Hugging Face's logo

faridkarimli
/
SWIN_Gaudi_60

Image Classification
Transformers
Safetensors
Habana
swinv2
vision
Generated from Trainer
Model card Files Community
faridkarimli commited on
Commit
430dc0e
·
verified ·
1 Parent(s): aac1a87

Training in progress, epoch 1

Browse files
Files changed (10) hide show
  1. README.md +99 -0
  2. all_results.json +16 -0
  3. config.json +0 -0
  4. eval_results.json +11 -0
  5. gaudi_config.json +10 -0
  6. model.safetensors +3 -0
  7. preprocessor_config.json +22 -0
  8. train_results.json +11 -0
  9. trainer_state.json +795 -0
  10. training_args.bin +3 -0
README.md ADDED
@@ -0,0 +1,99 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: transformers
3
+ license: apache-2.0
4
+ base_model: microsoft/swinv2-large-patch4-window12-192-22k
5
+ tags:
6
+ - image-classification
7
+ - vision
8
+ - generated_from_trainer
9
+ metrics:
10
+ - accuracy
11
+ model-index:
12
+ - name: SWIN_Gaudi_v1
13
+ results: []
14
+ ---
15
+
16
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
17
+ should probably proofread and complete it, then remove this comment. -->
18
+
19
+ # SWIN_Gaudi_v1
20
+
21
+ This model is a fine-tuned version of [microsoft/swinv2-large-patch4-window12-192-22k](https://huggingface.co/microsoft/swinv2-large-patch4-window12-192-22k) on an unknown dataset.
22
+ It achieves the following results on the evaluation set:
23
+ - Loss: 4.8090
24
+ - Accuracy: 0.2293
25
+ - Memory Allocated (gb): 1.84
26
+ - Max Memory Allocated (gb): 58.8
27
+ - Total Memory Available (gb): 94.62
28
+
29
+ ## Model description
30
+
31
+ More information needed
32
+
33
+ ## Intended uses & limitations
34
+
35
+ More information needed
36
+
37
+ ## Training and evaluation data
38
+
39
+ More information needed
40
+
41
+ ## Training procedure
42
+
43
+ ### Training hyperparameters
44
+
45
+ The following hyperparameters were used during training:
46
+ - learning_rate: 0.001
47
+ - train_batch_size: 128
48
+ - eval_batch_size: 128
49
+ - seed: 42
50
+ - distributed_type: multi-GPU
51
+ - num_devices: 8
52
+ - total_train_batch_size: 1024
53
+ - total_eval_batch_size: 1024
54
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-06
55
+ - lr_scheduler_type: linear
56
+ - num_epochs: 30.0
57
+
58
+ ### Training results
59
+
60
+ | Training Loss | Epoch | Step | Accuracy | Validation Loss | Memory Allocated (gb) | Allocated (gb) | Memory Available (gb) |
61
+ |:-------------:|:-----:|:-----:|:--------:|:---------------:|:---------------------:|:--------------:|:---------------------:|
62
+ | 5.3071 | 1.0 | 657 | 0.0801 | 6.2406 | 60.21 | 3.2 | 94.62 |
63
+ | 3.1366 | 2.0 | 1314 | 0.1047 | 5.8481 | 60.21 | 3.2 | 94.62 |
64
+ | 2.6048 | 3.0 | 1971 | 0.1238 | 5.5522 | 60.21 | 3.2 | 94.62 |
65
+ | 1.9918 | 4.0 | 2628 | 0.1301 | 5.5551 | 60.21 | 3.2 | 94.62 |
66
+ | 1.8353 | 5.0 | 3285 | 0.1415 | 5.4142 | 60.21 | 3.2 | 94.62 |
67
+ | 1.7262 | 6.0 | 3942 | 0.1495 | 5.4061 | 60.21 | 3.2 | 94.62 |
68
+ | 1.5135 | 7.0 | 4599 | 0.1468 | 5.4261 | 60.21 | 3.2 | 94.62 |
69
+ | 1.4225 | 8.0 | 5256 | 0.1573 | 5.3333 | 60.21 | 3.2 | 94.62 |
70
+ | 1.354 | 9.0 | 5913 | 0.1638 | 5.2205 | 60.21 | 3.2 | 94.62 |
71
+ | 1.2511 | 10.0 | 6570 | 0.1708 | 5.2129 | 60.21 | 3.2 | 94.62 |
72
+ | 1.1742 | 11.0 | 7227 | 0.1724 | 5.2002 | 60.21 | 3.2 | 94.62 |
73
+ | 1.1342 | 12.0 | 7884 | 0.1782 | 5.1635 | 60.21 | 3.2 | 94.62 |
74
+ | 1.0711 | 13.0 | 8541 | 0.1779 | 5.1436 | 60.21 | 3.2 | 94.62 |
75
+ | 0.9971 | 14.0 | 9198 | 0.1817 | 5.1076 | 60.21 | 3.2 | 94.62 |
76
+ | 0.9774 | 15.0 | 9855 | 0.1935 | 4.9076 | 60.21 | 3.2 | 94.62 |
77
+ | 0.9174 | 16.0 | 10512 | 0.1890 | 5.0318 | 60.21 | 3.2 | 94.62 |
78
+ | 0.8675 | 17.0 | 11169 | 0.1951 | 5.0392 | 60.21 | 3.2 | 94.62 |
79
+ | 0.8499 | 18.0 | 11826 | 0.1978 | 5.0243 | 60.21 | 3.2 | 94.62 |
80
+ | 0.8262 | 19.0 | 12483 | 0.1972 | 5.0843 | 60.21 | 3.2 | 94.62 |
81
