upload
Browse files- model.safetensors +1 -1
- optimizer.pt +1 -1
- rng_state.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +396 -66
- training_args.bin +2 -2
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 903834408
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:71ea3a66cfe45da07a597ca205eba00665f4df04e3824396fd21fdf36798ecd7
|
| 3 |
size 903834408
|
optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1807760570
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c71575ff302ecb138f99e2638444c58668beb1d781cf7ec9eba85f893fefa02f
|
| 3 |
size 1807760570
|
rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:91db2b1c3ce940f3f696e691b60556212689b422bcd4c4cd37a04b7e81bf65c8
|
| 3 |
size 14244
|
scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9ddbab7344f07bcd20249ea0fc4f895615fe9b3fe939a9e7a0035653172adb57
|
| 3 |
size 1064
|
trainer_state.json
CHANGED
|
@@ -1,93 +1,423 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 0.
|
| 3 |
-
"best_model_checkpoint": "./vit5_summary/checkpoint-
|
| 4 |
-
"epoch":
|
| 5 |
-
"eval_steps":
|
| 6 |
-
"global_step":
|
| 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.
|
| 13 |
-
"grad_norm":
|
| 14 |
-
"learning_rate":
|
| 15 |
-
"loss":
|
| 16 |
-
"step":
|
| 17 |
},
|
| 18 |
{
|
| 19 |
-
"epoch": 0.
|
| 20 |
-
"eval_loss": 0.
|
| 21 |
-
"eval_runtime":
|
| 22 |
-
"eval_samples_per_second":
|
| 23 |
-
"eval_steps_per_second":
|
| 24 |
-
"step":
|
| 25 |
},
|
| 26 |
{
|
| 27 |
-
"epoch":
|
| 28 |
-
"grad_norm":
|
| 29 |
-
"learning_rate": 1.
|
| 30 |
-
"loss": 0.
|
| 31 |
-
"step":
|
| 32 |
},
|
| 33 |
{
|
| 34 |
-
"epoch":
|
| 35 |
-
"eval_loss": 0.
|
| 36 |
-
"eval_runtime":
|
| 37 |
-
"eval_samples_per_second":
|
| 38 |
-
"eval_steps_per_second":
|
| 39 |
-
"step":
|
| 40 |
},
|
| 41 |
{
|
| 42 |
-
"epoch":
|
| 43 |
-
"grad_norm":
|
| 44 |
-
"learning_rate":
|
| 45 |
-
"loss": 0.
|
| 46 |
-
"step":
|
| 47 |
},
|
| 48 |
{
|
| 49 |
-
"epoch":
|
| 50 |
-
"eval_loss": 0.
|
| 51 |
-
"eval_runtime":
|
| 52 |
-
"eval_samples_per_second":
|
| 53 |
-
"eval_steps_per_second":
|
| 54 |
-
"step":
|
| 55 |
},
|
| 56 |
{
|
| 57 |
-
"epoch":
|
| 58 |
-
"grad_norm":
|
| 59 |
-
"learning_rate":
|
| 60 |
-
"loss": 0.
|
| 61 |
-
"step":
|
| 62 |
},
|
| 63 |
{
|
| 64 |
-
"epoch":
|
| 65 |
-
"eval_loss": 0.
|
| 66 |
-
"eval_runtime":
|
| 67 |
-
"eval_samples_per_second":
|
| 68 |
-
"eval_steps_per_second":
|
| 69 |
-
"step":
|
| 70 |
},
|
| 71 |
{
|
| 72 |
-
"epoch":
|
| 73 |
-
"grad_norm":
|
| 74 |
-
"learning_rate":
|
| 75 |
-
"loss": 0.
|
| 76 |
-
"step":
|
| 77 |
},
|
| 78 |
{
|
| 79 |
-
"epoch":
|
| 80 |
-
"eval_loss": 0.
