002,
"step": 4727
},
{
"epoch": 8.99,
"grad_norm": 0.02108675427734852,
"learning_rate": 1.5953140354287655e-06,
"loss": 0.0001,
"step": 4728
},
{
"epoch": 8.99,
"grad_norm": 0.017698531970381737,
"learning_rate": 1.5893757278748412e-06,
"loss": 0.0001,
"step": 4729
},
{
"epoch": 8.99,
"grad_norm": 0.5665483474731445,
"learning_rate": 1.5834481924016542e-06,
"loss": 0.0046,
"step": 4730
},
{
"epoch": 8.99,
"grad_norm": 0.4219640791416168,
"learning_rate": 1.5775314312566636e-06,
"loss": 0.0018,
"step": 4731
},
{
"epoch": 9.0,
"grad_norm": 0.4670141637325287,
"learning_rate": 1.5716254466832613e-06,
"loss": 0.001,
"step": 4732
},
{
"epoch": 9.0,
"grad_norm": 0.2617659270763397,
"learning_rate": 1.565730240920733e-06,
"loss": 0.0046,
"step": 4733
},
{
"epoch": 9.0,
"grad_norm": 0.03550489619374275,
"learning_rate": 1.5598458162042872e-06,
"loss": 0.0001,
"step": 4734
},
{
"epoch": 9.0,
"grad_norm": 0.025873105973005295,
"learning_rate": 1.553972174765046e-06,
"loss": 0.0002,
"step": 4735
},
{
"epoch": 9.0,
"grad_norm": 0.11365854740142822,
"learning_rate": 1.5481093188300344e-06,
"loss": 0.0004,
"step": 4736
},
{
"epoch": 9.01,
"grad_norm": 0.3686664402484894,
"learning_rate": 1.5422572506222043e-06,
"loss": 0.0011,
"step": 4737
},
{
"epoch": 9.01,
"grad_norm": 0.0580432265996933,
"learning_rate": 1.5364159723603942e-06,
"loss": 0.0005,
"step": 4738
},
{
"epoch": 9.01,
"grad_norm": 0.04454429820179939,
"learning_rate": 1.5305854862593726e-06,
"loss": 0.0002,
"step": 4739
},
{
"epoch": 9.01,
"grad_norm": 0.10450960695743561,
"learning_rate": 1.5247657945298077e-06,
"loss": 0.0004,
"step": 4740
},
{
"epoch": 9.01,
"grad_norm": 0.28574690222740173,
"learning_rate": 1.5189568993782743e-06,
"loss": 0.0007,
"step": 4741
},
{
"epoch": 9.02,
"grad_norm": 0.09119351208209991,
"learning_rate": 1.5131588030072474e-06,
"loss": 0.0001,
"step": 4742
},
{
"epoch": 9.02,
"grad_norm": 0.0542692132294178,
"learning_rate": 1.5073715076151317e-06,
"loss": 0.001,
"step": 4743
},
{
"epoch": 9.02,
"grad_norm": 0.020157869905233383,
"learning_rate": 1.5015950153961988e-06,
"loss": 0.0001,
"step": 4744
},
{
"epoch": 9.02,
"grad_norm": 0.06817253679037094,
"learning_rate": 1.4958293285406565e-06,
"loss": 0.0002,
"step": 4745
},
{
"epoch": 9.02,
"grad_norm": 0.06858240067958832,
"learning_rate": 1.490074449234603e-06,
"loss": 0.0002,
"step": 4746
},
{
"epoch": 9.02,
"grad_norm": 0.07347028702497482,
"learning_rate": 1.4843303796600427e-06,
"loss": 0.0003,
"step": 4747
},
{
"epoch": 9.03,
"grad_norm": 0.012713614851236343,
"learning_rate": 1.4785971219948702e-06,
"loss": 0.0001,
"step": 4748
},
{
"epoch": 9.03,
"grad_norm": 0.1248493418097496,
"learning_rate": 1.4728746784128965e-06,
"loss": 0.0015,
"step": 4749
},
{
"epoch": 9.03,
"grad_norm": 0.35022076964378357,
"learning_rate": 1.4671630510838196e-06,
"loss": 0.001,
"step": 4750
},
{
"epoch": 9.03,
"grad_norm": 0.19104939699172974,
"learning_rate": 1.4614622421732404e-06,
"loss": 0.0005,
"step": 4751
},
{
"epoch": 9.03,
"grad_norm": 0.049895383417606354,
"learning_rate": 1.4557722538426698e-06,
"loss": 0.0003,
"step": 4752
},
{
"epoch": 9.04,
"grad_norm": 0.6240670084953308,
"learning_rate": 1.4500930882494923e-06,
"loss": 0.0038,
"step": 4753
},
{
"epoch": 9.04,
"grad_norm": 0.0846959799528122,
"learning_rate": 1.4444247475470117e-06,
"loss": 0.0005,
"step": 4754
},
{
"epoch": 9.04,
"grad_norm": 0.007603269070386887,
"learning_rate": 1.438767233884416e-06,
"loss": 0.0001,
"step": 4755
},
{
"epoch": 9.04,
"grad_norm": 0.04665505886077881,
"learning_rate": 1.4331205494067857e-06,
"loss": 0.0004,
"step": 4756
},
{
"epoch": 9.04,
"grad_norm": 0.023231789469718933,
"learning_rate": 1.4274846962551014e-06,
"loss": 0.0002,
"step": 4757
},
{
"epoch": 9.05,
"grad_norm": 0.062225788831710815,
"learning_rate": 1.4218596765662372e-06,
"loss": 0.0008,
"step": 4758
},
{
"epoch": 9.05,
"grad_norm": 0.01815798506140709,
"learning_rate": 1.4162454924729573e-06,
"loss": 0.0001,
"step": 4759
},
{
"epoch": 9.05,
"grad_norm": 0.016672363504767418,
"learning_rate": 1.410642146103912e-06,
"loss": 0.0002,
"step": 4760
},
{
"epoch": 9.05,
"grad_norm": 0.06553399562835693,
"learning_rate": 1.4050496395836553e-06,
"loss": 0.0002,
"step": 4761
},
{
"epoch": 9.05,
"grad_norm": 0.0243124570697546,
"learning_rate": 1.3994679750326145e-06,
"loss": 0.0002,
"step": 4762
},
{
"epoch": 9.06,
"grad_norm": 0.020681560039520264,
"learning_rate": 1.3938971545671298e-06,
"loss": 0.0001,
"step": 4763
},
{
"epoch": 9.06,
"grad_norm": 0.1342475414276123,
"learning_rate": 1.3883371802994017e-06,
"loss": 0.0006,
"step": 4764
},
{
"epoch": 9.06,
"grad_norm": 0.09999359399080276,
"learning_rate": 1.3827880543375405e-06,
"loss": 0.0002,
"step": 4765
},
{
"epoch": 9.06,
"grad_norm": 0.06031373515725136,
"learning_rate": 1.3772497787855331e-06,
"loss": 0.0003,
"step": 4766
},
{
"epoch": 9.06,
"grad_norm": 0.8776128888130188,
"learning_rate": 1.3717223557432534e-06,
"loss": 0.003,
"step": 4767
},
{
"epoch": 9.06,
"grad_norm": 0.08178051561117172,
"learning_rate": 1.3662057873064615e-06,
"loss": 0.0011,
"step": 4768
},
{
"epoch": 9.07,
"grad_norm": 0.46873682737350464,
"learning_rate": 1.3607000755668009e-06,
"loss": 0.001,
"step": 4769
},
{
"epoch": 9.07,
"grad_norm": 0.2710656523704529,
"learning_rate": 1.3552052226118018e-06,
"loss": 0.002,
"step": 4770
},
{
"epoch": 9.07,
"grad_norm": 0.025585375726222992,
"learning_rate": 1.3497212305248675e-06,
"loss": 0.0003,
"step": 4771
},
{
"epoch": 9.07,
"grad_norm": 0.11600522696971893,
"learning_rate": 1.3442481013853048e-06,
"loss": 0.0006,
"step": 4772
},
{
"epoch": 9.07,
"grad_norm": 0.03956759721040726,
"learning_rate": 1.3387858372682737e-06,
"loss": 0.0003,
"step": 4773
},
{
"epoch": 9.08,
"grad_norm": 0.048587948083877563,
"learning_rate": 1.3333344402448378e-06,
"loss": 0.0003,
"step": 4774
},
{
"epoch": 9.08,
"grad_norm": 0.03531361743807793,
"learning_rate": 1.3278939123819267e-06,
"loss": 0.0001,
"step": 4775
},
{
"epoch": 9.08,
"grad_norm": 0.2309519499540329,
"learning_rate": 1.322464255742354e-06,
"loss": 0.001,
"step": 4776
},
{
"epoch": 9.08,
"grad_norm": 0.04637506604194641,
"learning_rate": 1.317045472384809e-06,
"loss": 0.0005,
"step": 4777
},
{
"epoch": 9.08,
"grad_norm": 0.018952086567878723,
"learning_rate": 1.311637564363869e-06,
"loss": 0.0001,
"step": 4778
},
{
"epoch": 9.09,
"grad_norm": 0.004110359586775303,
"learning_rate": 1.3062405337299699e-06,
"loss": 0.0001,
"step": 4779
},
{
"epoch": 9.09,
"grad_norm": 0.30696922540664673,
"learning_rate": 1.3008543825294316e-06,
"loss": 0.0008,
"step": 4780
},
{
"epoch": 9.09,
"grad_norm": 0.03733064606785774,
"learning_rate": 1.2954791128044608e-06,
"loss": 0.0002,
"step": 4781
},
{
"epoch": 9.09,
"grad_norm": 0.09673619270324707,
"learning_rate": 1.29011472659311e-06,
"loss": 0.0008,
"step": 4782
},
{
"epoch": 9.09,
"grad_norm": 0.4160933494567871,
"learning_rate": 1.2847612259293428e-06,
"loss": 0.001,
"step": 4783
},
{
"epoch": 9.1,
"grad_norm": 0.19258812069892883,
"learning_rate": 1.2794186128429587e-06,
"loss": 0.0005,
"step": 4784
},
{
"epoch": 9.1,
"grad_norm": 0.13979655504226685,
"learning_rate": 1.2740868893596536e-06,
"loss": 0.0003,
"step": 4785
},
{
"epoch": 9.1,
"grad_norm": 0.02205735631287098,
"learning_rate": 1.268766057500984e-06,
"loss": 0.0001,
"step": 4786
},
{
"epoch": 9.1,
"grad_norm": 0.057746522128582,
"learning_rate": 1.263456119284383e-06,
"loss": 0.0002,
"step": 4787
},
{
"epoch": 9.1,
"grad_norm": 0.12576134502887726,
"learning_rate": 1.2581570767231398e-06,
"loss": 0.0005,
"step": 4788
},
{
"epoch": 9.1,
"grad_norm": 0.02352857030928135,
"learning_rate": 1.252868931826441e-06,
"loss": 0.0001,
"step": 4789
},
{
"epoch": 9.11,
"grad_norm": 0.015999021008610725,
"learning_rate": 1.247591686599302e-06,
"loss": 0.0001,
"step": 4790
},
{
"epoch": 9.11,
"grad_norm": 0.11603011190891266,
"learning_rate": 1.2423253430426395e-06,
"loss": 0.0005,
"step": 4791
},
{
"epoch": 9.11,
"grad_norm": 0.030100291594862938,
"learning_rate": 1.2370699031532162e-06,
"loss": 0.0002,
"step": 4792
},
{
"epoch": 9.11,
"grad_norm": 0.13961903750896454,
"learning_rate": 1.2318253689236747e-06,
"loss": 0.0011,
"step": 4793
},
{
"epoch": 9.11,
"grad_norm": 0.04333731532096863,
"learning_rate": 1.2265917423425077e-06,
"loss": 0.0003,
"step": 4794
},
{
"epoch": 9.12,
"grad_norm": 0.901295006275177,
"learning_rate": 1.2213690253940847e-06,
"loss": 0.0018,
"step": 4795
},
{
"epoch": 9.12,
"grad_norm": 0.1933199167251587,
"learning_rate": 1.2161572200586345e-06,
"loss": 0.0005,
"step": 4796
},
{
"epoch": 9.12,
"grad_norm": 0.01970449462532997,