+ | 0.7623 | 20.0 | 13140 | 0.2048 | 5.0004 | 60.21 | 3.2 | 94.62 |
82
+ | 0.7481 | 21.0 | 13797 | 0.2132 | 4.8428 | 60.24 | 3.2 | 94.62 |
83
+ | 0.7284 | 22.0 | 14454 | 0.2149 | 4.8461 | 60.24 | 3.2 | 94.62 |
84
+ | 0.6834 | 23.0 | 15111 | 0.2159 | 4.8741 | 60.24 | 3.2 | 94.62 |
85
+ | 0.6591 | 24.0 | 15768 | 0.2187 | 4.8993 | 60.24 | 3.2 | 94.62 |
86
+ | 0.6447 | 25.0 | 16425 | 0.2196 | 4.8415 | 60.24 | 3.2 | 94.62 |
87
+ | 0.6107 | 26.0 | 17082 | 0.2216 | 4.8600 | 60.24 | 3.2 | 94.62 |
88
+ | 0.5958 | 27.0 | 17739 | 0.2245 | 4.8391 | 60.24 | 3.2 | 94.62 |
89
+ | 0.5836 | 28.0 | 18396 | 0.2265 | 4.8561 | 60.24 | 3.2 | 94.62 |
90
+ | 0.5547 | 29.0 | 19053 | 0.2295 | 4.7933 | 60.24 | 3.2 | 94.62 |
91
+ | 0.547 | 30.0 | 19710 | 0.2293 | 4.8090 | 60.24 | 3.2 | 94.62 |
92
+
93
+
94
+ ### Framework versions
95
+
96
+ - Transformers 4.45.2
97
+ - Pytorch 2.6.0+hpu_1.20.0-543.git4952fce
98
+ - Datasets 3.5.0
99
+ - Tokenizers 0.20.3
all_results.json ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 30.0,
3
+ "eval_accuracy": 0.22931691219483877,
4
+ "eval_loss": 4.809002876281738,
5
+ "eval_runtime": 1140.029,
6
+ "eval_samples_per_second": 589.298,
7
+ "eval_steps_per_second": 0.576,
8
+ "max_memory_allocated (GB)": 58.8,
9
+ "memory_allocated (GB)": 1.84,
10
+ "total_flos": 2.9333313524800244e+21,
11
+ "total_memory_available (GB)": 94.62,
12
+ "train_loss": 0.0,
13
+ "train_runtime": 0.2168,
14
+ "train_samples_per_second": 92964947.027,
15
+ "train_steps_per_second": 90914.595
16
+ }
config.json ADDED
The diff for this file is too large to render. See raw diff
 
eval_results.json ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 30.0,
3
+ "eval_accuracy": 0.22931691219483877,
4
+ "eval_loss": 4.809002876281738,
5
+ "eval_runtime": 1140.029,
6
+ "eval_samples_per_second": 589.298,
7
+ "eval_steps_per_second": 0.576,
8
+ "max_memory_allocated (GB)": 58.8,
9
+ "memory_allocated (GB)": 1.84,
10
+ "total_memory_available (GB)": 94.62
11
+ }
gaudi_config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "autocast_bf16_ops": null,
3
+ "autocast_fp32_ops": null,
4
+ "optimum_version": "1.24.0",
5
+ "transformers_version": "4.45.2",
6
+ "use_dynamic_shapes": false,
7
+ "use_fused_adam": true,
8
+ "use_fused_clip_norm": true,
9
+ "use_torch_autocast": true
10
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fa33f6a2bf7af4b9a6e430bc18e3a1f68f0f4b4fd5c793acf0470e8828bb912f
3
+ size 876174524
preprocessor_config.json ADDED
@@ -0,0 +1,22 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_normalize": true,
3
+ "do_rescale": true,
4
+ "do_resize": true,
5
+ "image_mean": [
6
+ 0.485,
7
+ 0.456,
8
+ 0.406
9
+ ],
10
+ "image_processor_type": "ViTImageProcessor",
11
+ "image_std": [
12
+ 0.229,
13
+ 0.224,
14
+ 0.225
15
+ ],
16
+ "resample": 3,
17
+ "rescale_factor": 0.00392156862745098,
18
+ "size": {
19
+ "height": 192,
20
+ "width": 192
21
+ }
22
+ }
train_results.json ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 30.0,
3
+ "max_memory_allocated (GB)": 1.42,
4
+ "memory_allocated (GB)": 1.42,
5
+ "total_flos": 2.9333313524800244e+21,
6
+ "total_memory_available (GB)": 94.62,
7
+ "train_loss": 0.0,
8
+ "train_runtime": 0.2168,
9
+ "train_samples_per_second": 92964947.027,
10
+ "train_steps_per_second": 90914.595
11
+ }
trainer_state.json ADDED
@@ -0,0 +1,795 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 30.0,
5
+ "eval_steps": 500,
6
+ "global_step": 19710,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.76103500761035,
13
+ "grad_norm": 1.1888039112091064,
14
+ "learning_rate": 0.0009746321664129883,
15
+ "loss": 5.3071,
16
+ "max_memory_allocated (GB)": 5.75,
17
+ "memory_allocated (GB)": 3.2,
18
+ "step": 500,
19
+ "total_memory_available (GB)": 94.62
20
+ },
21
+ {
22
+ "epoch": 1.0,
23
+ "eval_accuracy": 0.08006198116451355,
24
+ "eval_loss": 6.240572929382324,
25
+ "eval_runtime": 1138.1454,
26
+ "eval_samples_per_second": 590.273,
