End of training
Browse files- all_results.json +7 -0
- train_results.json +7 -0
- trainer_state.json +1147 -0
all_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 60.0,
|
| 3 |
+
"train_loss": 17.346778074085204,
|
| 4 |
+
"train_runtime": 5573.2397,
|
| 5 |
+
"train_samples_per_second": 26.807,
|
| 6 |
+
"train_steps_per_second": 3.359
|
| 7 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 60.0,
|
| 3 |
+
"train_loss": 17.346778074085204,
|
| 4 |
+
"train_runtime": 5573.2397,
|
| 5 |
+
"train_samples_per_second": 26.807,
|
| 6 |
+
"train_steps_per_second": 3.359
|
| 7 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1147 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 60.0,
|
| 5 |
+
"global_step": 18720,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 1.0,
|
| 12 |
+
"eval_accuracy": 0.516245487364621,
|
| 13 |
+
"eval_loss": 30.45633316040039,
|
| 14 |
+
"eval_runtime": 6.9248,
|
| 15 |
+
"eval_samples_per_second": 40.001,
|
| 16 |
+
"eval_steps_per_second": 5.054,
|
| 17 |
+
"step": 312
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"best_epoch": 0,
|
| 21 |
+
"best_eval_accuracy": 0.516245487364621,
|
| 22 |
+
"epoch": 1.0,
|
| 23 |
+
"step": 312
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 1.6,
|
| 27 |
+
"learning_rate": 0.0009732905982905983,
|
| 28 |
+
"loss": 31.8281,
|
| 29 |
+
"step": 500
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 2.0,
|
| 33 |
+
"eval_accuracy": 0.4729241877256318,
|
| 34 |
+
"eval_loss": 28.268251419067383,
|
| 35 |
+
"eval_runtime": 6.8798,
|
| 36 |
+
"eval_samples_per_second": 40.263,
|
| 37 |
+
"eval_steps_per_second": 5.087,
|
| 38 |
+
"step": 624
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"best_epoch": 0,
|
| 42 |
+
"best_eval_accuracy": 0.516245487364621,
|
| 43 |
+
"epoch": 2.0,
|
| 44 |
+
"step": 624
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 3.0,
|
| 48 |
+
"eval_accuracy": 0.4729241877256318,
|
| 49 |
+
"eval_loss": 22.482925415039062,
|
| 50 |
+
"eval_runtime": 6.8633,
|
| 51 |
+
"eval_samples_per_second": 40.36,
|
| 52 |
+
"eval_steps_per_second": 5.1,
|
| 53 |
+
"step": 936
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"best_epoch": 0,
|
| 57 |
+
"best_eval_accuracy": 0.516245487364621,
|
| 58 |
+
"epoch": 3.0,
|
| 59 |
+
"step": 936
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 3.21,
|
| 63 |
+
"learning_rate": 0.0009465811965811966,
|
| 64 |
+
"loss": 26.6026,
|
| 65 |
+
"step": 1000
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 4.0,
|
| 69 |
+
"eval_accuracy": 0.4729241877256318,
|
| 70 |
+
"eval_loss": 17.250818252563477,
|
| 71 |
+
"eval_runtime": 6.8768,
|
| 72 |
+
"eval_samples_per_second": 40.281,
|
| 73 |
+
"eval_steps_per_second": 5.09,
|
| 74 |
+
"step": 1248
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"best_epoch": 0,
|
| 78 |
+
"best_eval_accuracy": 0.516245487364621,
|
| 79 |
+
"epoch": 4.0,
|
| 80 |
+
"step": 1248
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 4.81,
|
| 84 |
+
"learning_rate": 0.0009198717948717949,
|
| 85 |
+
"loss": 20.7188,
|
| 86 |
+
"step": 1500
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 5.0,
|
| 90 |
+
"eval_accuracy": 0.5270758122743683,
|
| 91 |
+
"eval_loss": 15.695619583129883,
|
| 92 |
+
"eval_runtime": 6.9281,
|
| 93 |
+
"eval_samples_per_second": 39.982,
|
| 94 |
+
"eval_steps_per_second": 5.052,
|
| 95 |
+
"step": 1560
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"best_epoch": 4,
|
| 99 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 100 |
+
"epoch": 5.0,
|
| 101 |
+
"step": 1560
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 6.0,
|
| 105 |
+
"eval_accuracy": 0.4729241877256318,
|
| 106 |
+
"eval_loss": 14.759888648986816,
|
| 107 |
+
"eval_runtime": 6.832,
|
| 108 |
+
"eval_samples_per_second": 40.545,
|
| 109 |
+
"eval_steps_per_second": 5.123,
|
| 110 |
+
"step": 1872
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"best_epoch": 4,
|
| 114 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 115 |