"learning_rate": 1.2109563283122438e-06,
"loss": 0.0003,
"step": 4797
},
{
"epoch": 9.12,
"grad_norm": 0.16480223834514618,
"learning_rate": 1.205766352126878e-06,
"loss": 0.0013,
"step": 4798
},
{
"epoch": 9.12,
"grad_norm": 0.11919678747653961,
"learning_rate": 1.2005872934703366e-06,
"loss": 0.0007,
"step": 4799
},
{
"epoch": 9.13,
"grad_norm": 0.019419632852077484,
"learning_rate": 1.195419154306302e-06,
"loss": 0.0001,
"step": 4800
},
{
"epoch": 9.13,
"eval_blimp_filtered_avg": 0.716865671641791,
"eval_blimp_filtered_std": 0.00499095175766691,
"step": 4800
},
{
"epoch": 9.13,
"eval_blimp_supplement_avg": 0.7801724137931034,
"eval_blimp_supplement_std": 0.018094413033767815,
"step": 4800
},
{
"epoch": 9.13,
"eval_vqa_filtered_avg": 0.32,
"eval_vqa_filtered_std": 0.046882617226215034,
"step": 4800
},
{
"epoch": 9.13,
"eval_winoground_filtered_avg": 0.5,
"eval_winoground_filtered_std": 0.050251890762960605,
"step": 4800
},
{
"epoch": 9.13,
"grad_norm": 0.03386770561337471,
"learning_rate": 1.1902619365943101e-06,
"loss": 0.0003,
"step": 4801
},
{
"epoch": 9.13,
"grad_norm": 0.04488684609532356,
"learning_rate": 1.185115642289757e-06,
"loss": 0.0005,
"step": 4802
},
{
"epoch": 9.13,
"grad_norm": 0.26234015822410583,
"learning_rate": 1.179980273343888e-06,
"loss": 0.0015,
"step": 4803
},
{
"epoch": 9.13,
"grad_norm": 0.08464223891496658,
"learning_rate": 1.1748558317038161e-06,
"loss": 0.0002,
"step": 4804
},
{
"epoch": 9.13,
"grad_norm": 0.023310324177145958,
"learning_rate": 1.1697423193125101e-06,
"loss": 0.0002,
"step": 4805
},
{
"epoch": 9.14,
"grad_norm": 0.022350599989295006,
"learning_rate": 1.1646397381087893e-06,
"loss": 0.0001,
"step": 4806
},
{
"epoch": 9.14,
"grad_norm": 0.006106809247285128,
"learning_rate": 1.1595480900273393e-06,
"loss": 0.0001,
"step": 4807
},
{
"epoch": 9.14,
"grad_norm": 0.12292753159999847,
"learning_rate": 1.1544673769986825e-06,
"loss": 0.0002,
"step": 4808
},
{
"epoch": 9.14,
"grad_norm": 0.07004491984844208,
"learning_rate": 1.1493976009492179e-06,
"loss": 0.0009,
"step": 4809
},
{
"epoch": 9.14,
"grad_norm": 0.015628525987267494,
"learning_rate": 1.1443387638011681e-06,
"loss": 0.0001,
"step": 4810
},
{
"epoch": 9.15,
"grad_norm": 0.384322851896286,
"learning_rate": 1.1392908674726421e-06,
"loss": 0.0004,
"step": 4811
},
{
"epoch": 9.15,
"grad_norm": 0.44587504863739014,
"learning_rate": 1.1342539138775787e-06,
"loss": 0.0006,
"step": 4812
},
{
"epoch": 9.15,
"grad_norm": 0.012115846388041973,
"learning_rate": 1.1292279049257703e-06,
"loss": 0.0001,
"step": 4813
},
{
"epoch": 9.15,
"grad_norm": 0.01616806723177433,
"learning_rate": 1.124212842522866e-06,
"loss": 0.0001,
"step": 4814
},
{
"epoch": 9.15,
"grad_norm": 0.04483051225543022,
"learning_rate": 1.1192087285703578e-06,
"loss": 0.0002,
"step": 4815
},
{
"epoch": 9.16,
"grad_norm": 0.013437979854643345,
"learning_rate": 1.1142155649655883e-06,
"loss": 0.0002,
"step": 4816
},
{
"epoch": 9.16,
"grad_norm": 0.022331811487674713,
"learning_rate": 1.1092333536017496e-06,
"loss": 0.0001,
"step": 4817
},
{
"epoch": 9.16,
"grad_norm": 0.09167629480361938,
"learning_rate": 1.1042620963678907e-06,
"loss": 0.0002,
"step": 4818
},
{
"epoch": 9.16,
"grad_norm": 0.0342106968164444,
"learning_rate": 1.0993017951488805e-06,
"loss": 0.0001,
"step": 4819
},
{
"epoch": 9.16,
"grad_norm": 0.10716624557971954,
"learning_rate": 1.0943524518254644e-06,
"loss": 0.0003,
"step": 4820
},
{
"epoch": 9.17,
"grad_norm": 0.03595684468746185,
"learning_rate": 1.089414068274215e-06,
"loss": 0.0003,
"step": 4821
},
{
"epoch": 9.17,
"grad_norm": 0.05555866286158562,
"learning_rate": 1.084486646367554e-06,
"loss": 0.0001,
"step": 4822
},
{
"epoch": 9.17,
"grad_norm": 0.06728649139404297,
"learning_rate": 1.0795701879737442e-06,
"loss": 0.0005,
"step": 4823
},
{
"epoch": 9.17,
"grad_norm": 0.2041804939508438,
"learning_rate": 1.0746646949569006e-06,
"loss": 0.0009,
"step": 4824
},
{
"epoch": 9.17,
"grad_norm": 0.2819393575191498,
"learning_rate": 1.0697701691769691e-06,
"loss": 0.001,
"step": 4825
},
{
"epoch": 9.17,
"grad_norm": 0.10957486927509308,
"learning_rate": 1.0648866124897382e-06,
"loss": 0.0004,
"step": 4826
},
{
"epoch": 9.18,
"grad_norm": 0.15928010642528534,
"learning_rate": 1.0600140267468538e-06,
"loss": 0.0005,
"step": 4827
},
{
"epoch": 9.18,
"grad_norm": 0.10874047875404358,
"learning_rate": 1.0551524137957747e-06,
"loss": 0.0003,
"step": 4828
},
{
"epoch": 9.18,
"grad_norm": 0.3573073446750641,
"learning_rate": 1.050301775479826e-06,
"loss": 0.0008,
"step": 4829
},
{
"epoch": 9.18,
"grad_norm": 0.051142338663339615,
"learning_rate": 1.045462113638157e-06,
"loss": 0.0003,
"step": 4830
},
{
"epoch": 9.18,
"grad_norm": 0.024142645299434662,
"learning_rate": 1.0406334301057597e-06,
"loss": 0.0001,
"step": 4831
},
{
"epoch": 9.19,
"grad_norm": 0.16203027963638306,
"learning_rate": 1.035815726713456e-06,
"loss": 0.0002,
"step": 4832
},
{
"epoch": 9.19,
"grad_norm": 0.062344275414943695,
"learning_rate": 1.0310090052879185e-06,
"loss": 0.0004,
"step": 4833
},
{
"epoch": 9.19,
"grad_norm": 0.05828079953789711,
"learning_rate": 1.026213267651639e-06,
"loss": 0.0002,
"step": 4834
},
{
"epoch": 9.19,
"grad_norm": 0.05517910048365593,
"learning_rate": 1.0214285156229663e-06,
"loss": 0.0002,
"step": 4835
},
{
"epoch": 9.19,
"grad_norm": 0.08056183159351349,
"learning_rate": 1.0166547510160595e-06,
"loss": 0.0003,
"step": 4836
},
{
"epoch": 9.2,
"grad_norm": 0.02815484255552292,
"learning_rate": 1.0118919756409306e-06,
"loss": 0.0001,
"step": 4837
},
{
"epoch": 9.2,
"grad_norm": 0.04062763229012489,
"learning_rate": 1.0071401913034185e-06,
"loss": 0.0002,
"step": 4838
},
{
"epoch": 9.2,
"grad_norm": 0.03878653049468994,
"learning_rate": 1.0023993998051916e-06,
"loss": 0.0001,
"step": 4839
},
{
"epoch": 9.2,
"grad_norm": 0.016959821805357933,
"learning_rate": 9.97669602943756e-07,
"loss": 0.0002,
"step": 4840
},
{
"epoch": 9.2,
"grad_norm": 0.06112067028880119,
"learning_rate": 9.9295080251244e-07,
"loss": 0.0002,
"step": 4841
},
{
"epoch": 9.21,
"grad_norm": 0.007021947298198938,
"learning_rate": 9.882430003004162e-07,
"loss": 0.0001,
"step": 4842
},
{
"epoch": 9.21,
"grad_norm": 0.046339210122823715,
"learning_rate": 9.835461980926729e-07,
"loss": 0.0003,
"step": 4843
},
{
"epoch": 9.21,
"grad_norm": 0.04219613969326019,
"learning_rate": 9.788603976700461e-07,
"loss": 0.0001,
"step": 4844
},
{
"epoch": 9.21,
"grad_norm": 0.019039161503314972,
"learning_rate": 9.741856008091744e-07,
"loss": 0.0001,
"step": 4845
},
{
"epoch": 9.21,
"grad_norm": 0.03942270576953888,
"learning_rate": 9.695218092825465e-07,
"loss": 0.0002,
"step": 4846
},
{
"epoch": 9.21,
"grad_norm": 0.017830662429332733,
"learning_rate": 9.64869024858468e-07,
"loss": 0.0001,
"step": 4847
},
{
"epoch": 9.22,
"grad_norm": 0.015521096996963024,
"learning_rate": 9.602272493010779e-07,
"loss": 0.0001,
"step": 4848
},
{
"epoch": 9.22,
"grad_norm": 0.059850599616765976,
"learning_rate": 9.555964843703346e-07,
"loss": 0.0006,
"step": 4849
},
{
"epoch": 9.22,
"grad_norm": 0.4023211896419525,
"learning_rate": 9.509767318220209e-07,
"loss": 0.0028,
"step": 4850
},
{
"epoch": 9.22,
"grad_norm": 0.14527498185634613,
"learning_rate": 9.463679934077518e-07,
"loss": 0.0004,
"step": 4851
},
{
"epoch": 9.22,
"grad_norm": 0.02655901573598385,
"learning_rate": 9.417702708749531e-07,
"loss": 0.0002,
"step": 4852
},
{
"epoch": 9.23,
"grad_norm": 0.016608482226729393,
"learning_rate": 9.37183565966897e-07,
"loss": 0.0001,
"step": 4853
},
{
"epoch": 9.23,
"grad_norm": 0.1090182363986969,
"learning_rate": 9.32607880422649e-07,
"loss": 0.0014,
"step": 4854
},
{
"epoch": 9.23,
"grad_norm": 0.02858634665608406,
"learning_rate": 9.280432159771246e-07,
"loss": 0.0003,
"step": 4855
},
{
"epoch": 9.23,
"grad_norm": 0.06525914371013641,
"learning_rate": 9.23489574361036e-07,
"loss": 0.0002,
"step": 4856
},
{
"epoch": 9.23,
"grad_norm": 0.07824494689702988,
"learning_rate": 9.189469573009356e-07,
"loss": 0.0002,
"step": 4857
},
{
"epoch": 9.24,
"grad_norm": 0.32934170961380005,
"learning_rate": 9.14415366519179e-07,
"loss": 0.0016,
"step": 4858
},
{
"epoch": 9.24,
"grad_norm": 0.3883056938648224,
"learning_rate": 9.098948037339582e-07,
"loss": 0.0027,
"step": 4859
},
{
"epoch": 9.24,
"grad_norm": 0.17633987963199615,
"learning_rate": 9.053852706592725e-07,
"loss": 0.0007,
"step": 4860
},
{
"epoch": 9.24,
"grad_norm": 0.05800224468111992,
"learning_rate": 9.008867690049371e-07,
"loss": 0.0005,
"step": 4861
},
{
"epoch": 9.24,
"grad_norm": 0.035300113260746,
"learning_rate": 8.963993004765946e-07,
"loss": 0.0002,