27
+ "eval_steps_per_second": 0.577,
28
+ "max_memory_allocated (GB)": 60.21,
29
+ "memory_allocated (GB)": 3.2,
30
+ "step": 657,
31
+ "total_memory_available (GB)": 94.62
32
+ },
33
+ {
34
+ "epoch": 1.5220700152207,
35
+ "grad_norm": 1.0501078367233276,
36
+ "learning_rate": 0.0009492643328259766,
37
+ "loss": 3.1366,
38
+ "max_memory_allocated (GB)": 60.21,
39
+ "memory_allocated (GB)": 3.2,
40
+ "step": 1000,
41
+ "total_memory_available (GB)": 94.62
42
+ },
43
+ {
44
+ "epoch": 2.0,
45
+ "eval_accuracy": 0.10465945339281382,
46
+ "eval_loss": 5.848066329956055,
47
+ "eval_runtime": 1134.9424,
48
+ "eval_samples_per_second": 591.939,
49
+ "eval_steps_per_second": 0.579,
50
+ "max_memory_allocated (GB)": 60.21,
51
+ "memory_allocated (GB)": 3.2,
52
+ "step": 1314,
53
+ "total_memory_available (GB)": 94.62
54
+ },
55
+ {
56
+ "epoch": 2.2831050228310503,
57
+ "grad_norm": 0.972637414932251,
58
+ "learning_rate": 0.0009238964992389651,
59
+ "loss": 2.6048,
60
+ "max_memory_allocated (GB)": 60.21,
61
+ "memory_allocated (GB)": 3.2,
62
+ "step": 1500,
63
+ "total_memory_available (GB)": 94.62
64
+ },
65
+ {
66
+ "epoch": 3.0,
67
+ "eval_accuracy": 0.12382538697294054,
68
+ "eval_loss": 5.552162170410156,
69
+ "eval_runtime": 1132.6347,
70
+ "eval_samples_per_second": 593.145,
71
+ "eval_steps_per_second": 0.58,
72
+ "max_memory_allocated (GB)": 60.21,
73
+ "memory_allocated (GB)": 3.2,
74
+ "step": 1971,
75
+ "total_memory_available (GB)": 94.62
76
+ },
77
+ {
78
+ "epoch": 3.0441400304414,
79
+ "grad_norm": 0.8711762428283691,
80
+ "learning_rate": 0.0008985286656519534,
81
+ "loss": 2.3103,
82
+ "max_memory_allocated (GB)": 60.21,
83
+ "memory_allocated (GB)": 3.2,
84
+ "step": 2000,
85
+ "total_memory_available (GB)": 94.62
86
+ },
87
+ {
88
+ "epoch": 3.8051750380517504,
89
+ "grad_norm": 0.859586775302887,
90
+ "learning_rate": 0.0008731608320649417,
91
+ "loss": 1.9918,
92
+ "max_memory_allocated (GB)": 60.21,
93
+ "memory_allocated (GB)": 3.2,
94
+ "step": 2500,
95
+ "total_memory_available (GB)": 94.62
96
+ },
97
+ {
98
+ "epoch": 4.0,
99
+ "eval_accuracy": 0.1300875089496098,
100
+ "eval_loss": 5.555095672607422,
101
+ "eval_runtime": 1133.2064,
102
+ "eval_samples_per_second": 592.846,
103
+ "eval_steps_per_second": 0.58,
104
+ "max_memory_allocated (GB)": 60.21,
105
+ "memory_allocated (GB)": 3.2,
106
+ "step": 2628,
107
+ "total_memory_available (GB)": 94.62
108
+ },
109
+ {
110
+ "epoch": 4.566210045662101,
111
+ "grad_norm": 0.8556590676307678,
112
+ "learning_rate": 0.00084779299847793,
113
+ "loss": 1.8353,
114
+ "max_memory_allocated (GB)": 60.21,
115
+ "memory_allocated (GB)": 3.2,
116
+ "step": 3000,
117
+ "total_memory_available (GB)": 94.62
118
+ },
119
+ {
120
+ "epoch": 5.0,
121
+ "eval_accuracy": 0.141504308464954,
122
+ "eval_loss": 5.414204120635986,
123
+ "eval_runtime": 1132.3487,
124
+ "eval_samples_per_second": 593.295,
125
+ "eval_steps_per_second": 0.58,
126
+ "max_memory_allocated (GB)": 60.21,
127
+ "memory_allocated (GB)": 3.2,
128
+ "step": 3285,
129
+ "total_memory_available (GB)": 94.62
130
+ },
131
+ {
132
+ "epoch": 5.327245053272451,
133
+ "grad_norm": 0.8324838280677795,
134
+ "learning_rate": 0.0008224251648909183,
135
+ "loss": 1.7262,
136
+ "max_memory_allocated (GB)": 60.21,
137
+ "memory_allocated (GB)": 3.2,
138
+ "step": 3500,
139
+ "total_memory_available (GB)": 94.62
140
+ },
141
+ {
142
+ "epoch": 6.0,
143
+ "eval_accuracy": 0.14951988413511416,
144
+ "eval_loss": 5.40610933303833,
145
+ "eval_runtime": 1133.7065,
146
+ "eval_samples_per_second": 592.585,
147
+ "eval_steps_per_second": 0.58,
148
+ "max_memory_allocated (GB)": 60.21,
149
+ "memory_allocated (GB)": 3.2,
150
+ "step": 3942,
151
+ "total_memory_available (GB)": 94.62
152
+ },
153
+ {
154
+ "epoch": 6.0882800608828,
155
+ "grad_norm": 0.7686742544174194,
156
+ "learning_rate": 0.0007970573313039067,
157