+
"epoch": 6.0,
|
| 116 |
+
"step": 1872
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 6.41,
|
| 120 |
+
"learning_rate": 0.0008931623931623932,
|
| 121 |
+
"loss": 18.7808,
|
| 122 |
+
"step": 2000
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 7.0,
|
| 126 |
+
"eval_accuracy": 0.5270758122743683,
|
| 127 |
+
"eval_loss": 14.433088302612305,
|
| 128 |
+
"eval_runtime": 6.8638,
|
| 129 |
+
"eval_samples_per_second": 40.356,
|
| 130 |
+
"eval_steps_per_second": 5.099,
|
| 131 |
+
"step": 2184
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"best_epoch": 4,
|
| 135 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 136 |
+
"epoch": 7.0,
|
| 137 |
+
"step": 2184
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 8.0,
|
| 141 |
+
"eval_accuracy": 0.5270758122743683,
|
| 142 |
+
"eval_loss": 13.9365873336792,
|
| 143 |
+
"eval_runtime": 6.8545,
|
| 144 |
+
"eval_samples_per_second": 40.411,
|
| 145 |
+
"eval_steps_per_second": 5.106,
|
| 146 |
+
"step": 2496
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"best_epoch": 4,
|
| 150 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 151 |
+
"epoch": 8.0,
|
| 152 |
+
"step": 2496
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 8.01,
|
| 156 |
+
"learning_rate": 0.0008664529914529915,
|
| 157 |
+
"loss": 18.0838,
|
| 158 |
+
"step": 2500
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 9.0,
|
| 162 |
+
"eval_accuracy": 0.4729241877256318,
|
| 163 |
+
"eval_loss": 13.634007453918457,
|
| 164 |
+
"eval_runtime": 6.7394,
|
| 165 |
+
"eval_samples_per_second": 41.102,
|
| 166 |
+
"eval_steps_per_second": 5.193,
|
| 167 |
+
"step": 2808
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"best_epoch": 4,
|
| 171 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 172 |
+
"epoch": 9.0,
|
| 173 |
+
"step": 2808
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 9.62,
|
| 177 |
+
"learning_rate": 0.0008397435897435898,
|
| 178 |
+
"loss": 17.722,
|
| 179 |
+
"step": 3000
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 10.0,
|
| 183 |
+
"eval_accuracy": 0.4729241877256318,
|
| 184 |
+
"eval_loss": 13.437921524047852,
|
| 185 |
+
"eval_runtime": 6.8897,
|
| 186 |
+
"eval_samples_per_second": 40.205,
|
| 187 |
+
"eval_steps_per_second": 5.08,
|
| 188 |
+
"step": 3120
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"best_epoch": 4,
|
| 192 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 193 |
+
"epoch": 10.0,
|
| 194 |
+
"step": 3120
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 11.0,
|
| 198 |
+
"eval_accuracy": 0.4729241877256318,
|
| 199 |
+
"eval_loss": 13.439308166503906,
|
| 200 |
+
"eval_runtime": 6.8739,
|
| 201 |
+
"eval_samples_per_second": 40.297,
|
| 202 |
+
"eval_steps_per_second": 5.092,
|
| 203 |
+
"step": 3432
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"best_epoch": 4,
|
| 207 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 208 |
+
"epoch": 11.0,
|
| 209 |
+
"step": 3432
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 11.22,
|
| 213 |
+
"learning_rate": 0.0008130341880341881,
|
| 214 |
+
"loss": 17.4783,
|
| 215 |
+
"step": 3500
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 12.0,
|
| 219 |
+
"eval_accuracy": 0.4729241877256318,
|
| 220 |
+
"eval_loss": 13.137593269348145,
|
| 221 |
+
"eval_runtime": 6.874,
|
| 222 |
+
"eval_samples_per_second": 40.297,
|
| 223 |
+
"eval_steps_per_second": 5.092,
|
| 224 |
+
"step": 3744
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"best_epoch": 4,
|
| 228 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 229 |
+
"epoch": 12.0,
|
| 230 |
+
"step": 3744
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 12.82,
|
| 234 |
+
"learning_rate": 0.0007863247863247863,
|
| 235 |
+
"loss": 17.2699,
|
| 236 |
+
"step": 4000
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 13.0,
|
| 240 |
+
"eval_accuracy": 0.4729241877256318,
|
| 241 |
+
"eval_loss": 12.95992374420166,
|
| 242 |
+
"eval_runtime": 6.8727,
|
| 243 |
+