|
| 81 |
-
"eval_runtime":
|
| 82 |
-
"eval_samples_per_second":
|
| 83 |
-
"eval_steps_per_second":
|
| 84 |
-
"step":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 85 |
}
|
| 86 |
],
|
| 87 |
-
"logging_steps":
|
| 88 |
-
"max_steps":
|
| 89 |
"num_input_tokens_seen": 0,
|
| 90 |
-
"num_train_epochs":
|
| 91 |
"save_steps": 500,
|
| 92 |
"stateful_callbacks": {
|
| 93 |
"EarlyStoppingCallback": {
|
|
@@ -96,7 +426,7 @@
|
|
| 96 |
"early_stopping_threshold": 0.001
|
| 97 |
},
|
| 98 |
"attributes": {
|
| 99 |
-
"early_stopping_patience_counter":
|
| 100 |
}
|
| 101 |
},
|
| 102 |
"TrainerControl": {
|
|
@@ -110,8 +440,8 @@
|
|
| 110 |
"attributes": {}
|
| 111 |
}
|
| 112 |
},
|
| 113 |
-
"total_flos":
|
| 114 |
-
"train_batch_size":
|
| 115 |
"trial_name": null,
|
| 116 |
"trial_params": null
|
| 117 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.22952628135681152,
|
| 3 |
+
"best_model_checkpoint": "./vit5_summary/checkpoint-2500",
|
| 4 |
+
"epoch": 5.990254466702761,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 2766,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
+
"epoch": 0.2165674066053059,
|
| 13 |
+
"grad_norm": 109060.0546875,
|
| 14 |
+
"learning_rate": 7.220216606498196e-06,
|
| 15 |
+
"loss": 12.7509,
|
| 16 |
+
"step": 100
|
| 17 |
},
|
| 18 |
{
|
| 19 |
+
"epoch": 0.2165674066053059,
|
| 20 |
+
"eval_loss": 0.5880528688430786,
|
| 21 |
+
"eval_runtime": 137.2063,
|
| 22 |
+
"eval_samples_per_second": 23.075,
|
| 23 |
+
"eval_steps_per_second": 1.924,
|
| 24 |
+
"step": 100
|
| 25 |
},
|
| 26 |
{
|
| 27 |
+
"epoch": 0.4331348132106118,
|
| 28 |
+
"grad_norm": 125010.6640625,
|
| 29 |
+
"learning_rate": 1.4440433212996392e-05,
|
| 30 |
+
"loss": 0.5052,
|
| 31 |
+
"step": 200
|
| 32 |
},
|
| 33 |
{
|
| 34 |
+
"epoch": 0.4331348132106118,
|
| 35 |
+
"eval_loss": 0.36430519819259644,
|
| 36 |
+
"eval_runtime": 137.901,
|
| 37 |
+
"eval_samples_per_second": 22.959,
|
| 38 |
+
"eval_steps_per_second": 1.914,
|
| 39 |
+
"step": 200
|
| 40 |
},
|
| 41 |
{
|
| 42 |
+
"epoch": 0.6497022198159177,
|
| 43 |
+
"grad_norm": 65910.375,
|
| 44 |
+
"learning_rate": 1.9815186822016877e-05,
|
| 45 |
+
"loss": 0.3734,
|
| 46 |
+
"step": 300
|
| 47 |
},
|
| 48 |
{
|
| 49 |
+
"epoch": 0.6497022198159177,
|
| 50 |
+
"eval_loss": 0.33203062415122986,
|
| 51 |
+
"eval_runtime": 138.0962,
|
| 52 |
+
"eval_samples_per_second": 22.926,
|
| 53 |
+
"eval_steps_per_second": 1.912,
|
| 54 |
+
"step": 300
|
| 55 |
},
|
| 56 |
{
|
| 57 |
+
"epoch": 0.8662696264212236,
|
| 58 |
+
"grad_norm": 43025.13671875,
|
| 59 |
+
"learning_rate": 1.9011651265568504e-05,
|
| 60 |
+
"loss": 0.3584,
|
| 61 |
+
"step": 400
|
| 62 |
},
|
| 63 |
{
|
| 64 |
+
"epoch": 0.8662696264212236,
|
| 65 |
+
"eval_loss": 0.3121868371963501,
|
| 66 |
+
"eval_runtime": 142.7377,
|
| 67 |
+
"eval_samples_per_second": 22.181,
|
| 68 |
+
"eval_steps_per_second": 1.85,
|
| 69 |
+
"step": 400
|
| 70 |
},
|
| 71 |
{
|
| 72 |
+
"epoch": 1.0828370330265296,
|
| 73 |
+