+ "loss": 1.6381,
158
+ "max_memory_allocated (GB)": 60.21,
159
+ "memory_allocated (GB)": 3.2,
160
+ "step": 4000,
161
+ "total_memory_available (GB)": 94.62
162
+ },
163
+ {
164
+ "epoch": 6.8493150684931505,
165
+ "grad_norm": 0.7999989986419678,
166
+ "learning_rate": 0.000771689497716895,
167
+ "loss": 1.5135,
168
+ "max_memory_allocated (GB)": 60.21,
169
+ "memory_allocated (GB)": 3.2,
170
+ "step": 4500,
171
+ "total_memory_available (GB)": 94.62
172
+ },
173
+ {
174
+ "epoch": 7.0,
175
+ "eval_accuracy": 0.14675722704248328,
176
+ "eval_loss": 5.426120758056641,
177
+ "eval_runtime": 1131.9083,
178
+ "eval_samples_per_second": 593.526,
179
+ "eval_steps_per_second": 0.58,
180
+ "max_memory_allocated (GB)": 60.21,
181
+ "memory_allocated (GB)": 3.2,
182
+ "step": 4599,
183
+ "total_memory_available (GB)": 94.62
184
+ },
185
+ {
186
+ "epoch": 7.610350076103501,
187
+ "grad_norm": 0.8491269946098328,
188
+ "learning_rate": 0.0007463216641298833,
189
+ "loss": 1.4225,
190
+ "max_memory_allocated (GB)": 60.21,
191
+ "memory_allocated (GB)": 3.2,
192
+ "step": 5000,
193
+ "total_memory_available (GB)": 94.62
194
+ },
195
+ {
196
+ "epoch": 8.0,
197
+ "eval_accuracy": 0.15733153522462218,
198
+ "eval_loss": 5.333346843719482,
199
+ "eval_runtime": 1130.7533,
200
+ "eval_samples_per_second": 594.132,
201
+ "eval_steps_per_second": 0.581,
202
+ "max_memory_allocated (GB)": 60.21,
203
+ "memory_allocated (GB)": 3.2,
204
+ "step": 5256,
205
+ "total_memory_available (GB)": 94.62
206
+ },
207
+ {
208
+ "epoch": 8.37138508371385,
209
+ "grad_norm": 0.7948514819145203,
210
+ "learning_rate": 0.0007209538305428717,
211
+ "loss": 1.354,
212
+ "max_memory_allocated (GB)": 60.21,
213
+ "memory_allocated (GB)": 3.2,
214
+ "step": 5500,
215
+ "total_memory_available (GB)": 94.62
216
+ },
217
+ {
218
+ "epoch": 9.0,
219
+ "eval_accuracy": 0.16383479429666115,
220
+ "eval_loss": 5.220494747161865,
221
+ "eval_runtime": 1131.3928,
222
+ "eval_samples_per_second": 593.796,
223
+ "eval_steps_per_second": 0.581,
224
+ "max_memory_allocated (GB)": 60.21,
225
+ "memory_allocated (GB)": 3.2,
226
+ "step": 5913,
227
+ "total_memory_available (GB)": 94.62
228
+ },
229
+ {
230
+ "epoch": 9.132420091324201,
231
+ "grad_norm": 0.7878388166427612,
232
+ "learning_rate": 0.00069558599695586,
233
+ "loss": 1.3172,
234
+ "max_memory_allocated (GB)": 60.21,
235
+ "memory_allocated (GB)": 3.2,
236
+ "step": 6000,
237
+ "total_memory_available (GB)": 94.62
238
+ },
239
+ {
240
+ "epoch": 9.89345509893455,
241
+ "grad_norm": 0.7506768703460693,
242
+ "learning_rate": 0.0006702181633688484,
243
+ "loss": 1.2511,
244
+ "max_memory_allocated (GB)": 60.21,
245
+ "memory_allocated (GB)": 3.2,
246
+ "step": 6500,
247
+ "total_memory_available (GB)": 94.62
248
+ },
249
+ {
250
+ "epoch": 10.0,
251
+ "eval_accuracy": 0.17084414356885877,
252
+ "eval_loss": 5.212928295135498,
253
+ "eval_runtime": 1133.5622,
254
+ "eval_samples_per_second": 592.66,
255
+ "eval_steps_per_second": 0.58,
256
+ "max_memory_allocated (GB)": 60.21,
257
+ "memory_allocated (GB)": 3.2,
258
+ "step": 6570,
259
+ "total_memory_available (GB)": 94.62
260
+ },
261
+ {
262
+ "epoch": 10.654490106544902,
263
+ "grad_norm": 0.7633622288703918,
264
+ "learning_rate": 0.0006448503297818367,
265
+ "loss": 1.1742,
266
+ "max_memory_allocated (GB)": 60.21,
267
+ "memory_allocated (GB)": 3.2,
268
+ "step": 7000,
269
+ "total_memory_available (GB)": 94.62
270
+ },
271
+ {
272
+ "epoch": 11.0,
273
+ "eval_accuracy": 0.17239664968287494,
274
+ "eval_loss": 5.200212001800537,
275
+ "eval_runtime": 1129.7037,
276
+ "eval_samples_per_second": 594.684,
277
+ "eval_steps_per_second": 0.582,
278
+ "max_memory_allocated (GB)": 60.21,
279
+ "memory_allocated (GB)": 3.2,
280
+ "step": 7227,
281
+ "total_memory_available (GB)": 94.62
282
+ },
283
+ {
284
+ "epoch": 11.415525114155251,
285
+ "grad_norm": 0.7618717551231384,