"step": 4862
},
{
"epoch": 9.25,
"grad_norm": 0.013261295855045319,
"learning_rate": 8.919228667756941e-07,
"loss": 0.0001,
"step": 4863
},
{
"epoch": 9.25,
"grad_norm": 0.03929504379630089,
"learning_rate": 8.874574695995175e-07,
"loss": 0.0002,
"step": 4864
},
{
"epoch": 9.25,
"grad_norm": 0.01957102119922638,
"learning_rate": 8.830031106411341e-07,
"loss": 0.0001,
"step": 4865
},
{
"epoch": 9.25,
"grad_norm": 0.07102957367897034,
"learning_rate": 8.785597915894561e-07,
"loss": 0.0002,
"step": 4866
},
{
"epoch": 9.25,
"grad_norm": 0.02123275212943554,
"learning_rate": 8.741275141291926e-07,
"loss": 0.0001,
"step": 4867
},
{
"epoch": 9.25,
"grad_norm": 0.295491099357605,
"learning_rate": 8.697062799408762e-07,
"loss": 0.0025,
"step": 4868
},
{
"epoch": 9.26,
"grad_norm": 0.04763409495353699,
"learning_rate": 8.652960907008389e-07,
"loss": 0.0001,
"step": 4869
},
{
"epoch": 9.26,
"grad_norm": 0.025864984840154648,
"learning_rate": 8.608969480812468e-07,
"loss": 0.0001,
"step": 4870
},
{
"epoch": 9.26,
"grad_norm": 0.0130376061424613,
"learning_rate": 8.565088537500554e-07,
"loss": 0.0001,
"step": 4871
},
{
"epoch": 9.26,
"grad_norm": 0.248138427734375,
"learning_rate": 8.521318093710373e-07,
"loss": 0.0021,
"step": 4872
},
{
"epoch": 9.26,
"grad_norm": 0.32735610008239746,
"learning_rate": 8.477658166037949e-07,
"loss": 0.0009,
"step": 4873
},
{
"epoch": 9.27,
"grad_norm": 0.5425891876220703,
"learning_rate": 8.434108771037042e-07,
"loss": 0.0018,
"step": 4874
},
{
"epoch": 9.27,
"grad_norm": 0.10806083679199219,
"learning_rate": 8.390669925219807e-07,
"loss": 0.001,
"step": 4875
},
{
"epoch": 9.27,
"grad_norm": 0.04632095247507095,
"learning_rate": 8.347341645056372e-07,
"loss": 0.0004,
"step": 4876
},
{
"epoch": 9.27,
"grad_norm": 0.15193146467208862,
"learning_rate": 8.304123946974929e-07,
"loss": 0.0005,
"step": 4877
},
{
"epoch": 9.27,
"grad_norm": 0.030862674117088318,
"learning_rate": 8.261016847361736e-07,
"loss": 0.0001,
"step": 4878
},
{
"epoch": 9.28,
"grad_norm": 0.07090532779693604,
"learning_rate": 8.218020362561218e-07,
"loss": 0.0002,
"step": 4879
},
{
"epoch": 9.28,
"grad_norm": 0.06303498893976212,
"learning_rate": 8.175134508875703e-07,
"loss": 0.0004,
"step": 4880
},
{
"epoch": 9.28,
"grad_norm": 0.03159960359334946,
"learning_rate": 8.132359302565717e-07,
"loss": 0.0002,
"step": 4881
},
{
"epoch": 9.28,
"grad_norm": 0.022098660469055176,
"learning_rate": 8.08969475984972e-07,
"loss": 0.0002,
"step": 4882
},
{
"epoch": 9.28,
"grad_norm": 0.025954945012927055,
"learning_rate": 8.04714089690427e-07,
"loss": 0.0002,
"step": 4883
},
{
"epoch": 9.29,
"grad_norm": 0.065904401242733,
"learning_rate": 8.004697729863997e-07,
"loss": 0.0004,
"step": 4884
},
{
"epoch": 9.29,
"grad_norm": 0.0639222264289856,
"learning_rate": 7.962365274821459e-07,
"loss": 0.0006,
"step": 4885
},
{
"epoch": 9.29,
"grad_norm": 0.03785884752869606,
"learning_rate": 7.920143547827352e-07,
"loss": 0.0001,
"step": 4886
},
{
"epoch": 9.29,
"grad_norm": 0.04562464728951454,
"learning_rate": 7.878032564890269e-07,
"loss": 0.0002,
"step": 4887
},
{
"epoch": 9.29,
"grad_norm": 0.05709375441074371,
"learning_rate": 7.836032341976907e-07,
"loss": 0.0004,
"step": 4888
},
{
"epoch": 9.29,
"grad_norm": 0.03707835078239441,
"learning_rate": 7.794142895011891e-07,
"loss": 0.0001,
"step": 4889
},
{
"epoch": 9.3,
"grad_norm": 0.01873670145869255,
"learning_rate": 7.752364239878018e-07,
"loss": 0.0001,
"step": 4890
},
{
"epoch": 9.3,
"grad_norm": 0.1427837312221527,
"learning_rate": 7.710696392415751e-07,
"loss": 0.0005,
"step": 4891
},
{
"epoch": 9.3,
"grad_norm": 0.010899386368691921,
"learning_rate": 7.669139368423916e-07,
"loss": 0.0002,
"step": 4892
},
{
"epoch": 9.3,
"grad_norm": 0.31745079159736633,
"learning_rate": 7.627693183659046e-07,
"loss": 0.0006,
"step": 4893
},
{
"epoch": 9.3,
"grad_norm": 0.12909862399101257,
"learning_rate": 7.586357853835768e-07,
"loss": 0.0004,
"step": 4894
},
{
"epoch": 9.31,
"grad_norm": 0.11416283994913101,
"learning_rate": 7.545133394626647e-07,
"loss": 0.0007,
"step": 4895
},
{
"epoch": 9.31,
"grad_norm": 0.0051935408264398575,
"learning_rate": 7.504019821662211e-07,
"loss": 0.0001,
"step": 4896
},
{
"epoch": 9.31,
"grad_norm": 0.13234646618366241,
"learning_rate": 7.46301715053096e-07,
"loss": 0.0003,
"step": 4897
},
{
"epoch": 9.31,
"grad_norm": 0.9631243348121643,
"learning_rate": 7.422125396779323e-07,
"loss": 0.0017,
"step": 4898
},
{
"epoch": 9.31,
"grad_norm": 0.034660227596759796,
"learning_rate": 7.38134457591173e-07,
"loss": 0.0001,
"step": 4899
},
{
"epoch": 9.32,
"grad_norm": 0.04951455071568489,
"learning_rate": 7.340674703390415e-07,
"loss": 0.0001,
"step": 4900
},
{
"epoch": 9.32,
"eval_blimp_filtered_avg": 0.7161194029850746,
"eval_blimp_filtered_std": 0.005002457705955971,
"step": 4900
},
{
"epoch": 9.32,
"eval_blimp_supplement_avg": 0.7758620689655172,
"eval_blimp_supplement_std": 0.01810997589189169,
"step": 4900
},
{
"epoch": 9.32,
"eval_vqa_filtered_avg": 0.32,
"eval_vqa_filtered_std": 0.046882617226215034,
"step": 4900
},
{
"epoch": 9.32,
"eval_winoground_filtered_avg": 0.5,
"eval_winoground_filtered_std": 0.050251890762960605,
"step": 4900
},
{
"epoch": 9.32,
"grad_norm": 0.24500563740730286,
"learning_rate": 7.300115794635709e-07,
"loss": 0.0005,
"step": 4901
},
{
"epoch": 9.32,
"grad_norm": 0.13096943497657776,
"learning_rate": 7.259667865025776e-07,
"loss": 0.0002,
"step": 4902
},
{
"epoch": 9.32,
"grad_norm": 0.05500175803899765,
"learning_rate": 7.21933092989675e-07,
"loss": 0.0003,
"step": 4903
},
{
"epoch": 9.32,
"grad_norm": 0.23469074070453644,
"learning_rate": 7.179105004542629e-07,
"loss": 0.0012,
"step": 4904
},
{
"epoch": 9.33,
"grad_norm": 0.01216525211930275,
"learning_rate": 7.138990104215315e-07,
"loss": 0.0002,
"step": 4905
},
{
"epoch": 9.33,
"grad_norm": 0.10102542489767075,
"learning_rate": 7.098986244124706e-07,
"loss": 0.0002,
"step": 4906
},
{
"epoch": 9.33,
"grad_norm": 0.03193369507789612,
"learning_rate": 7.059093439438469e-07,
"loss": 0.0001,
"step": 4907
},
{
"epoch": 9.33,
"grad_norm": 0.015103736892342567,
"learning_rate": 7.019311705282305e-07,
"loss": 0.0002,
"step": 4908
},
{
"epoch": 9.33,
"grad_norm": 0.034484460949897766,
"learning_rate": 6.97964105673965e-07,
"loss": 0.0002,
"step": 4909
},
{
"epoch": 9.33,
"grad_norm": 0.06266232579946518,
"learning_rate": 6.940081508851969e-07,
"loss": 0.0004,
"step": 4910
},
{
"epoch": 9.34,
"grad_norm": 0.07277169823646545,
"learning_rate": 6.900633076618468e-07,
"loss": 0.0003,
"step": 4911
},
{
"epoch": 9.34,
"grad_norm": 0.09892090409994125,
"learning_rate": 6.861295774996346e-07,
"loss": 0.0005,
"step": 4912
},
{
"epoch": 9.34,
"grad_norm": 0.007053976412862539,
"learning_rate": 6.822069618900606e-07,
"loss": 0.0,
"step": 4913
},
{
"epoch": 9.34,
"grad_norm": 0.02495928481221199,
"learning_rate": 6.782954623204052e-07,
"loss": 0.0002,
"step": 4914
},
{
"epoch": 9.34,
"grad_norm": 0.033663976937532425,
"learning_rate": 6.74395080273742e-07,
"loss": 0.0001,
"step": 4915
},
{
"epoch": 9.35,
"grad_norm": 0.014074107632040977,
"learning_rate": 6.705058172289347e-07,
"loss": 0.0001,
"step": 4916
},
{
"epoch": 9.35,
"grad_norm": 0.046267323195934296,
"learning_rate": 6.66627674660617e-07,
"loss": 0.0002,
"step": 4917
},
{
"epoch": 9.35,
"grad_norm": 0.3941512107849121,
"learning_rate": 6.627606540392095e-07,
"loss": 0.0007,
"step": 4918
},
{
"epoch": 9.35,
"grad_norm": 0.027814319357275963,
"learning_rate": 6.589047568309292e-07,
"loss": 0.0002,
"step": 4919
},
{
"epoch": 9.35,
"grad_norm": 0.028373029083013535,
"learning_rate": 6.550599844977534e-07,
"loss": 0.0002,
"step": 4920
},
{
"epoch": 9.36,
"grad_norm": 0.009086941368877888,
"learning_rate": 6.512263384974659e-07,
"loss": 0.0001,
"step": 4921
},
{
"epoch": 9.36,
"grad_norm": 0.019746480509638786,
"learning_rate": 6.474038202836108e-07,
"loss": 0.0001,
"step": 4922
},
{
"epoch": 9.36,
"grad_norm": 0.03492172062397003,
"learning_rate": 6.435924313055286e-07,
"loss": 0.0002,
"step": 4923
},
{
"epoch": 9.36,
"grad_norm": 0.1553676873445511,
"learning_rate": 6.397921730083234e-07,
"loss": 0.0005,
"step": 4924
},
{
"epoch": 9.36,
"grad_norm": 0.28312820196151733,
"learning_rate": 6.360030468329026e-07,
"loss": 0.0018,
"step": 4925
},
{
"epoch": 9.37,
"grad_norm": 0.02677205018699169,
"learning_rate": 6.322250542159269e-07,
"loss": 0.0003,
"step": 4926
},
{
"epoch": 9.37,
"grad_norm": 0.6547946929931641,
"learning_rate": 6.284581965898539e-07,
"loss": 0.0047,
"step": 4927
},
{
"epoch": 9.37,
"grad_norm": 0.044770289212465286,
"learning_rate": 6.247024753829111e-07,