"grad_norm": 58838.125,
|
| 74 |
+
"learning_rate": 1.820811570912013e-05,
|
| 75 |
+
"loss": 0.3364,
|
| 76 |
+
"step": 500
|
| 77 |
},
|
| 78 |
{
|
| 79 |
+
"epoch": 1.0828370330265296,
|
| 80 |
+
"eval_loss": 0.2999902665615082,
|
| 81 |
+
"eval_runtime": 138.8144,
|
| 82 |
+
"eval_samples_per_second": 22.807,
|
| 83 |
+
"eval_steps_per_second": 1.902,
|
| 84 |
+
"step": 500
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.2994044396318354,
|
| 88 |
+
"grad_norm": 77278.2890625,
|
| 89 |
+
"learning_rate": 1.7404580152671757e-05,
|
| 90 |
+
"loss": 0.317,
|
| 91 |
+
"step": 600
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"epoch": 1.2994044396318354,
|
| 95 |
+
"eval_loss": 0.2913927137851715,
|
| 96 |
+
"eval_runtime": 138.1948,
|
| 97 |
+
"eval_samples_per_second": 22.91,
|
| 98 |
+
"eval_steps_per_second": 1.91,
|
| 99 |
+
"step": 600
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 1.5159718462371412,
|
| 103 |
+
"grad_norm": 50982.0703125,
|
| 104 |
+
"learning_rate": 1.6601044596223384e-05,
|
| 105 |
+
"loss": 0.3006,
|
| 106 |
+
"step": 700
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"epoch": 1.5159718462371412,
|
| 110 |
+
"eval_loss": 0.2813930809497833,
|
| 111 |
+
"eval_runtime": 137.5997,
|
| 112 |
+
"eval_samples_per_second": 23.009,
|
| 113 |
+
"eval_steps_per_second": 1.919,
|
| 114 |
+
"step": 700
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 1.7325392528424473,
|
| 118 |
+
"grad_norm": 63450.98046875,
|
| 119 |
+
"learning_rate": 1.579750903977501e-05,
|
| 120 |
+
"loss": 0.2931,
|
| 121 |
+
"step": 800
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 1.7325392528424473,
|
| 125 |
+
"eval_loss": 0.2716469168663025,
|
| 126 |
+
"eval_runtime": 137.578,
|
| 127 |
+
"eval_samples_per_second": 23.012,
|
| 128 |
+
"eval_steps_per_second": 1.919,
|
| 129 |
+
"step": 800
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 1.949106659447753,
|
| 133 |
+
"grad_norm": 46679.9140625,
|
| 134 |
+
"learning_rate": 1.4993973483326637e-05,
|
| 135 |
+
"loss": 0.2925,
|
| 136 |
+
"step": 900
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 1.949106659447753,
|
| 140 |
+
"eval_loss": 0.26637548208236694,
|
| 141 |
+
"eval_runtime": 142.275,
|
| 142 |
+
"eval_samples_per_second": 22.253,
|
| 143 |
+
"eval_steps_per_second": 1.856,
|
| 144 |
+
"step": 900
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 2.165674066053059,
|
| 148 |
+
"grad_norm": 51095.6640625,
|
| 149 |
+
"learning_rate": 1.4190437926878266e-05,
|
| 150 |
+
"loss": 0.2752,
|
| 151 |
+
"step": 1000
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 2.165674066053059,
|
| 155 |
+
"eval_loss": 0.2650776505470276,
|
| 156 |
+
"eval_runtime": 144.2721,
|
| 157 |
+
"eval_samples_per_second": 21.945,
|
| 158 |
+
"eval_steps_per_second": 1.83,
|
| 159 |
+
"step": 1000
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 2.382241472658365,
|
| 163 |
+
"grad_norm": 50603.19921875,
|
| 164 |
+
"learning_rate": 1.3386902370429893e-05,
|
| 165 |