286
+ "learning_rate": 0.000619482496194825,
287
+ "loss": 1.1342,
288
+ "max_memory_allocated (GB)": 60.21,
289
+ "memory_allocated (GB)": 3.2,
290
+ "step": 7500,
291
+ "total_memory_available (GB)": 94.62
292
+ },
293
+ {
294
+ "epoch": 12.0,
295
+ "eval_accuracy": 0.17819584797645788,
296
+ "eval_loss": 5.163547039031982,
297
+ "eval_runtime": 1128.6949,
298
+ "eval_samples_per_second": 595.216,
299
+ "eval_steps_per_second": 0.582,
300
+ "max_memory_allocated (GB)": 60.21,
301
+ "memory_allocated (GB)": 3.2,
302
+ "step": 7884,
303
+ "total_memory_available (GB)": 94.62
304
+ },
305
+ {
306
+ "epoch": 12.1765601217656,
307
+ "grad_norm": 0.7090550661087036,
308
+ "learning_rate": 0.0005941146626078133,
309
+ "loss": 1.1111,
310
+ "max_memory_allocated (GB)": 60.21,
311
+ "memory_allocated (GB)": 3.2,
312
+ "step": 8000,
313
+ "total_memory_available (GB)": 94.62
314
+ },
315
+ {
316
+ "epoch": 12.937595129375952,
317
+ "grad_norm": 0.7710525393486023,
318
+ "learning_rate": 0.0005687468290208016,
319
+ "loss": 1.0711,
320
+ "max_memory_allocated (GB)": 60.21,
321
+ "memory_allocated (GB)": 3.2,
322
+ "step": 8500,
323
+ "total_memory_available (GB)": 94.62
324
+ },
325
+ {
326
+ "epoch": 13.0,
327
+ "eval_accuracy": 0.17787879735106435,
328
+ "eval_loss": 5.143550872802734,
329
+ "eval_runtime": 1133.3031,
330
+ "eval_samples_per_second": 592.796,
331
+ "eval_steps_per_second": 0.58,
332
+ "max_memory_allocated (GB)": 60.21,
333
+ "memory_allocated (GB)": 3.2,
334
+ "step": 8541,
335
+ "total_memory_available (GB)": 94.62
336
+ },
337
+ {
338
+ "epoch": 13.698630136986301,
339
+ "grad_norm": 0.7961007952690125,
340
+ "learning_rate": 0.00054337899543379,
341
+ "loss": 0.9971,
342
+ "max_memory_allocated (GB)": 60.21,
343
+ "memory_allocated (GB)": 3.2,
344
+ "step": 9000,
345
+ "total_memory_available (GB)": 94.62
346
+ },
347
+ {
348
+ "epoch": 14.0,
349
+ "eval_accuracy": 0.18167893935402052,
350
+ "eval_loss": 5.107571125030518,
351
+ "eval_runtime": 1132.5925,
352
+ "eval_samples_per_second": 593.167,
353
+ "eval_steps_per_second": 0.58,
354
+ "max_memory_allocated (GB)": 60.21,
355
+ "memory_allocated (GB)": 3.2,
356
+ "step": 9198,
357
+ "total_memory_available (GB)": 94.62
358
+ },
359
+ {
360
+ "epoch": 14.459665144596652,
361
+ "grad_norm": 0.7081454992294312,
362
+ "learning_rate": 0.0005180111618467784,
363
+ "loss": 0.9774,
364
+ "max_memory_allocated (GB)": 60.21,
365
+ "memory_allocated (GB)": 3.2,
366
+ "step": 9500,
367
+ "total_memory_available (GB)": 94.62
368
+ },
369
+ {
370
+ "epoch": 15.0,
371
+ "eval_accuracy": 0.19349168002595946,
372
+ "eval_loss": 4.9076433181762695,
373
+ "eval_runtime": 1135.4783,
374
+ "eval_samples_per_second": 591.66,
375
+ "eval_steps_per_second": 0.579,
376
+ "max_memory_allocated (GB)": 60.21,
377
+ "memory_allocated (GB)": 3.2,
378
+ "step": 9855,
379
+ "total_memory_available (GB)": 94.62
380
+ },
381
+ {
382
+ "epoch": 15.220700152207002,
383
+ "grad_norm": 0.706643283367157,
384
+ "learning_rate": 0.0004926433282597666,
385
+ "loss": 0.9457,
386
+ "max_memory_allocated (GB)": 60.21,
387
+ "memory_allocated (GB)": 3.2,
388
+ "step": 10000,
389
+ "total_memory_available (GB)": 94.62
390
+ },
391
+ {
392
+ "epoch": 15.981735159817351,
393
+ "grad_norm": 0.7469919323921204,
394
+ "learning_rate": 0.0004672754946727549,
395
+ "loss": 0.9174,
396
+ "max_memory_allocated (GB)": 60.21,
397
+ "memory_allocated (GB)": 3.2,
398
+ "step": 10500,
399
+ "total_memory_available (GB)": 94.62
400
+ },
401
+ {
402
+ "epoch": 16.0,
403
+ "eval_accuracy": 0.18904255176632923,
404
+ "eval_loss": 5.03179407119751,
405
+ "eval_runtime": 1127.5619,
406
+ "eval_samples_per_second": 595.814,
407
+ "eval_steps_per_second": 0.583,
408
+ "max_memory_allocated (GB)": 60.21,
409
+ "memory_allocated (GB)": 3.2,
410
+ "step": 10512,
411
+ "total_memory_available (GB)": 94.62
412
+ },
413
+ {
414