"loss": 0.0002,
"step": 4928
},
{
"epoch": 9.37,
"grad_norm": 0.02055669017136097,
"learning_rate": 6.209578920191061e-07,
"loss": 0.0001,
"step": 4929
},
{
"epoch": 9.37,
"grad_norm": 0.015533977188169956,
"learning_rate": 6.172244479182232e-07,
"loss": 0.0001,
"step": 4930
},
{
"epoch": 9.37,
"grad_norm": 0.005255147349089384,
"learning_rate": 6.135021444958266e-07,
"loss": 0.0001,
"step": 4931
},
{
"epoch": 9.38,
"grad_norm": 0.011647654697299004,
"learning_rate": 6.097909831632476e-07,
"loss": 0.0001,
"step": 4932
},
{
"epoch": 9.38,
"grad_norm": 0.023795343935489655,
"learning_rate": 6.06090965327597e-07,
"loss": 0.0002,
"step": 4933
},
{
"epoch": 9.38,
"grad_norm": 0.006099463440477848,
"learning_rate": 6.024020923917695e-07,
"loss": 0.0001,
"step": 4934
},
{
"epoch": 9.38,
"grad_norm": 0.06438902020454407,
"learning_rate": 5.987243657544161e-07,
"loss": 0.0003,
"step": 4935
},
{
"epoch": 9.38,
"grad_norm": 0.04437414929270744,
"learning_rate": 5.950577868099816e-07,
"loss": 0.0003,
"step": 4936
},
{
"epoch": 9.39,
"grad_norm": 0.013762859627604485,
"learning_rate": 5.914023569486637e-07,
"loss": 0.0001,
"step": 4937
},
{
"epoch": 9.39,
"grad_norm": 0.18521875143051147,
"learning_rate": 5.877580775564539e-07,
"loss": 0.0016,
"step": 4938
},
{
"epoch": 9.39,
"grad_norm": 0.010554071515798569,
"learning_rate": 5.841249500151003e-07,
"loss": 0.0001,
"step": 4939
},
{
"epoch": 9.39,
"grad_norm": 0.02354867197573185,
"learning_rate": 5.805029757021307e-07,
"loss": 0.0001,
"step": 4940
},
{
"epoch": 9.39,
"grad_norm": 0.05421895906329155,
"learning_rate": 5.768921559908369e-07,
"loss": 0.0003,
"step": 4941
},
{
"epoch": 9.4,
"grad_norm": 0.01537521742284298,
"learning_rate": 5.732924922502869e-07,
"loss": 0.0001,
"step": 4942
},
{
"epoch": 9.4,
"grad_norm": 0.016649531200528145,
"learning_rate": 5.697039858453191e-07,
"loss": 0.0001,
"step": 4943
},
{
"epoch": 9.4,
"grad_norm": 0.033578064292669296,
"learning_rate": 5.661266381365383e-07,
"loss": 0.0002,
"step": 4944
},
{
"epoch": 9.4,
"grad_norm": 0.02272622287273407,
"learning_rate": 5.625604504803261e-07,
"loss": 0.0002,
"step": 4945
},
{
"epoch": 9.4,
"grad_norm": 0.02870124951004982,
"learning_rate": 5.590054242288145e-07,
"loss": 0.0002,
"step": 4946
},
{
"epoch": 9.4,
"grad_norm": 0.015264256857335567,
"learning_rate": 5.554615607299284e-07,
"loss": 0.0002,
"step": 4947
},
{
"epoch": 9.41,
"grad_norm": 0.10699054598808289,
"learning_rate": 5.519288613273398e-07,
"loss": 0.0005,
"step": 4948
},
{
"epoch": 9.41,
"grad_norm": 0.027672646567225456,
"learning_rate": 5.484073273605006e-07,
"loss": 0.0001,
"step": 4949
},
{
"epoch": 9.41,
"grad_norm": 0.10436788946390152,
"learning_rate": 5.448969601646225e-07,
"loss": 0.0008,
"step": 4950
},
{
"epoch": 9.41,
"grad_norm": 0.24571040272712708,
"learning_rate": 5.41397761070681e-07,
"loss": 0.001,
"step": 4951
},
{
"epoch": 9.41,
"grad_norm": 0.05248498544096947,
"learning_rate": 5.379097314054249e-07,
"loss": 0.0007,
"step": 4952
},
{
"epoch": 9.42,
"grad_norm": 0.308822363615036,
"learning_rate": 5.344328724913628e-07,
"loss": 0.0009,
"step": 4953
},
{
"epoch": 9.42,
"grad_norm": 0.10393083840608597,
"learning_rate": 5.309671856467768e-07,
"loss": 0.0018,
"step": 4954
},
{
"epoch": 9.42,
"grad_norm": 0.0413050502538681,
"learning_rate": 5.275126721856926e-07,
"loss": 0.0002,
"step": 4955
},
{
"epoch": 9.42,
"grad_norm": 0.0076439268887043,
"learning_rate": 5.240693334179259e-07,
"loss": 0.0001,
"step": 4956
},
{
"epoch": 9.42,
"grad_norm": 0.044137198477983475,
"learning_rate": 5.206371706490254e-07,
"loss": 0.0003,
"step": 4957
},
{
"epoch": 9.43,
"grad_norm": 0.05612409487366676,
"learning_rate": 5.17216185180337e-07,
"loss": 0.0001,
"step": 4958
},
{
"epoch": 9.43,
"grad_norm": 0.013735366053879261,
"learning_rate": 5.138063783089364e-07,
"loss": 0.0002,
"step": 4959
},
{
"epoch": 9.43,
"grad_norm": 0.057997267693281174,
"learning_rate": 5.104077513276828e-07,
"loss": 0.0003,
"step": 4960
},
{
"epoch": 9.43,
"grad_norm": 0.08178876340389252,
"learning_rate": 5.070203055251855e-07,
"loss": 0.0002,
"step": 4961
},
{
"epoch": 9.43,
"grad_norm": 0.004258302040398121,
"learning_rate": 5.036440421858235e-07,
"loss": 0.0,
"step": 4962
},
{
"epoch": 9.44,
"grad_norm": 0.11706515401601791,
"learning_rate": 5.002789625897197e-07,
"loss": 0.0002,
"step": 4963
},
{
"epoch": 9.44,
"grad_norm": 0.047744639217853546,
"learning_rate": 4.969250680127701e-07,
"loss": 0.0001,
"step": 4964
},
{
"epoch": 9.44,
"grad_norm": 0.046125758439302444,
"learning_rate": 4.935823597266342e-07,
"loss": 0.0002,
"step": 4965
},
{
"epoch": 9.44,
"grad_norm": 0.05309942364692688,
"learning_rate": 4.902508389987148e-07,
"loss": 0.0004,
"step": 4966
},
{
"epoch": 9.44,
"grad_norm": 0.012455823831260204,
"learning_rate": 4.869305070921815e-07,
"loss": 0.0001,
"step": 4967
},
{
"epoch": 9.44,
"grad_norm": 0.011095402762293816,
"learning_rate": 4.836213652659638e-07,
"loss": 0.0001,
"step": 4968
},
{
"epoch": 9.45,
"grad_norm": 0.008355801925063133,
"learning_rate": 4.803234147747415e-07,
"loss": 0.0001,
"step": 4969
},
{
"epoch": 9.45,
"grad_norm": 0.14192909002304077,
"learning_rate": 4.770366568689544e-07,
"loss": 0.0003,
"step": 4970
},
{
"epoch": 9.45,
"grad_norm": 0.032362814992666245,
"learning_rate": 4.737610927948055e-07,
"loss": 0.0002,
"step": 4971
},
{
"epoch": 9.45,
"grad_norm": 0.10664493590593338,
"learning_rate": 4.7049672379423813e-07,
"loss": 0.0015,
"step": 4972
},
{
"epoch": 9.45,
"grad_norm": 0.021787380799651146,
"learning_rate": 4.6724355110496883e-07,
"loss": 0.0002,
"step": 4973
},
{
"epoch": 9.46,
"grad_norm": 0.017144711688160896,
"learning_rate": 4.6400157596045433e-07,
"loss": 0.0001,
"step": 4974
},
{
"epoch": 9.46,
"grad_norm": 0.021533582359552383,
"learning_rate": 4.607707995899113e-07,
"loss": 0.0001,
"step": 4975
},
{
"epoch": 9.46,
"grad_norm": 0.11535833775997162,
"learning_rate": 4.5755122321831656e-07,
"loss": 0.0004,
"step": 4976
},
{
"epoch": 9.46,
"grad_norm": 0.0491127148270607,
"learning_rate": 4.5434284806638693e-07,
"loss": 0.0002,
"step": 4977
},
{
"epoch": 9.46,
"grad_norm": 0.005234102252870798,
"learning_rate": 4.511456753506027e-07,
"loss": 0.0001,
"step": 4978
},
{
"epoch": 9.47,
"grad_norm": 0.10590368509292603,
"learning_rate": 4.479597062831942e-07,
"loss": 0.003,
"step": 4979
},
{
"epoch": 9.47,
"grad_norm": 0.015835540369153023,
"learning_rate": 4.447849420721417e-07,
"loss": 0.0001,
"step": 4980
},
{
"epoch": 9.47,
"grad_norm": 0.42672422528266907,
"learning_rate": 4.4162138392117914e-07,
"loss": 0.0017,
"step": 4981
},
{
"epoch": 9.47,
"grad_norm": 0.02100340463221073,
"learning_rate": 4.3846903302979356e-07,
"loss": 0.0001,
"step": 4982
},
{
"epoch": 9.47,
"grad_norm": 0.009627659805119038,
"learning_rate": 4.353278905932123e-07,
"loss": 0.0001,
"step": 4983
},
{
"epoch": 9.48,
"grad_norm": 0.013160955160856247,
"learning_rate": 4.321979578024293e-07,
"loss": 0.0001,
"step": 4984
},
{
"epoch": 9.48,
"grad_norm": 0.3345682919025421,
"learning_rate": 4.290792358441786e-07,
"loss": 0.0012,
"step": 4985
},
{
"epoch": 9.48,
"grad_norm": 0.029298601672053337,
"learning_rate": 4.2597172590094104e-07,
"loss": 0.0002,
"step": 4986
},
{
"epoch": 9.48,
"grad_norm": 0.006016085390001535,
"learning_rate": 4.22875429150954e-07,
"loss": 0.0001,
"step": 4987
},
{
"epoch": 9.48,
"grad_norm": 0.05769047141075134,
"learning_rate": 4.197903467681985e-07,
"loss": 0.0003,
"step": 4988
},
{
"epoch": 9.48,
"grad_norm": 0.027840126305818558,
"learning_rate": 4.167164799223988e-07,
"loss": 0.0001,
"step": 4989
},
{
"epoch": 9.49,
"grad_norm": 0.048520807176828384,
"learning_rate": 4.136538297790393e-07,
"loss": 0.0003,
"step": 4990
},
{
"epoch": 9.49,
"grad_norm": 0.06846524775028229,
"learning_rate": 4.1060239749934445e-07,
"loss": 0.0007,
"step": 4991
},
{
"epoch": 9.49,
"grad_norm": 0.04547520726919174,
"learning_rate": 4.0756218424027547e-07,
"loss": 0.0002,
"step": 4992
},
{
"epoch": 9.49,
"grad_norm": 0.03560828045010567,
"learning_rate": 4.0453319115456357e-07,
"loss": 0.0003,
"step": 4993
},
{
"epoch": 9.49,
"grad_norm": 0.4070132374763489,
"learning_rate": 4.015154193906667e-07,
"loss": 0.0006,
"step": 4994
},
{
"epoch": 9.5,
"grad_norm": 0.06427042931318283,
"learning_rate": 3.9850887009278967e-07,
"loss": 0.0002,
"step": 4995
},
{
"epoch": 9.5,
"grad_norm": 0.03472144901752472,
"learning_rate": 3.955135444008906e-07,
"loss": 0.0001,
"step": 4996
},
{
"epoch": 9.5,
"grad_norm": 0.026199299842119217,
"learning_rate": 3.9252944345066766e-07,
"loss": 0.0002,
"step": 4997
},
{
"epoch": 9.5,