+
"loss": 0.2675,
|
| 166 |
+
"step": 1100
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"epoch": 2.382241472658365,
|
| 170 |
+
"eval_loss": 0.2589792311191559,
|
| 171 |
+
"eval_runtime": 137.6329,
|
| 172 |
+
"eval_samples_per_second": 23.003,
|
| 173 |
+
"eval_steps_per_second": 1.918,
|
| 174 |
+
"step": 1100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 2.5988088792636708,
|
| 178 |
+
"grad_norm": 53034.58984375,
|
| 179 |
+
"learning_rate": 1.258336681398152e-05,
|
| 180 |
+
"loss": 0.259,
|
| 181 |
+
"step": 1200
|
| 182 |
+
},
|
| 183 |
+
{
|
| 184 |
+
"epoch": 2.5988088792636708,
|
| 185 |
+
"eval_loss": 0.2544151544570923,
|
| 186 |
+
"eval_runtime": 137.1871,
|
| 187 |
+
"eval_samples_per_second": 23.078,
|
| 188 |
+
"eval_steps_per_second": 1.924,
|
| 189 |
+
"step": 1200
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 2.8153762858689766,
|
| 193 |
+
"grad_norm": 51014.87890625,
|
| 194 |
+
"learning_rate": 1.1779831257533148e-05,
|
| 195 |
+
"loss": 0.2639,
|
| 196 |
+
"step": 1300
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"epoch": 2.8153762858689766,
|
| 200 |
+
"eval_loss": 0.2480245977640152,
|
| 201 |
+
"eval_runtime": 138.2167,
|
| 202 |
+
"eval_samples_per_second": 22.906,
|
| 203 |
+
"eval_steps_per_second": 1.91,
|
| 204 |
+
"step": 1300
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 3.0319436924742824,
|
| 208 |
+
"grad_norm": 52027.0625,
|
| 209 |
+
"learning_rate": 1.0976295701084774e-05,
|
| 210 |
+
"loss": 0.2515,
|
| 211 |
+
"step": 1400
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 3.0319436924742824,
|
| 215 |
+
"eval_loss": 0.24613255262374878,
|
| 216 |
+
"eval_runtime": 138.1819,
|
| 217 |
+
"eval_samples_per_second": 22.912,
|
| 218 |
+
"eval_steps_per_second": 1.911,
|
| 219 |
+
"step": 1400
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 3.2485110990795887,
|
| 223 |
+
"grad_norm": 64106.3203125,
|
| 224 |
+
"learning_rate": 1.01727601446364e-05,
|
| 225 |
+
"loss": 0.2349,
|
| 226 |
+
"step": 1500
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 3.2485110990795887,
|
| 230 |
+
"eval_loss": 0.24559645354747772,
|
| 231 |
+
"eval_runtime": 142.67,
|
| 232 |
+
"eval_samples_per_second": 22.191,
|
| 233 |
+
"eval_steps_per_second": 1.85,
|
| 234 |
+
"step": 1500
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 3.4650785056848945,
|
| 238 |
+
"grad_norm": 53759.48046875,
|
| 239 |
+
"learning_rate": 9.369224588188028e-06,
|
| 240 |
+
"loss": 0.2386,
|
| 241 |
+
"step": 1600
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 3.4650785056848945,
|
| 245 |
+
"eval_loss": 0.24301083385944366,
|
| 246 |
+
"eval_runtime": 138.4069,
|
| 247 |
+
"eval_samples_per_second": 22.875,
|
| 248 |
+
"eval_steps_per_second": 1.907,
|
| 249 |
+
"step": 1600
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 3.6816459122902003,
|
| 253 |
+
"grad_norm": 53480.16015625,
|
| 254 |
+
"learning_rate": 8.565689031739656e-06,
|
| 255 |
+
"loss": 0.2343,
|
| 256 |
+
"step": 1700