+ "epoch": 16.7427701674277,
415
+ "grad_norm": 0.7092038989067078,
416
+ "learning_rate": 0.0004419076610857433,
417
+ "loss": 0.8675,
418
+ "max_memory_allocated (GB)": 60.21,
419
+ "memory_allocated (GB)": 3.2,
420
+ "step": 11000,
421
+ "total_memory_available (GB)": 94.62
422
+ },
423
+ {
424
+ "epoch": 17.0,
425
+ "eval_accuracy": 0.19512456517176552,
426
+ "eval_loss": 5.039154052734375,
427
+ "eval_runtime": 1129.1255,
428
+ "eval_samples_per_second": 594.989,
429
+ "eval_steps_per_second": 0.582,
430
+ "max_memory_allocated (GB)": 60.21,
431
+ "memory_allocated (GB)": 3.2,
432
+ "step": 11169,
433
+ "total_memory_available (GB)": 94.62
434
+ },
435
+ {
436
+ "epoch": 17.503805175038053,
437
+ "grad_norm": 0.722985029220581,
438
+ "learning_rate": 0.0004165398274987316,
439
+ "loss": 0.8499,
440
+ "max_memory_allocated (GB)": 60.21,
441
+ "memory_allocated (GB)": 3.2,
442
+ "step": 11500,
443
+ "total_memory_available (GB)": 94.62
444
+ },
445
+ {
446
+ "epoch": 18.0,
447
+ "eval_accuracy": 0.19776069971435672,
448
+ "eval_loss": 5.024279594421387,
449
+ "eval_runtime": 1134.2545,
450
+ "eval_samples_per_second": 592.298,
451
+ "eval_steps_per_second": 0.579,
452
+ "max_memory_allocated (GB)": 60.21,
453
+ "memory_allocated (GB)": 3.2,
454
+ "step": 11826,
455
+ "total_memory_available (GB)": 94.62
456
+ },
457
+ {
458
+ "epoch": 18.264840182648403,
459
+ "grad_norm": 0.703146755695343,
460
+ "learning_rate": 0.0003911719939117199,
461
+ "loss": 0.8262,
462
+ "max_memory_allocated (GB)": 60.21,
463
+ "memory_allocated (GB)": 3.2,
464
+ "step": 12000,
465
+ "total_memory_available (GB)": 94.62
466
+ },
467
+ {
468
+ "epoch": 19.0,
469
+ "eval_accuracy": 0.19716678797946466,
470
+ "eval_loss": 5.084349632263184,
471
+ "eval_runtime": 1134.1402,
472
+ "eval_samples_per_second": 592.358,
473
+ "eval_steps_per_second": 0.579,
474
+ "max_memory_allocated (GB)": 60.21,
475
+ "memory_allocated (GB)": 3.2,
476
+ "step": 12483,
477
+ "total_memory_available (GB)": 94.62
478
+ },
479
+ {
480
+ "epoch": 19.025875190258752,
481
+ "grad_norm": 0.6394225358963013,
482
+ "learning_rate": 0.00036580416032470827,
483
+ "loss": 0.8039,
484
+ "max_memory_allocated (GB)": 60.21,
485
+ "memory_allocated (GB)": 3.2,
486
+ "step": 12500,
487
+ "total_memory_available (GB)": 94.62
488
+ },
489
+ {
490
+ "epoch": 19.7869101978691,
491
+ "grad_norm": 0.6850036978721619,
492
+ "learning_rate": 0.0003404363267376966,
493
+ "loss": 0.7623,
494
+ "max_memory_allocated (GB)": 60.21,
495
+ "memory_allocated (GB)": 3.2,
496
+ "step": 13000,
497
+ "total_memory_available (GB)": 94.62
498
+ },
499
+ {
500
+ "epoch": 20.0,
501
+ "eval_accuracy": 0.20482214650715894,
502
+ "eval_loss": 5.0004353523254395,
503
+ "eval_runtime": 1132.7333,
504
+ "eval_samples_per_second": 593.094,
505
+ "eval_steps_per_second": 0.58,
506
+ "max_memory_allocated (GB)": 60.21,
507
+ "memory_allocated (GB)": 3.2,
508
+ "step": 13140,
509
+ "total_memory_available (GB)": 94.62
510
+ },
511
+ {
512
+ "epoch": 20.54794520547945,
513
+ "grad_norm": 0.7249587178230286,
514
+ "learning_rate": 0.00031506849315068495,
515
+ "loss": 0.7481,
516
+ "max_memory_allocated (GB)": 60.21,
517
+ "memory_allocated (GB)": 3.2,
518
+ "step": 13500,
519
+ "total_memory_available (GB)": 94.62
520
+ },
521
+ {
522
+ "epoch": 21.0,
523
+ "eval_accuracy": 0.21318007731272057,
524
+ "eval_loss": 4.842759609222412,
525
+ "eval_runtime": 1132.229,
526
+ "eval_samples_per_second": 593.358,
527
+ "eval_steps_per_second": 0.58,
528
+ "max_memory_allocated (GB)": 60.24,
529
+ "memory_allocated (GB)": 3.2,
530
+ "step": 13797,
531
+ "total_memory_available (GB)": 94.62
532
+ },
533
+ {
534
+ "epoch": 21.308980213089804,
535
+ "grad_norm": 0.651644766330719,
536
+ "learning_rate": 0.00028970065956367326,
537
+ "loss": 0.7284,
538
+ "max_memory_allocated (GB)": 60.24,
539
+ "memory_allocated (GB)": 3.2,
540
+ "step": 14000,