"grad_norm": 0.16900603473186493,
"learning_rate": 3.8955656837355915e-07,
"loss": 0.0004,
"step": 4998
},
{
"epoch": 9.5,
"grad_norm": 0.013709093444049358,
"learning_rate": 3.8659492029675003e-07,
"loss": 0.0001,
"step": 4999
},
{
"epoch": 9.51,
"grad_norm": 0.09828565269708633,
"learning_rate": 3.8364450034317876e-07,
"loss": 0.0003,
"step": 5000
},
{
"epoch": 9.51,
"eval_blimp_filtered_avg": 0.7171641791044776,
"eval_blimp_filtered_std": 0.0050058526403290605,
"step": 5000
},
{
"epoch": 9.51,
"eval_blimp_supplement_avg": 0.7780172413793104,
"eval_blimp_supplement_std": 0.018103492993566235,
"step": 5000
},
{
"epoch": 9.51,
"eval_vqa_filtered_avg": 0.33,
"eval_vqa_filtered_std": 0.04725815626252604,
"step": 5000
},
{
"epoch": 9.51,
"eval_winoground_filtered_avg": 0.5,
"eval_winoground_filtered_std": 0.050251890762960605,
"step": 5000
},
{
"epoch": 9.51,
"grad_norm": 0.020593522116541862,
"learning_rate": 3.8070530963150716e-07,
"loss": 0.0002,
"step": 5001
},
{
"epoch": 9.51,
"grad_norm": 0.27989304065704346,
"learning_rate": 3.777773492761505e-07,
"loss": 0.001,
"step": 5002
},
{
"epoch": 9.51,
"grad_norm": 0.013927803374826908,
"learning_rate": 3.7486062038727065e-07,
"loss": 0.0001,
"step": 5003
},
{
"epoch": 9.51,
"grad_norm": 0.009101859293878078,
"learning_rate": 3.719551240707597e-07,
"loss": 0.0001,
"step": 5004
},
{
"epoch": 9.52,
"grad_norm": 0.010758004151284695,
"learning_rate": 3.6906086142825647e-07,
"loss": 0.0001,
"step": 5005
},
{
"epoch": 9.52,
"grad_norm": 0.06445468962192535,
"learning_rate": 3.661778335571464e-07,
"loss": 0.0002,
"step": 5006
},
{
"epoch": 9.52,
"grad_norm": 0.4593485891819,
"learning_rate": 3.6330604155053846e-07,
"loss": 0.0068,
"step": 5007
},
{
"epoch": 9.52,
"grad_norm": 0.016885166987776756,
"learning_rate": 3.604454864973017e-07,
"loss": 0.0001,
"step": 5008
},
{
"epoch": 9.52,
"grad_norm": 0.017787782475352287,
"learning_rate": 3.575961694820318e-07,
"loss": 0.0001,
"step": 5009
},
{
"epoch": 9.52,
"grad_norm": 0.04883182793855667,
"learning_rate": 3.547580915850612e-07,
"loss": 0.0003,
"step": 5010
},
{
"epoch": 9.53,
"grad_norm": 0.16256083548069,
"learning_rate": 3.519312538824759e-07,
"loss": 0.0006,
"step": 5011
},
{
"epoch": 9.53,
"grad_norm": 0.07412020117044449,
"learning_rate": 3.491156574460852e-07,
"loss": 0.0007,
"step": 5012
},
{
"epoch": 9.53,
"grad_norm": 0.32659345865249634,
"learning_rate": 3.4631130334344173e-07,
"loss": 0.0009,
"step": 5013
},
{
"epoch": 9.53,
"grad_norm": 0.2741422951221466,
"learning_rate": 3.4351819263783503e-07,
"loss": 0.0007,
"step": 5014
},
{
"epoch": 9.53,
"grad_norm": 0.041785940527915955,
"learning_rate": 3.4073632638829455e-07,
"loss": 0.0001,
"step": 5015
},
{
"epoch": 9.54,
"grad_norm": 0.07768747210502625,
"learning_rate": 3.3796570564957997e-07,
"loss": 0.0009,
"step": 5016
},
{
"epoch": 9.54,
"grad_norm": 0.3752177059650421,
"learning_rate": 3.3520633147219427e-07,
"loss": 0.0007,
"step": 5017
},
{
"epoch": 9.54,
"grad_norm": 0.011692519299685955,
"learning_rate": 3.3245820490237056e-07,
"loss": 0.0001,
"step": 5018
},
{
"epoch": 9.54,
"grad_norm": 0.08288253098726273,
"learning_rate": 3.2972132698208204e-07,
"loss": 0.0003,
"step": 5019
},
{
"epoch": 9.54,
"grad_norm": 0.017185943201184273,
"learning_rate": 3.269956987490319e-07,
"loss": 0.0001,
"step": 5020
},
{
"epoch": 9.55,
"grad_norm": 0.03415300324559212,
"learning_rate": 3.2428132123666023e-07,
"loss": 0.0002,
"step": 5021
},
{
"epoch": 9.55,
"grad_norm": 0.021807318553328514,
"learning_rate": 3.2157819547414704e-07,
"loss": 0.0001,
"step": 5022
},
{
"epoch": 9.55,
"grad_norm": 0.07413138449192047,
"learning_rate": 3.188863224863958e-07,
"loss": 0.0002,
"step": 5023
},
{
"epoch": 9.55,
"grad_norm": 0.014432262629270554,
"learning_rate": 3.162057032940469e-07,
"loss": 0.0001,
"step": 5024
},
{
"epoch": 9.55,
"grad_norm": 0.09639009088277817,
"learning_rate": 3.1353633891348045e-07,
"loss": 0.0003,
"step": 5025
},
{
"epoch": 9.56,
"grad_norm": 0.16770534217357635,
"learning_rate": 3.108782303568003e-07,
"loss": 0.0009,
"step": 5026
},
{
"epoch": 9.56,
"grad_norm": 0.07264440506696701,
"learning_rate": 3.082313786318469e-07,
"loss": 0.0005,
"step": 5027
},
{
"epoch": 9.56,
"grad_norm": 0.10646619647741318,
"learning_rate": 3.055957847421975e-07,
"loss": 0.0007,
"step": 5028
},
{
"epoch": 9.56,
"grad_norm": 0.02914976142346859,
"learning_rate": 3.029714496871461e-07,
"loss": 0.0001,
"step": 5029
},
{
"epoch": 9.56,
"grad_norm": 0.4346679747104645,
"learning_rate": 3.0035837446173e-07,
"loss": 0.0008,
"step": 5030
},
{
"epoch": 9.56,
"grad_norm": 0.13064277172088623,
"learning_rate": 2.9775656005672003e-07,
"loss": 0.0008,
"step": 5031
},
{
"epoch": 9.57,
"grad_norm": 0.06084633618593216,
"learning_rate": 2.951660074586071e-07,
"loss": 0.0007,
"step": 5032
},
{
"epoch": 9.57,
"grad_norm": 0.051732324063777924,
"learning_rate": 2.925867176496155e-07,
"loss": 0.0001,
"step": 5033
},
{
"epoch": 9.57,
"grad_norm": 0.03135710954666138,
"learning_rate": 2.90018691607703e-07,
"loss": 0.0002,
"step": 5034
},
{
"epoch": 9.57,
"grad_norm": 0.1287924200296402,
"learning_rate": 2.87461930306554e-07,
"loss": 0.001,
"step": 5035
},
{
"epoch": 9.57,
"grad_norm": 0.011827484704554081,
"learning_rate": 2.8491643471557637e-07,
"loss": 0.0002,
"step": 5036
},
{
"epoch": 9.58,
"grad_norm": 0.022266222164034843,
"learning_rate": 2.823822057999248e-07,
"loss": 0.0001,
"step": 5037
},
{
"epoch": 9.58,
"grad_norm": 0.038436055183410645,
"learning_rate": 2.798592445204573e-07,
"loss": 0.0002,
"step": 5038
},
{
"epoch": 9.58,
"grad_norm": 0.04642660915851593,
"learning_rate": 2.7734755183377534e-07,
"loss": 0.0002,
"step": 5039
},
{
"epoch": 9.58,
"grad_norm": 0.013666420243680477,
"learning_rate": 2.748471286922072e-07,
"loss": 0.0001,
"step": 5040
},
{
"epoch": 9.58,
"grad_norm": 0.029122812673449516,
"learning_rate": 2.723579760438011e-07,
"loss": 0.0002,
"step": 5041
},
{
"epoch": 9.59,
"grad_norm": 0.02065826579928398,
"learning_rate": 2.698800948323388e-07,
"loss": 0.0002,
"step": 5042
},
{
"epoch": 9.59,
"grad_norm": 0.288931667804718,
"learning_rate": 2.674134859973221e-07,
"loss": 0.0008,
"step": 5043
},
{
"epoch": 9.59,
"grad_norm": 0.03982876241207123,
"learning_rate": 2.6495815047398955e-07,
"loss": 0.0003,
"step": 5044
},
{
"epoch": 9.59,
"grad_norm": 0.011180124245584011,
"learning_rate": 2.625140891932898e-07,
"loss": 0.0001,
"step": 5045
},
{
"epoch": 9.59,
"grad_norm": 0.02884078398346901,
"learning_rate": 2.600813030819116e-07,
"loss": 0.0002,
"step": 5046
},
{
"epoch": 9.6,
"grad_norm": 0.007019575219601393,
"learning_rate": 2.5765979306225706e-07,
"loss": 0.0001,
"step": 5047
},
{
"epoch": 9.6,
"grad_norm": 0.018649091944098473,
"learning_rate": 2.552495600524651e-07,
"loss": 0.0001,
"step": 5048
},
{
"epoch": 9.6,
"grad_norm": 0.011655835434794426,
"learning_rate": 2.5285060496638477e-07,
"loss": 0.0001,
"step": 5049
},
{
"epoch": 9.6,
"grad_norm": 0.043903231620788574,
"learning_rate": 2.5046292871360177e-07,
"loss": 0.0004,
"step": 5050
},
{
"epoch": 9.6,
"grad_norm": 0.048086900264024734,
"learning_rate": 2.4808653219941523e-07,
"loss": 0.0004,
"step": 5051
},
{
"epoch": 9.6,
"grad_norm": 0.30672046542167664,
"learning_rate": 2.457214163248578e-07,
"loss": 0.0022,
"step": 5052
},
{
"epoch": 9.61,
"grad_norm": 0.01766509935259819,
"learning_rate": 2.4336758198666877e-07,
"loss": 0.0001,
"step": 5053
},
{
"epoch": 9.61,
"grad_norm": 0.01162694301456213,
"learning_rate": 2.410250300773342e-07,
"loss": 0.0001,
"step": 5054
},
{
"epoch": 9.61,
"grad_norm": 0.018695848062634468,
"learning_rate": 2.3869376148503707e-07,
"loss": 0.0001,
"step": 5055
},
{
"epoch": 9.61,
"grad_norm": 0.006725541781634092,
"learning_rate": 2.3637377709369689e-07,
"loss": 0.0001,
"step": 5056
},
{
"epoch": 9.61,
"grad_norm": 0.010596629232168198,
"learning_rate": 2.3406507778295672e-07,
"loss": 0.0001,
"step": 5057
},
{
"epoch": 9.62,
"grad_norm": 0.1358325481414795,
"learning_rate": 2.3176766442816633e-07,
"loss": 0.0003,
"step": 5058
},
{
"epoch": 9.62,
"grad_norm": 0.011646448634564877,
"learning_rate": 2.294815379004156e-07,
"loss": 0.0,
"step": 5059
},
{
"epoch": 9.62,
"grad_norm": 0.059001922607421875,
"learning_rate": 2.2720669906649783e-07,
"loss": 0.0008,
"step": 5060
},
{
"epoch": 9.62,
"grad_norm": 0.2614803910255432,
"learning_rate": 2.2494314878893306e-07,
"loss": 0.0006,
"step": 5061
},
{
"epoch": 9.62,
"grad_norm": 0.017351586371660233,
"learning_rate": 2.226908879259648e-07,
"loss": 0.0001,
"step": 5062
},
{
"epoch": 9.63,
"grad_norm": 0.011136366054415703,
"learning_rate": 2.204499173315566e-07,