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 3.6816459122902003,
|
| 260 |
+
"eval_loss": 0.23925070464611053,
|
| 261 |
+
"eval_runtime": 137.9969,
|
| 262 |
+
"eval_samples_per_second": 22.943,
|
| 263 |
+
"eval_steps_per_second": 1.913,
|
| 264 |
+
"step": 1700
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 3.898213318895506,
|
| 268 |
+
"grad_norm": 58331.94921875,
|
| 269 |
+
"learning_rate": 7.762153475291283e-06,
|
| 270 |
+
"loss": 0.2362,
|
| 271 |
+
"step": 1800
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 3.898213318895506,
|
| 275 |
+
"eval_loss": 0.23644807934761047,
|
| 276 |
+
"eval_runtime": 137.3893,
|
| 277 |
+
"eval_samples_per_second": 23.044,
|
| 278 |
+
"eval_steps_per_second": 1.922,
|
| 279 |
+
"step": 1800
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 4.114780725500812,
|
| 283 |
+
"grad_norm": 53067.78125,
|
| 284 |
+
"learning_rate": 6.958617918842909e-06,
|
| 285 |
+
"loss": 0.2264,
|
| 286 |
+
"step": 1900
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 4.114780725500812,
|
| 290 |
+
"eval_loss": 0.23520290851593018,
|
| 291 |
+
"eval_runtime": 137.4102,
|
| 292 |
+
"eval_samples_per_second": 23.041,
|
| 293 |
+
"eval_steps_per_second": 1.921,
|
| 294 |
+
"step": 1900
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 4.331348132106118,
|
| 298 |
+
"grad_norm": 63341.34765625,
|
| 299 |
+
"learning_rate": 6.155082362394537e-06,
|
| 300 |
+
"loss": 0.2209,
|
| 301 |
+
"step": 2000
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 4.331348132106118,
|
| 305 |
+
"eval_loss": 0.23410087823867798,
|
| 306 |
+
"eval_runtime": 142.6996,
|
| 307 |
+
"eval_samples_per_second": 22.186,
|
| 308 |
+
"eval_steps_per_second": 1.85,
|
| 309 |
+
"step": 2000
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 4.547915538711424,
|
| 313 |
+
"grad_norm": 59952.515625,
|
| 314 |
+
"learning_rate": 5.3515468059461635e-06,
|
| 315 |
+
"loss": 0.2209,
|
| 316 |
+
"step": 2100
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 4.547915538711424,
|
| 320 |
+
"eval_loss": 0.23309889435768127,
|
| 321 |
+
"eval_runtime": 142.5841,
|
| 322 |
+
"eval_samples_per_second": 22.204,
|
| 323 |
+
"eval_steps_per_second": 1.852,
|
| 324 |
+
"step": 2100
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 4.76448294531673,
|
| 328 |
+
"grad_norm": 61383.71484375,
|
| 329 |
+
"learning_rate": 4.548011249497791e-06,
|
| 330 |
+
"loss": 0.2204,
|
| 331 |
+
"step": 2200
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 4.76448294531673,
|
| 335 |
+
"eval_loss": 0.23156361281871796,
|
| 336 |
+
"eval_runtime": 137.9316,
|
| 337 |
+
"eval_samples_per_second": 22.953,
|
| 338 |
+
"eval_steps_per_second": 1.914,
|
| 339 |
+
"step": 2200
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 4.981050351922036,
|
| 343 |
+
"grad_norm": 57406.8125,
|
| 344 |
+
"learning_rate": 3.7444756930494173e-06,
|
| 345 |
+
"loss": 0.2185,
|
| 346 |
+
"step": 2300
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 4.981050351922036,