541
+ "total_memory_available (GB)": 94.62
542
+ },
543
+ {
544
+ "epoch": 22.0,
545
+ "eval_accuracy": 0.2148576174761877,
546
+ "eval_loss": 4.846081733703613,
547
+ "eval_runtime": 1134.6589,
548
+ "eval_samples_per_second": 592.087,
549
+ "eval_steps_per_second": 0.579,
550
+ "max_memory_allocated (GB)": 60.24,
551
+ "memory_allocated (GB)": 3.2,
552
+ "step": 14454,
553
+ "total_memory_available (GB)": 94.62
554
+ },
555
+ {
556
+ "epoch": 22.070015220700153,
557
+ "grad_norm": 0.6403504610061646,
558
+ "learning_rate": 0.00026433282597666157,
559
+ "loss": 0.706,
560
+ "max_memory_allocated (GB)": 60.24,
561
+ "memory_allocated (GB)": 3.2,
562
+ "step": 14500,
563
+ "total_memory_available (GB)": 94.62
564
+ },
565
+ {
566
+ "epoch": 22.831050228310502,
567
+ "grad_norm": 0.6770262718200684,
568
+ "learning_rate": 0.0002389649923896499,
569
+ "loss": 0.6834,
570
+ "max_memory_allocated (GB)": 60.24,
571
+ "memory_allocated (GB)": 3.2,
572
+ "step": 15000,
573
+ "total_memory_available (GB)": 94.62
574
+ },
575
+ {
576
+ "epoch": 23.0,
577
+ "eval_accuracy": 0.2159159413947548,
578
+ "eval_loss": 4.8741374015808105,
579
+ "eval_runtime": 1130.2975,
580
+ "eval_samples_per_second": 594.372,
581
+ "eval_steps_per_second": 0.581,
582
+ "max_memory_allocated (GB)": 60.24,
583
+ "memory_allocated (GB)": 3.2,
584
+ "step": 15111,
585
+ "total_memory_available (GB)": 94.62
586
+ },
587
+ {
588
+ "epoch": 23.59208523592085,
589
+ "grad_norm": 0.6229885816574097,
590
+ "learning_rate": 0.00021359715880263824,
591
+ "loss": 0.6591,
592
+ "max_memory_allocated (GB)": 60.24,
593
+ "memory_allocated (GB)": 3.2,
594
+ "step": 15500,
595
+ "total_memory_available (GB)": 94.62
596
+ },
597
+ {
598
+ "epoch": 24.0,
599
+ "eval_accuracy": 0.2186681789832648,
600
+ "eval_loss": 4.89931058883667,
601
+ "eval_runtime": 1133.9373,
602
+ "eval_samples_per_second": 592.464,
603
+ "eval_steps_per_second": 0.579,
604
+ "max_memory_allocated (GB)": 60.24,
605
+ "memory_allocated (GB)": 3.2,
606
+ "step": 15768,
607
+ "total_memory_available (GB)": 94.62
608
+ },
609
+ {
610
+ "epoch": 24.3531202435312,
611
+ "grad_norm": 0.6464186310768127,
612
+ "learning_rate": 0.00018822932521562658,
613
+ "loss": 0.6447,
614
+ "max_memory_allocated (GB)": 60.24,
615
+ "memory_allocated (GB)": 3.2,
616
+ "step": 16000,
617
+ "total_memory_available (GB)": 94.62
618
+ },
619
+ {
620
+ "epoch": 25.0,
621
+ "eval_accuracy": 0.21962528486180016,
622
+ "eval_loss": 4.8415398597717285,
623
+ "eval_runtime": 1126.1392,
624
+ "eval_samples_per_second": 596.567,
625
+ "eval_steps_per_second": 0.583,
626
+ "max_memory_allocated (GB)": 60.24,
627
+ "memory_allocated (GB)": 3.2,
628
+ "step": 16425,
629
+ "total_memory_available (GB)": 94.62
630
+ },
631
+ {
632
+ "epoch": 25.114155251141554,
633
+ "grad_norm": 0.695124626159668,
634
+ "learning_rate": 0.00016286149162861492,
635
+ "loss": 0.6323,
636
+ "max_memory_allocated (GB)": 60.24,
637
+ "memory_allocated (GB)": 3.2,
638
+ "step": 16500,
639
+ "total_memory_available (GB)": 94.62
640
+ },
641
+ {
642
+ "epoch": 25.875190258751903,
643
+ "grad_norm": 0.7219062447547913,
644
+ "learning_rate": 0.00013749365804160323,
645
+ "loss": 0.6107,
646
+ "max_memory_allocated (GB)": 60.24,
647
+ "memory_allocated (GB)": 3.2,
648
+ "step": 17000,
649
+ "total_memory_available (GB)": 94.62
650
+ },
651
+ {
652
+ "epoch": 26.0,
653
+ "eval_accuracy": 0.22164369166008005,
654
+ "eval_loss": 4.859982967376709,
655
+ "eval_runtime": 1131.8158,
656
+ "eval_samples_per_second": 593.574,
657
+ "eval_steps_per_second": 0.58,
658
+ "max_memory_allocated (GB)": 60.24,
659
+ "memory_allocated (GB)": 3.2,
660
+ "step": 17082,
661
+ "total_memory_available (GB)": 94.62
662
+ },
663
+ {
664
+ "epoch": 26.636225266362253,
665
+ "grad_norm": 0.6680580377578735,
666
+ "learning_rate": 0.00011212582445459158,
667
+ "loss": 0.5958,
668