"loss": 0.0001,
"step": 5063
},
{
"epoch": 9.63,
"grad_norm": 0.013217462226748466,
"learning_rate": 2.182202378553788e-07,
"loss": 0.0001,
"step": 5064
},
{
"epoch": 9.63,
"grad_norm": 0.36353641748428345,
"learning_rate": 2.1600185034283515e-07,
"loss": 0.0011,
"step": 5065
},
{
"epoch": 9.63,
"grad_norm": 0.018435906618833542,
"learning_rate": 2.137947556350428e-07,
"loss": 0.0001,
"step": 5066
},
{
"epoch": 9.63,
"grad_norm": 0.004592916462570429,
"learning_rate": 2.115989545688357e-07,
"loss": 0.0001,
"step": 5067
},
{
"epoch": 9.63,
"grad_norm": 0.0943123996257782,
"learning_rate": 2.0941444797676457e-07,
"loss": 0.0003,
"step": 5068
},
{
"epoch": 9.64,
"grad_norm": 0.0550452396273613,
"learning_rate": 2.072412366871035e-07,
"loss": 0.0003,
"step": 5069
},
{
"epoch": 9.64,
"grad_norm": 0.023152349516749382,
"learning_rate": 2.0507932152383668e-07,
"loss": 0.0001,
"step": 5070
},
{
"epoch": 9.64,
"grad_norm": 0.040073834359645844,
"learning_rate": 2.0292870330667178e-07,
"loss": 0.0001,
"step": 5071
},
{
"epoch": 9.64,
"grad_norm": 0.022386405616998672,
"learning_rate": 2.0078938285102988e-07,
"loss": 0.0001,
"step": 5072
},
{
"epoch": 9.64,
"grad_norm": 0.05149734765291214,
"learning_rate": 1.9866136096804544e-07,
"loss": 0.0003,
"step": 5073
},
{
"epoch": 9.65,
"grad_norm": 0.08191116899251938,
"learning_rate": 1.9654463846457637e-07,
"loss": 0.0003,
"step": 5074
},
{
"epoch": 9.65,
"grad_norm": 0.09250998497009277,
"learning_rate": 1.944392161431907e-07,
"loss": 0.0009,
"step": 5075
},
{
"epoch": 9.65,
"grad_norm": 0.10803099721670151,
"learning_rate": 1.9234509480217323e-07,
"loss": 0.0011,
"step": 5076
},
{
"epoch": 9.65,
"grad_norm": 0.04536407068371773,
"learning_rate": 1.9026227523552543e-07,
"loss": 0.0002,
"step": 5077
},
{
"epoch": 9.65,
"grad_norm": 0.24139192700386047,
"learning_rate": 1.8819075823295896e-07,
"loss": 0.0003,
"step": 5078
},
{
"epoch": 9.66,
"grad_norm": 0.1547284871339798,
"learning_rate": 1.861305445799122e-07,
"loss": 0.0008,
"step": 5079
},
{
"epoch": 9.66,
"grad_norm": 0.021302590146660805,
"learning_rate": 1.8408163505751697e-07,
"loss": 0.0001,
"step": 5080
},
{
"epoch": 9.66,
"grad_norm": 0.03541111946105957,
"learning_rate": 1.8204403044264183e-07,
"loss": 0.0002,
"step": 5081
},
{
"epoch": 9.66,
"grad_norm": 0.04827214777469635,
"learning_rate": 1.8001773150785218e-07,
"loss": 0.0005,
"step": 5082
},
{
"epoch": 9.66,
"grad_norm": 0.07363447546958923,
"learning_rate": 1.780027390214367e-07,
"loss": 0.0002,
"step": 5083
},
{
"epoch": 9.67,
"grad_norm": 0.01855512335896492,
"learning_rate": 1.75999053747391e-07,
"loss": 0.0001,
"step": 5084
},
{
"epoch": 9.67,
"grad_norm": 0.16889849305152893,
"learning_rate": 1.74006676445424e-07,
"loss": 0.0015,
"step": 5085
},
{
"epoch": 9.67,
"grad_norm": 0.037495117634534836,
"learning_rate": 1.7202560787096476e-07,
"loss": 0.0002,
"step": 5086
},
{
"epoch": 9.67,
"grad_norm": 0.05085058882832527,
"learning_rate": 1.700558487751458e-07,
"loss": 0.0003,
"step": 5087
},
{
"epoch": 9.67,
"grad_norm": 0.2236286848783493,
"learning_rate": 1.68097399904813e-07,
"loss": 0.0003,
"step": 5088
},
{
"epoch": 9.67,
"grad_norm": 0.05796332284808159,
"learning_rate": 1.6615026200252904e-07,
"loss": 0.0002,
"step": 5089
},
{
"epoch": 9.68,
"grad_norm": 0.0529366061091423,
"learning_rate": 1.6421443580656003e-07,
"loss": 0.0003,
"step": 5090
},
{
"epoch": 9.68,
"grad_norm": 0.05811135470867157,
"learning_rate": 1.6228992205088888e-07,
"loss": 0.0004,
"step": 5091
},
{
"epoch": 9.68,
"grad_norm": 0.017768077552318573,
"learning_rate": 1.603767214652152e-07,
"loss": 0.0001,
"step": 5092
},
{
"epoch": 9.68,
"grad_norm": 0.061207979917526245,
"learning_rate": 1.5847483477493207e-07,
"loss": 0.0002,
"step": 5093
},
{
"epoch": 9.68,
"grad_norm": 0.1127825528383255,
"learning_rate": 1.56584262701156e-07,
"loss": 0.0008,
"step": 5094
},
{
"epoch": 9.69,
"grad_norm": 0.17140009999275208,
"learning_rate": 1.5470500596071025e-07,
"loss": 0.0006,
"step": 5095
},
{
"epoch": 9.69,
"grad_norm": 0.1590358465909958,
"learning_rate": 1.528370652661315e-07,
"loss": 0.0008,
"step": 5096
},
{
"epoch": 9.69,
"grad_norm": 0.10543307662010193,
"learning_rate": 1.509804413256566e-07,
"loss": 0.0003,
"step": 5097
},
{
"epoch": 9.69,
"grad_norm": 0.019123028963804245,
"learning_rate": 1.4913513484323904e-07,
"loss": 0.0001,
"step": 5098
},
{
"epoch": 9.69,
"grad_norm": 0.05143130570650101,
"learning_rate": 1.4730114651854254e-07,
"loss": 0.0002,
"step": 5099
},
{
"epoch": 9.7,
"grad_norm": 0.3774498403072357,
"learning_rate": 1.4547847704693085e-07,
"loss": 0.0009,
"step": 5100
},
{
"epoch": 9.7,
"eval_blimp_filtered_avg": 0.716865671641791,
"eval_blimp_filtered_std": 0.0050013555432687025,
"step": 5100
},
{
"epoch": 9.7,
"eval_blimp_supplement_avg": 0.7823275862068966,
"eval_blimp_supplement_std": 0.01801770012633094,
"step": 5100
},
{
"epoch": 9.7,
"eval_vqa_filtered_avg": 0.32,
"eval_vqa_filtered_std": 0.046882617226215034,
"step": 5100
},
{
"epoch": 9.7,
"eval_winoground_filtered_avg": 0.52,
"eval_winoground_filtered_std": 0.05021167315686779,
"step": 5100
},
{
"epoch": 9.7,
"grad_norm": 0.041449833661317825,
"learning_rate": 1.4366712711948448e-07,
"loss": 0.0002,
"step": 5101
},
{
"epoch": 9.7,
"grad_norm": 0.318409264087677,
"learning_rate": 1.418670974229874e-07,
"loss": 0.0005,
"step": 5102
},
{
"epoch": 9.7,
"grad_norm": 0.03648805990815163,
"learning_rate": 1.4007838863993704e-07,
"loss": 0.0003,
"step": 5103
},
{
"epoch": 9.7,
"grad_norm": 0.14228251576423645,
"learning_rate": 1.3830100144853086e-07,
"loss": 0.0007,
"step": 5104
},
{
"epoch": 9.71,
"grad_norm": 0.059719327837228775,
"learning_rate": 1.3653493652267645e-07,
"loss": 0.0002,
"step": 5105
},
{
"epoch": 9.71,
"grad_norm": 0.01144147478044033,
"learning_rate": 1.3478019453199152e-07,
"loss": 0.0001,
"step": 5106
},
{
"epoch": 9.71,
"grad_norm": 0.047510016709566116,
"learning_rate": 1.3303677614179722e-07,
"loss": 0.0003,
"step": 5107
},
{
"epoch": 9.71,
"grad_norm": 0.1313295215368271,
"learning_rate": 1.313046820131214e-07,
"loss": 0.0017,
"step": 5108
},
{
"epoch": 9.71,
"grad_norm": 0.4057360589504242,
"learning_rate": 1.2958391280270544e-07,
"loss": 0.0009,
"step": 5109
},
{
"epoch": 9.71,
"grad_norm": 0.08212651312351227,
"learning_rate": 1.2787446916298407e-07,
"loss": 0.0004,
"step": 5110
},
{
"epoch": 9.72,
"grad_norm": 0.10862787067890167,
"learning_rate": 1.261763517421055e-07,
"loss": 0.0008,
"step": 5111
},
{
"epoch": 9.72,
"grad_norm": 0.06154458597302437,
"learning_rate": 1.2448956118392142e-07,
"loss": 0.0003,
"step": 5112
},
{
"epoch": 9.72,
"grad_norm": 0.11704808473587036,
"learning_rate": 1.2281409812799683e-07,
"loss": 0.0003,
"step": 5113
},
{
"epoch": 9.72,
"grad_norm": 0.22119879722595215,
"learning_rate": 1.2114996320958694e-07,
"loss": 0.0009,
"step": 5114
},
{
"epoch": 9.72,
"grad_norm": 0.08210884779691696,
"learning_rate": 1.1949715705966036e-07,
"loss": 0.0004,
"step": 5115
},
{
"epoch": 9.73,
"grad_norm": 0.17698217928409576,
"learning_rate": 1.1785568030489579e-07,
"loss": 0.0003,
"step": 5116
},
{
"epoch": 9.73,
"grad_norm": 0.09891340881586075,
"learning_rate": 1.1622553356766541e-07,
"loss": 0.0002,
"step": 5117
},
{
"epoch": 9.73,
"grad_norm": 0.43650150299072266,
"learning_rate": 1.1460671746605144e-07,
"loss": 0.0018,
"step": 5118
},
{
"epoch": 9.73,
"grad_norm": 0.04277857020497322,
"learning_rate": 1.129992326138396e-07,
"loss": 0.0003,
"step": 5119
},
{
"epoch": 9.73,
"grad_norm": 0.033652398735284805,
"learning_rate": 1.1140307962051565e-07,
"loss": 0.0001,
"step": 5120
},
{
"epoch": 9.74,
"grad_norm": 0.03484876826405525,
"learning_rate": 1.0981825909127885e-07,
"loss": 0.0002,
"step": 5121
},
{
"epoch": 9.74,
"grad_norm": 0.028026839718222618,
"learning_rate": 1.0824477162701851e-07,
"loss": 0.0003,
"step": 5122
},
{
"epoch": 9.74,
"grad_norm": 0.02924318052828312,
"learning_rate": 1.0668261782433408e-07,
"loss": 0.0002,
"step": 5123
},
{
"epoch": 9.74,
"grad_norm": 0.08317159116268158,
"learning_rate": 1.0513179827552844e-07,
"loss": 0.0003,
"step": 5124
},
{
"epoch": 9.74,
"grad_norm": 0.07549761980772018,
"learning_rate": 1.0359231356860122e-07,
"loss": 0.0006,
"step": 5125
},
{
"epoch": 9.75,
"grad_norm": 0.07184870541095734,
"learning_rate": 1.020641642872655e-07,
"loss": 0.0005,
"step": 5126
},
{
"epoch": 9.75,
"grad_norm": 0.014542185701429844,
"learning_rate": 1.0054735101092116e-07,
"loss": 0.0001,
"step": 5127
},
{
"epoch": 9.75,
"grad_norm": 0.020369479432702065,
"learning_rate": 9.904187431468481e-08,
"loss": 0.0001,
"step": 5128
},
{
"epoch": 9.75,