|
| 350 |
+
"eval_loss": 0.22999364137649536,
|
| 351 |
+
"eval_runtime": 137.9105,
|
| 352 |
+
"eval_samples_per_second": 22.957,
|
| 353 |
+
"eval_steps_per_second": 1.914,
|
| 354 |
+
"step": 2300
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 5.1976177585273415,
|
| 358 |
+
"grad_norm": 58135.14453125,
|
| 359 |
+
"learning_rate": 2.940940136601045e-06,
|
| 360 |
+
"loss": 0.2136,
|
| 361 |
+
"step": 2400
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 5.1976177585273415,
|
| 365 |
+
"eval_loss": 0.23013143241405487,
|
| 366 |
+
"eval_runtime": 137.4929,
|
| 367 |
+
"eval_samples_per_second": 23.027,
|
| 368 |
+
"eval_steps_per_second": 1.92,
|
| 369 |
+
"step": 2400
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 5.414185165132648,
|
| 373 |
+
"grad_norm": 54848.12890625,
|
| 374 |
+
"learning_rate": 2.137404580152672e-06,
|
| 375 |
+
"loss": 0.2102,
|
| 376 |
+
"step": 2500
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 5.414185165132648,
|
| 380 |
+
"eval_loss": 0.22952628135681152,
|
| 381 |
+
"eval_runtime": 142.5207,
|
| 382 |
+
"eval_samples_per_second": 22.214,
|
| 383 |
+
"eval_steps_per_second": 1.852,
|
| 384 |
+
"step": 2500
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 5.630752571737953,
|
| 388 |
+
"grad_norm": 62785.00390625,
|
| 389 |
+
"learning_rate": 1.333869023704299e-06,
|
| 390 |
+
"loss": 0.2141,
|
| 391 |
+
"step": 2600
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 5.630752571737953,
|
| 395 |
+
"eval_loss": 0.2286754995584488,
|
| 396 |
+
"eval_runtime": 142.4263,
|
| 397 |
+
"eval_samples_per_second": 22.229,
|
| 398 |
+
"eval_steps_per_second": 1.854,
|
| 399 |
+
"step": 2600
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 5.8473199783432595,
|
| 403 |
+
"grad_norm": 63349.1015625,
|
| 404 |
+
"learning_rate": 5.303334672559261e-07,
|
| 405 |
+
"loss": 0.2115,
|
| 406 |
+
"step": 2700
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 5.8473199783432595,
|
| 410 |
+
"eval_loss": 0.22872701287269592,
|
| 411 |
+
"eval_runtime": 137.9701,
|
| 412 |
+
"eval_samples_per_second": 22.947,
|
| 413 |
+
"eval_steps_per_second": 1.913,
|
| 414 |
+
"step": 2700
|
| 415 |
}
|
| 416 |
],
|
| 417 |
+
"logging_steps": 100,
|
| 418 |
+
"max_steps": 2766,
|
| 419 |
"num_input_tokens_seen": 0,
|
| 420 |
+
"num_train_epochs": 6,
|
| 421 |
"save_steps": 500,
|
| 422 |
"stateful_callbacks": {
|
| 423 |
"EarlyStoppingCallback": {
|
|
|
|
| 426 |
"early_stopping_threshold": 0.001
|
| 427 |
},
|
| 428 |
"attributes": {
|
| 429 |
+
"early_stopping_patience_counter": 2
|
| 430 |
}
|
| 431 |
},
|
| 432 |
"TrainerControl": {
|
|
|
|
| 440 |
"attributes": {}
|
| 441 |
}
|
| 442 |
},
|
| 443 |
+
"total_flos": 8.084098684551168e+16,
|
| 444 |
+
"train_batch_size": 12,
|
| 445 |
"trial_name": null,
|
| 446 |
"trial_params": null
|
| 447 |
}
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6ca4919d0287634f6c7fd251b5ff3b6a23dd7b3f5954cc55d9f8604243903648
|
| 3 |
+
size 5368
|