+ "max_memory_allocated (GB)": 60.24,
669
+ "memory_allocated (GB)": 3.2,
670
+ "step": 17500,
671
+ "total_memory_available (GB)": 94.62
672
+ },
673
+ {
674
+ "epoch": 27.0,
675
+ "eval_accuracy": 0.22447184277861382,
676
+ "eval_loss": 4.839137554168701,
677
+ "eval_runtime": 1135.3766,
678
+ "eval_samples_per_second": 591.713,
679
+ "eval_steps_per_second": 0.579,
680
+ "max_memory_allocated (GB)": 60.24,
681
+ "memory_allocated (GB)": 3.2,
682
+ "step": 17739,
683
+ "total_memory_available (GB)": 94.62
684
+ },
685
+ {
686
+ "epoch": 27.397260273972602,
687
+ "grad_norm": 0.6511676907539368,
688
+ "learning_rate": 8.67579908675799e-05,
689
+ "loss": 0.5836,
690
+ "max_memory_allocated (GB)": 60.24,
691
+ "memory_allocated (GB)": 3.2,
692
+ "step": 18000,
693
+ "total_memory_available (GB)": 94.62
694
+ },
695
+ {
696
+ "epoch": 28.0,
697
+ "eval_accuracy": 0.22654234709749826,
698
+ "eval_loss": 4.856111526489258,
699
+ "eval_runtime": 1131.6209,
700
+ "eval_samples_per_second": 593.677,
701
+ "eval_steps_per_second": 0.581,
702
+ "max_memory_allocated (GB)": 60.24,
703
+ "memory_allocated (GB)": 3.2,
704
+ "step": 18396,
705
+ "total_memory_available (GB)": 94.62
706
+ },
707
+ {
708
+ "epoch": 28.15829528158295,
709
+ "grad_norm": 0.6694862842559814,
710
+ "learning_rate": 6.139015728056824e-05,
711
+ "loss": 0.5713,
712
+ "max_memory_allocated (GB)": 60.24,
713
+ "memory_allocated (GB)": 3.2,
714
+ "step": 18500,
715
+ "total_memory_available (GB)": 94.62
716
+ },
717
+ {
718
+ "epoch": 28.919330289193304,
719
+ "grad_norm": 0.6698545813560486,
720
+ "learning_rate": 3.6022323693556566e-05,
721
+ "loss": 0.5547,
722
+ "max_memory_allocated (GB)": 60.24,
723
+ "memory_allocated (GB)": 3.2,
724
+ "step": 19000,
725
+ "total_memory_available (GB)": 94.62
726
+ },
727
+ {
728
+ "epoch": 29.0,
729
+ "eval_accuracy": 0.2294940437648943,
730
+ "eval_loss": 4.793288230895996,
731
+ "eval_runtime": 1134.2517,
732
+ "eval_samples_per_second": 592.3,
733
+ "eval_steps_per_second": 0.579,
734
+ "max_memory_allocated (GB)": 60.24,
735
+ "memory_allocated (GB)": 3.2,
736
+ "step": 19053,
737
+ "total_memory_available (GB)": 94.62
738
+ },
739
+ {
740
+ "epoch": 29.680365296803654,
741
+ "grad_norm": 0.5940834879875183,
742
+ "learning_rate": 1.06544901065449e-05,
743
+ "loss": 0.547,
744
+ "max_memory_allocated (GB)": 60.24,
745
+ "memory_allocated (GB)": 3.2,
746
+ "step": 19500,
747
+ "total_memory_available (GB)": 94.62
748
+ },
749
+ {
750
+ "epoch": 30.0,
751
+ "eval_accuracy": 0.22931691219483877,
752
+ "eval_loss": 4.809002876281738,
753
+ "eval_runtime": 1130.6604,
754
+ "eval_samples_per_second": 594.181,
755
+ "eval_steps_per_second": 0.581,
756
+ "max_memory_allocated (GB)": 60.24,
757
+ "memory_allocated (GB)": 3.2,
758
+ "step": 19710,
759
+ "total_memory_available (GB)": 94.62
760
+ },
761
+ {
762
+ "epoch": 30.0,
763
+ "max_memory_allocated (GB)": 1.42,
764
+ "memory_allocated (GB)": 1.42,
765
+ "step": 19710,
766
+ "total_flos": 2.9333313524800244e+21,
767
+ "total_memory_available (GB)": 94.62,
768
+ "train_loss": 0.0,
769
+ "train_runtime": 0.2168,
770
+ "train_samples_per_second": 92964947.027,
771
+ "train_steps_per_second": 90914.595
772
+ }
773
+ ],
774
+ "logging_steps": 500,
775
+ "max_steps": 19710,
776
+ "num_input_tokens_seen": 0,
777
+ "num_train_epochs": 30,
778
+ "save_steps": 500,
779
+ "stateful_callbacks": {
780
+ "TrainerControl": {
781
+ "args": {
782
+ "should_epoch_stop": false,
783
+ "should_evaluate": false,
784
+ "should_log": false,
785
+ "should_save": true,
786
+ "should_training_stop": true
787
+ },
788
+ "attributes": {}
789
+ }
790
+ },
791
+ "total_flos": 2.9333313524800244e+21,
792
+ "train_batch_size": 128,
793
+ "trial_name": null,
794
+ "trial_params": null
795
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:18fb3937f0f0009027af236f4b671cbba0784ec4a098d28ebd4f00aa9708249c
3
+ size 6136