"grad_norm": 0.1350410431623459,
"learning_rate": 9.754773476936652e-08,
"loss": 0.0012,
"step": 5129
},
{
"epoch": 9.75,
"grad_norm": 0.028608694672584534,
"learning_rate": 9.606493294147978e-08,
"loss": 0.0002,
"step": 5130
},
{
"epoch": 9.75,
"grad_norm": 0.06712980568408966,
"learning_rate": 9.459346939323487e-08,
"loss": 0.0005,
"step": 5131
},
{
"epoch": 9.76,
"grad_norm": 0.44646647572517395,
"learning_rate": 9.313334468255219e-08,
"loss": 0.0013,
"step": 5132
},
{
"epoch": 9.76,
"grad_norm": 0.017497753724455833,
"learning_rate": 9.168455936304554e-08,
"loss": 0.0002,
"step": 5133
},
{
"epoch": 9.76,
"grad_norm": 0.10285653173923492,
"learning_rate": 9.024711398403218e-08,
"loss": 0.0008,
"step": 5134
},
{
"epoch": 9.76,
"grad_norm": 0.014093542471528053,
"learning_rate": 8.882100909052948e-08,
"loss": 0.0001,
"step": 5135
},
{
"epoch": 9.76,
"grad_norm": 0.020939577370882034,
"learning_rate": 8.74062452232549e-08,
"loss": 0.0001,
"step": 5136
},
{
"epoch": 9.77,
"grad_norm": 0.2866107225418091,
"learning_rate": 8.600282291862605e-08,
"loss": 0.0005,
"step": 5137
},
{
"epoch": 9.77,
"grad_norm": 0.050880853086709976,
"learning_rate": 8.461074270875723e-08,
"loss": 0.0002,
"step": 5138
},
{
"epoch": 9.77,
"grad_norm": 0.03482217341661453,
"learning_rate": 8.323000512146627e-08,
"loss": 0.0002,
"step": 5139
},
{
"epoch": 9.77,
"grad_norm": 0.038232676684856415,
"learning_rate": 8.186061068027439e-08,
"loss": 0.0002,
"step": 5140
},
{
"epoch": 9.77,
"grad_norm": 0.02192629687488079,
"learning_rate": 8.050255990438959e-08,
"loss": 0.0001,
"step": 5141
},
{
"epoch": 9.78,
"grad_norm": 0.033521585166454315,
"learning_rate": 7.915585330872999e-08,
"loss": 0.0002,
"step": 5142
},
{
"epoch": 9.78,
"grad_norm": 0.05203203484416008,
"learning_rate": 7.78204914039038e-08,
"loss": 0.0001,
"step": 5143
},
{
"epoch": 9.78,
"grad_norm": 0.13386893272399902,
"learning_rate": 7.649647469623267e-08,
"loss": 0.0005,
"step": 5144
},
{
"epoch": 9.78,
"grad_norm": 0.04584323614835739,
"learning_rate": 7.518380368771504e-08,
"loss": 0.0002,
"step": 5145
},
{
"epoch": 9.78,
"grad_norm": 0.009608347900211811,
"learning_rate": 7.388247887606614e-08,
"loss": 0.0001,
"step": 5146
},
{
"epoch": 9.79,
"grad_norm": 0.29903072118759155,
"learning_rate": 7.259250075469126e-08,
"loss": 0.0011,
"step": 5147
},
{
"epoch": 9.79,
"grad_norm": 0.017031922936439514,
"learning_rate": 7.131386981268918e-08,
"loss": 0.0001,
"step": 5148
},
{
"epoch": 9.79,
"grad_norm": 0.06049073860049248,
"learning_rate": 7.004658653486873e-08,
"loss": 0.0007,
"step": 5149
},
{
"epoch": 9.79,
"grad_norm": 0.030173858627676964,
"learning_rate": 6.879065140172225e-08,
"loss": 0.0002,
"step": 5150
},
{
"epoch": 9.79,
"grad_norm": 0.3084075152873993,
"learning_rate": 6.754606488945214e-08,
"loss": 0.0007,
"step": 5151
},
{
"epoch": 9.79,
"grad_norm": 0.031630370765924454,
"learning_rate": 6.63128274699476e-08,
"loss": 0.0002,
"step": 5152
},
{
"epoch": 9.8,
"grad_norm": 0.056497927755117416,
"learning_rate": 6.509093961080126e-08,
"loss": 0.0001,
"step": 5153
},
{
"epoch": 9.8,
"grad_norm": 0.06531836092472076,
"learning_rate": 6.38804017752992e-08,
"loss": 0.0008,
"step": 5154
},
{
"epoch": 9.8,
"grad_norm": 0.019538387656211853,
"learning_rate": 6.268121442242425e-08,
"loss": 0.0001,
"step": 5155
},
{
"epoch": 9.8,
"grad_norm": 0.08553823083639145,
"learning_rate": 6.14933780068594e-08,
"loss": 0.0008,
"step": 5156
},
{
"epoch": 9.8,
"grad_norm": 0.08395964652299881,
"learning_rate": 6.03168929789777e-08,
"loss": 0.0003,
"step": 5157
},
{
"epoch": 9.81,
"grad_norm": 0.02502269297838211,
"learning_rate": 5.915175978485232e-08,
"loss": 0.0001,
"step": 5158
},
{
"epoch": 9.81,
"grad_norm": 0.053093165159225464,
"learning_rate": 5.7997978866249915e-08,
"loss": 0.0002,
"step": 5159
},
{
"epoch": 9.81,
"grad_norm": 0.02422378398478031,
"learning_rate": 5.685555066064052e-08,
"loss": 0.0001,
"step": 5160
},
{
"epoch": 9.81,
"grad_norm": 0.07182827591896057,
"learning_rate": 5.572447560117433e-08,
"loss": 0.0002,
"step": 5161
},
{
"epoch": 9.81,
"grad_norm": 0.04788190871477127,
"learning_rate": 5.460475411671495e-08,
"loss": 0.0002,
"step": 5162
},
{
"epoch": 9.82,
"grad_norm": 0.05097171291708946,
"learning_rate": 5.349638663180611e-08,
"loss": 0.0008,
"step": 5163
},
{
"epoch": 9.82,
"grad_norm": 0.013338749296963215,
"learning_rate": 5.2399373566694994e-08,
"loss": 0.0001,
"step": 5164
},
{
"epoch": 9.82,
"grad_norm": 0.09776651114225388,
"learning_rate": 5.1313715337322206e-08,
"loss": 0.0006,
"step": 5165
},
{
"epoch": 9.82,
"grad_norm": 0.20525234937667847,
"learning_rate": 5.0239412355321814e-08,
"loss": 0.0009,
"step": 5166
},
{
"epoch": 9.82,
"grad_norm": 0.014263163320720196,
"learning_rate": 4.917646502802131e-08,
"loss": 0.0001,
"step": 5167
},
{
"epoch": 9.83,
"grad_norm": 0.016629548743367195,
"learning_rate": 4.812487375844832e-08,
"loss": 0.0001,
"step": 5168
},
{
"epoch": 9.83,
"grad_norm": 0.044356685131788254,
"learning_rate": 4.708463894531723e-08,
"loss": 0.0003,
"step": 5169
},
{
"epoch": 9.83,
"grad_norm": 0.012764827348291874,
"learning_rate": 4.605576098304254e-08,
"loss": 0.0002,
"step": 5170
},
{
"epoch": 9.83,
"grad_norm": 0.01902877539396286,
"learning_rate": 4.5038240261728866e-08,
"loss": 0.0001,
"step": 5171
},
{
"epoch": 9.83,
"grad_norm": 0.036682385951280594,
"learning_rate": 4.4032077167174275e-08,
"loss": 0.0003,
"step": 5172
},
{
"epoch": 9.83,
"grad_norm": 0.5175368785858154,
"learning_rate": 4.3037272080876933e-08,
"loss": 0.0007,
"step": 5173
},
{
"epoch": 9.84,
"grad_norm": 0.009186072275042534,
"learning_rate": 4.205382538002178e-08,
"loss": 0.0001,
"step": 5174
},
{
"epoch": 9.84,
"grad_norm": 0.04685364291071892,
"learning_rate": 4.108173743748722e-08,
"loss": 0.0005,
"step": 5175
},
{
"epoch": 9.84,
"grad_norm": 0.19305293262004852,
"learning_rate": 4.0121008621851754e-08,
"loss": 0.0007,
"step": 5176
},
{
"epoch": 9.84,
"grad_norm": 0.1314833015203476,
"learning_rate": 3.917163929737733e-08,
"loss": 0.0004,
"step": 5177
},
{
"epoch": 9.84,
"grad_norm": 0.0060163820162415504,
"learning_rate": 3.8233629824026005e-08,
"loss": 0.0,
"step": 5178
},
{
"epoch": 9.85,
"grad_norm": 0.010106309317052364,
"learning_rate": 3.730698055744996e-08,
"loss": 0.0001,
"step": 5179
},
{
"epoch": 9.85,
"grad_norm": 0.10005632042884827,
"learning_rate": 3.6391691848994824e-08,
"loss": 0.0008,
"step": 5180
},
{
"epoch": 9.85,
"grad_norm": 0.06749608367681503,
"learning_rate": 3.548776404569631e-08,
"loss": 0.0003,
"step": 5181
},
{
"epoch": 9.85,
"grad_norm": 0.014072658494114876,
"learning_rate": 3.459519749029028e-08,
"loss": 0.0001,
"step": 5182
},
{
"epoch": 9.85,
"grad_norm": 0.11204958707094193,
"learning_rate": 3.371399252119267e-08,
"loss": 0.0007,
"step": 5183
},
{
"epoch": 9.86,
"grad_norm": 0.023086877539753914,
"learning_rate": 3.28441494725229e-08,
"loss": 0.0001,
"step": 5184
},
{
"epoch": 9.86,
"grad_norm": 0.017782552167773247,
"learning_rate": 3.198566867408714e-08,
"loss": 0.0001,
"step": 5185
},
{
"epoch": 9.86,
"grad_norm": 0.21713554859161377,
"learning_rate": 3.113855045138503e-08,
"loss": 0.0005,
"step": 5186
},
{
"epoch": 9.86,
"grad_norm": 0.19883304834365845,
"learning_rate": 3.030279512560297e-08,
"loss": 0.0006,
"step": 5187
},
{
"epoch": 9.86,
"grad_norm": 0.006980234291404486,
"learning_rate": 2.9478403013627477e-08,
"loss": 0.0001,
"step": 5188
},
{
"epoch": 9.87,
"grad_norm": 0.03666188195347786,
"learning_rate": 2.8665374428031853e-08,
"loss": 0.0002,
"step": 5189
},
{
"epoch": 9.87,
"grad_norm": 0.030899271368980408,
"learning_rate": 2.786370967707952e-08,
"loss": 0.0002,
"step": 5190
},
{
"epoch": 9.87,
"grad_norm": 0.01776890456676483,
"learning_rate": 2.7073409064727328e-08,
"loss": 0.0001,
"step": 5191
},
{
"epoch": 9.87,
"grad_norm": 0.0329170860350132,
"learning_rate": 2.6294472890625586e-08,
"loss": 0.0001,
"step": 5192
},
{
"epoch": 9.87,
"grad_norm": 0.018565546721220016,
"learning_rate": 2.5526901450111385e-08,
"loss": 0.0001,
"step": 5193
},
{
"epoch": 9.87,
"grad_norm": 0.05696948617696762,
"learning_rate": 2.477069503421192e-08,
"loss": 0.0004,
"step": 5194
},
{
"epoch": 9.88,
"grad_norm": 0.17604228854179382,
"learning_rate": 2.4025853929654505e-08,
"loss": 0.0013,
"step": 5195
},
{
"epoch": 9.88,
"grad_norm": 0.034971173852682114,
"learning_rate": 2.3292378418846572e-08,
"loss": 0.0003,
"step": 5196
},
{
"epoch": 9.88,
"grad_norm": 0.01594848744571209,
"learning_rate": 2.2570268779888992e-08,
"loss": 0.0002,
"step": 5197
},
{
"epoch": 9.88,
"grad_norm": 0.0030603313352912664,
"learning_rate": 2.1859525286576087e-08,
