End of training
Browse files- all_results.json +7 -0
- train_results.json +7 -0
- trainer_state.json +112 -0
all_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.99,
|
| 3 |
+
"train_loss": 1.6275942337818634,
|
| 4 |
+
"train_runtime": 5933.9925,
|
| 5 |
+
"train_samples_per_second": 33.704,
|
| 6 |
+
"train_steps_per_second": 0.263
|
| 7 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.99,
|
| 3 |
+
"train_loss": 1.6275942337818634,
|
| 4 |
+
"train_runtime": 5933.9925,
|
| 5 |
+
"train_samples_per_second": 33.704,
|
| 6 |
+
"train_steps_per_second": 0.263
|
| 7 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,112 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.9923224568138194,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 1560,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 1.0,
|
| 13 |
+
"eval_f1": 0.8966,
|
| 14 |
+
"eval_gen_len": 19.970909090909092,
|
| 15 |
+
"eval_loss": 1.5708835124969482,
|
| 16 |
+
"eval_precision": 0.9093,
|
| 17 |
+
"eval_recall": 0.8846,
|
| 18 |
+
"eval_rouge1": 0.4119,
|
| 19 |
+
"eval_rouge2": 0.2002,
|
| 20 |
+
"eval_rougeL": 0.3529,
|
| 21 |
+
"eval_rougeLsum": 0.3527,
|
| 22 |
+
"eval_runtime": 231.9368,
|
| 23 |
+
"eval_samples_per_second": 11.857,
|
| 24 |
+
"eval_steps_per_second": 0.742,
|
| 25 |
+
"step": 390
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"epoch": 1.28,
|
| 29 |
+
"learning_rate": 1.3589743589743592e-05,
|
| 30 |
+
"loss": 1.8155,
|
| 31 |
+
"step": 500
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 2.0,
|
| 35 |
+
"eval_f1": 0.9003,
|
| 36 |
+
"eval_gen_len": 19.918545454545455,
|
| 37 |
+
"eval_loss": 1.5360783338546753,
|
| 38 |
+
"eval_precision": 0.9123,
|
| 39 |
+
"eval_recall": 0.8889,
|
| 40 |
+
"eval_rouge1": 0.4331,
|
| 41 |
+
"eval_rouge2": 0.2157,
|
| 42 |
+
"eval_rougeL": 0.3717,
|
| 43 |
+
"eval_rougeLsum": 0.3717,
|
| 44 |
+
"eval_runtime": 228.1491,
|
| 45 |
+
"eval_samples_per_second": 12.054,
|
| 46 |
+
"eval_steps_per_second": 0.754,
|
| 47 |
+
"step": 781
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"epoch": 2.56,
|
| 51 |
+
"learning_rate": 7.17948717948718e-06,
|
| 52 |
+
"loss": 1.5875,
|
| 53 |
+
"step": 1000
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 3.0,
|
| 57 |
+
"eval_f1": 0.899,
|
| 58 |
+
"eval_gen_len": 19.954545454545453,
|
| 59 |
+
"eval_loss": 1.5030012130737305,
|
| 60 |
+
"eval_precision": 0.9117,
|
| 61 |
+
"eval_recall": 0.8871,
|
| 62 |
+
"eval_rouge1": 0.4263,
|
| 63 |
+
"eval_rouge2": 0.2129,
|
| 64 |
+
"eval_rougeL": 0.3671,
|
| 65 |
+
"eval_rougeLsum": 0.3673,
|
| 66 |
+
"eval_runtime": 228.8181,
|
| 67 |
+
"eval_samples_per_second": 12.018,
|
| 68 |
+
"eval_steps_per_second": 0.752,
|
| 69 |
+
"step": 1172
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 3.84,
|
| 73 |
+
"learning_rate": 7.692307692307694e-07,
|
| 74 |
+
"loss": 1.4978,
|
| 75 |
+
"step": 1500
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 3.99,
|
| 79 |
+
"eval_f1": 0.9002,
|
| 80 |
+
"eval_gen_len": 19.925454545454546,
|
| 81 |
+
"eval_loss": 1.499870777130127,
|
| 82 |
+
"eval_precision": 0.9125,
|
| 83 |
+
"eval_recall": 0.8885,
|
| 84 |
+
"eval_rouge1": 0.4331,
|
| 85 |
+
"eval_rouge2": 0.2164,
|
| 86 |
+
"eval_rougeL": 0.3724,
|
| 87 |
+
"eval_rougeLsum": 0.3725,
|
| 88 |
+
"eval_runtime": 229.5833,
|
| 89 |
+
"eval_samples_per_second": 11.978,
|
| 90 |
+
"eval_steps_per_second": 0.749,
|
| 91 |
+
"step": 1560
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"epoch": 3.99,
|
| 95 |
+
"step": 1560,
|
| 96 |
+
"total_flos": 2.163117129644114e+17,
|
| 97 |
+
"train_loss": 1.6275942337818634,
|
| 98 |
+
"train_runtime": 5933.9925,
|
| 99 |
+
"train_samples_per_second": 33.704,
|
| 100 |
+
"train_steps_per_second": 0.263
|
| 101 |
+
}
|
| 102 |
+
],
|
| 103 |
+
"logging_steps": 500,
|
| 104 |
+
"max_steps": 1560,
|
| 105 |
+
"num_input_tokens_seen": 0,
|
| 106 |
+
"num_train_epochs": 4,
|
| 107 |
+
"save_steps": 500,
|
| 108 |
+
"total_flos": 2.163117129644114e+17,
|
| 109 |
+
"train_batch_size": 32,
|
| 110 |
+
"trial_name": null,
|
| 111 |
+
"trial_params": null
|
| 112 |
+
}
|