"loss": 0.0,
"step": 5198
},
{
"epoch": 9.88,
"grad_norm": 0.13626140356063843,
"learning_rate": 2.1160148208392293e-08,
"loss": 0.0009,
"step": 5199
},
{
"epoch": 9.89,
"grad_norm": 0.5455458760261536,
"learning_rate": 2.0472137810508827e-08,
"loss": 0.001,
"step": 5200
},
{
"epoch": 9.89,
"eval_blimp_filtered_avg": 0.7165671641791045,
"eval_blimp_filtered_std": 0.005005537975046647,
"step": 5200
},
{
"epoch": 9.89,
"eval_blimp_supplement_avg": 0.7801724137931034,
"eval_blimp_supplement_std": 0.018091817922526954,
"step": 5200
},
{
"epoch": 9.89,
"eval_vqa_filtered_avg": 0.32,
"eval_vqa_filtered_std": 0.046882617226215034,
"step": 5200
},
{
"epoch": 9.89,
"eval_winoground_filtered_avg": 0.5,
"eval_winoground_filtered_std": 0.050251890762960605,
"step": 5200
},
{
"epoch": 9.89,
"grad_norm": 0.017485307529568672,
"learning_rate": 1.9795494353793686e-08,
"loss": 0.0001,
"step": 5201
},
{
"epoch": 9.89,
"grad_norm": 0.20075646042823792,
"learning_rate": 1.9130218094798313e-08,
"loss": 0.0009,
"step": 5202
},
{
"epoch": 9.89,
"grad_norm": 0.01598513126373291,
"learning_rate": 1.8476309285764272e-08,
"loss": 0.0001,
"step": 5203
},
{
"epoch": 9.89,
"grad_norm": 0.12029172480106354,
"learning_rate": 1.7833768174626564e-08,
"loss": 0.0013,
"step": 5204
},
{
"epoch": 9.9,
"grad_norm": 1.0377858877182007,
"learning_rate": 1.7202595005013643e-08,
"loss": 0.0019,
"step": 5205
},
{
"epoch": 9.9,
"grad_norm": 0.12031227350234985,
"learning_rate": 1.6582790016234084e-08,
"loss": 0.001,
"step": 5206
},
{
"epoch": 9.9,
"grad_norm": 0.23325763642787933,
"learning_rate": 1.5974353443289903e-08,
"loss": 0.0006,
"step": 5207
},
{
"epoch": 9.9,
"grad_norm": 1.3542128801345825,
"learning_rate": 1.537728551687989e-08,
"loss": 0.0014,
"step": 5208
},
{
"epoch": 9.9,
"grad_norm": 0.2169056385755539,
"learning_rate": 1.4791586463379636e-08,
"loss": 0.0005,
"step": 5209
},
{
"epoch": 9.9,
"grad_norm": 0.14060214161872864,
"learning_rate": 1.421725650486816e-08,
"loss": 0.0016,
"step": 5210
},
{
"epoch": 9.91,
"grad_norm": 0.8145201802253723,
"learning_rate": 1.3654295859101273e-08,
"loss": 0.0017,
"step": 5211
},
{
"epoch": 9.91,
"grad_norm": 0.09021039307117462,
"learning_rate": 1.3102704739531569e-08,
"loss": 0.0003,
"step": 5212
},
{
"epoch": 9.91,
"grad_norm": 0.009653858840465546,
"learning_rate": 1.2562483355298416e-08,
"loss": 0.0001,
"step": 5213
},
{
"epoch": 9.91,
"grad_norm": 0.3286755681037903,
"learning_rate": 1.2033631911227972e-08,
"loss": 0.0022,
"step": 5214
},
{
"epoch": 9.91,
"grad_norm": 0.09239126741886139,
"learning_rate": 1.1516150607839838e-08,
"loss": 0.0003,
"step": 5215
},
{
"epoch": 9.92,
"grad_norm": 0.0209198035299778,
"learning_rate": 1.1010039641343728e-08,
"loss": 0.0001,
"step": 5216
},
{
"epoch": 9.92,
"grad_norm": 0.6407065987586975,
"learning_rate": 1.0515299203629481e-08,
"loss": 0.0012,
"step": 5217
},
{
"epoch": 9.92,
"grad_norm": 0.03524121269583702,
"learning_rate": 1.0031929482280378e-08,
"loss": 0.0003,
"step": 5218
},
{
"epoch": 9.92,
"grad_norm": 0.6369953155517578,
"learning_rate": 9.559930660576476e-09,
"loss": 0.003,
"step": 5219
},
{
"epoch": 9.92,
"grad_norm": 0.054926712065935135,
"learning_rate": 9.099302917474627e-09,
"loss": 0.0002,
"step": 5220
},
{
"epoch": 9.93,
"grad_norm": 0.043360788375139236,
"learning_rate": 8.650046427625124e-09,
"loss": 0.0002,
"step": 5221
},
{
"epoch": 9.93,
"grad_norm": 0.03207186236977577,
"learning_rate": 8.212161361368375e-09,
"loss": 0.0003,
"step": 5222
},
{
"epoch": 9.93,
"grad_norm": 0.016229044646024704,
"learning_rate": 7.785647884728242e-09,
"loss": 0.0001,
"step": 5223
},
{
"epoch": 9.93,
"grad_norm": 0.004804561380296946,
"learning_rate": 7.370506159425361e-09,
"loss": 0.0,
"step": 5224
},
{
"epoch": 9.93,
"grad_norm": 0.012342993170022964,
"learning_rate": 6.9667363428571605e-09,
"loss": 0.0001,
"step": 5225
},
{
"epoch": 9.94,
"grad_norm": 0.08992448449134827,
"learning_rate": 6.574338588121176e-09,
"loss": 0.0003,
"step": 5226
},
{
"epoch": 9.94,
"grad_norm": 0.10938418656587601,
"learning_rate": 6.1933130439950635e-09,
"loss": 0.0012,
"step": 5227
},
{
"epoch": 9.94,
"grad_norm": 0.2587052881717682,
"learning_rate": 5.823659854946594e-09,
"loss": 0.0023,
"step": 5228
},
{
"epoch": 9.94,
"grad_norm": 0.030825519934296608,
"learning_rate": 5.465379161133655e-09,
"loss": 0.0002,
"step": 5229
},
{
"epoch": 9.94,
"grad_norm": 0.0207489263266325,
"learning_rate": 5.118471098400912e-09,
"loss": 0.0001,
"step": 5230
},
{
"epoch": 9.94,
"grad_norm": 0.008567447774112225,
"learning_rate": 4.782935798279819e-09,
"loss": 0.0001,
"step": 5231
},
{
"epoch": 9.95,
"grad_norm": 0.02328609488904476,
"learning_rate": 4.458773387991943e-09,
"loss": 0.0001,
"step": 5232
},
{
"epoch": 9.95,
"grad_norm": 0.018100500106811523,
"learning_rate": 4.145983990445634e-09,
"loss": 0.0002,
"step": 5233
},
{
"epoch": 9.95,
"grad_norm": 0.03456133231520653,
"learning_rate": 3.844567724236026e-09,
"loss": 0.0003,
"step": 5234
},
{
"epoch": 9.95,
"grad_norm": 0.293224036693573,
"learning_rate": 3.5545247036483654e-09,
"loss": 0.0009,
"step": 5235
},
{
"epoch": 9.95,
"grad_norm": 0.011678933165967464,
"learning_rate": 3.2758550386546848e-09,
"loss": 0.0001,
"step": 5236
},
{
"epoch": 9.96,
"grad_norm": 0.6271359920501709,
"learning_rate": 3.0085588349104685e-09,
"loss": 0.0025,
"step": 5237
},
{
"epoch": 9.96,
"grad_norm": 0.004568392876535654,
"learning_rate": 2.7526361937679767e-09,
"loss": 0.0,
"step": 5238
},
{
"epoch": 9.96,
"grad_norm": 0.006656745448708534,
"learning_rate": 2.5080872122629217e-09,
"loss": 0.0001,
"step": 5239
},
{
"epoch": 9.96,
"grad_norm": 0.01631256565451622,
"learning_rate": 2.274911983111139e-09,
"loss": 0.0001,
"step": 5240
},
{
"epoch": 9.96,
"grad_norm": 0.039879798889160156,
"learning_rate": 2.053110594728569e-09,
"loss": 0.0002,
"step": 5241
},
{
"epoch": 9.97,
"grad_norm": 0.10285880416631699,
"learning_rate": 1.842683131207945e-09,
"loss": 0.0009,
"step": 5242
},
{
"epoch": 9.97,
"grad_norm": 0.6335331201553345,
"learning_rate": 1.6436296723387757e-09,
"loss": 0.0012,
"step": 5243
},
{
"epoch": 9.97,
"grad_norm": 0.04950598254799843,
"learning_rate": 1.455950293590691e-09,
"loss": 0.0001,
"step": 5244
},
{
"epoch": 9.97,
"grad_norm": 0.05579832196235657,
"learning_rate": 1.2796450661267667e-09,
"loss": 0.0005,
"step": 5245
},
{
"epoch": 9.97,
"grad_norm": 0.06540787220001221,
"learning_rate": 1.1147140567868697e-09,
"loss": 0.0008,
"step": 5246
},
{
"epoch": 9.98,
"grad_norm": 0.10935112833976746,
"learning_rate": 9.611573281143038e-10,
"loss": 0.0002,
"step": 5247
},
{
"epoch": 9.98,
"grad_norm": 0.024695981293916702,
"learning_rate": 8.189749383291644e-10,
"loss": 0.0001,
"step": 5248
},
{
"epoch": 9.98,
"grad_norm": 0.03446861356496811,
"learning_rate": 6.881669413383307e-10,
"loss": 0.0003,
"step": 5249
},
{
"epoch": 9.98,
"grad_norm": 1.6899726390838623,
"learning_rate": 5.687333867387956e-10,
"loss": 0.0019,
"step": 5250
},
{
"epoch": 9.98,
"grad_norm": 0.08394700288772583,
"learning_rate": 4.6067431981433593e-10,
"loss": 0.0002,
"step": 5251
},
{
"epoch": 9.98,
"grad_norm": 0.18875232338905334,
"learning_rate": 3.6398978153884266e-10,
"loss": 0.0009,
"step": 5252
},
{
"epoch": 9.99,
"grad_norm": 0.5613245368003845,
"learning_rate": 2.786798085696596e-10,
"loss": 0.0022,
"step": 5253
},
{
"epoch": 9.99,
"grad_norm": 0.008934354409575462,
"learning_rate": 2.0474443325424474e-10,
"loss": 0.0001,
"step": 5254
},
{
"epoch": 9.99,
"grad_norm": 0.020037241280078888,
"learning_rate": 1.421836836201784e-10,
"loss": 0.0001,
"step": 5255
},
{
"epoch": 9.99,
"grad_norm": 0.00964090321213007,
"learning_rate": 9.099758339514708e-11,
"loss": 0.0001,
"step": 5256
},
{
"epoch": 9.99,
"grad_norm": 0.0876900851726532,
"learning_rate": 5.118615198029808e-11,
"loss": 0.0003,
"step": 5257
},
{
"epoch": 10.0,
"grad_norm": 0.15568570792675018,
"learning_rate": 2.2749404473554336e-11,
"loss": 0.0005,
"step": 5258
},
{
"epoch": 10.0,
"grad_norm": 0.006819253321737051,
"learning_rate": 5.68735165629164e-12,
"loss": 0.0001,
"step": 5259
},
{
"epoch": 10.0,
"grad_norm": 0.04345171898603439,
"learning_rate": 0.0,
"loss": 0.0001,
"step": 5260
},
{
"epoch": 10.0,
"step": 5260,
"total_flos": 1.325353398408577e+18,
"train_loss": 0.06126316024861676,
"train_runtime": 12820.2093,
"train_samples_per_second": 52.48,
"train_steps_per_second": 0.41
}
],
"logging_steps": 1.0,
"max_steps": 5260,
"num_input_tokens_seen": 0,
"num_train_epochs": 10,
"save_steps": 1000,
"total_flos": 1.325353398408577e+18,
"train_batch_size": 32,
"trial_name": null,
"trial_params": null
}