"eval_samples_per_second": 40.304,
|
| 244 |
+
"eval_steps_per_second": 5.093,
|
| 245 |
+
"step": 4056
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"best_epoch": 4,
|
| 249 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 250 |
+
"epoch": 13.0,
|
| 251 |
+
"step": 4056
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 14.0,
|
| 255 |
+
"eval_accuracy": 0.4729241877256318,
|
| 256 |
+
"eval_loss": 12.848048210144043,
|
| 257 |
+
"eval_runtime": 6.8528,
|
| 258 |
+
"eval_samples_per_second": 40.421,
|
| 259 |
+
"eval_steps_per_second": 5.107,
|
| 260 |
+
"step": 4368
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"best_epoch": 4,
|
| 264 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 265 |
+
"epoch": 14.0,
|
| 266 |
+
"step": 4368
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 14.42,
|
| 270 |
+
"learning_rate": 0.0007596153846153846,
|
| 271 |
+
"loss": 17.0966,
|
| 272 |
+
"step": 4500
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 15.0,
|
| 276 |
+
"eval_accuracy": 0.4729241877256318,
|
| 277 |
+
"eval_loss": 12.781341552734375,
|
| 278 |
+
"eval_runtime": 6.8525,
|
| 279 |
+
"eval_samples_per_second": 40.423,
|
| 280 |
+
"eval_steps_per_second": 5.108,
|
| 281 |
+
"step": 4680
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"best_epoch": 4,
|
| 285 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 286 |
+
"epoch": 15.0,
|
| 287 |
+
"step": 4680
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 16.0,
|
| 291 |
+
"eval_accuracy": 0.5270758122743683,
|
| 292 |
+
"eval_loss": 12.691984176635742,
|
| 293 |
+
"eval_runtime": 6.9197,
|
| 294 |
+
"eval_samples_per_second": 40.03,
|
| 295 |
+
"eval_steps_per_second": 5.058,
|
| 296 |
+
"step": 4992
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"best_epoch": 4,
|
| 300 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 301 |
+
"epoch": 16.0,
|
| 302 |
+
"step": 4992
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 16.03,
|
| 306 |
+
"learning_rate": 0.0007329059829059829,
|
| 307 |
+
"loss": 16.9613,
|
| 308 |
+
"step": 5000
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 17.0,
|
| 312 |
+
"eval_accuracy": 0.5270758122743683,
|
| 313 |
+
"eval_loss": 12.569376945495605,
|
| 314 |
+
"eval_runtime": 6.837,
|
| 315 |
+
"eval_samples_per_second": 40.515,
|
| 316 |
+
"eval_steps_per_second": 5.119,
|
| 317 |
+
"step": 5304
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"best_epoch": 4,
|
| 321 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 322 |
+
"epoch": 17.0,
|
| 323 |
+
"step": 5304
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 17.63,
|
| 327 |
+
"learning_rate": 0.0007061965811965813,
|
| 328 |
+
"loss": 16.848,
|
| 329 |
+
"step": 5500
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 18.0,
|
| 333 |
+
"eval_accuracy": 0.5270758122743683,
|
| 334 |
+
"eval_loss": 12.51942253112793,
|
| 335 |
+
"eval_runtime": 6.9126,
|
| 336 |
+
"eval_samples_per_second": 40.072,
|
| 337 |
+
"eval_steps_per_second": 5.063,
|
| 338 |
+
"step": 5616
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"best_epoch": 4,
|
| 342 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 343 |
+
"epoch": 18.0,
|
| 344 |
+
"step": 5616
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 19.0,
|
| 348 |
+
"eval_accuracy": 0.4729241877256318,
|
| 349 |
+
"eval_loss": 12.459056854248047,
|
| 350 |
+
"eval_runtime": 6.8379,
|
| 351 |
+
"eval_samples_per_second": 40.509,
|
| 352 |
+
"eval_steps_per_second": 5.119,
|
| 353 |
+
"step": 5928
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"best_epoch": 4,
|
| 357 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 358 |
+
"epoch": 19.0,
|
| 359 |
+
"step": 5928
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 19.23,
|
| 363 |
+
"learning_rate": 0.0006794871794871796,
|
| 364 |
+
"loss": 16.7661,
|
| 365 |
+
"step": 6000
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 20.0,
|
| 369 |
+
"eval_accuracy": 0.5270758122743683,
|
| 370 |
+
"eval_loss": 12.382737159729004,
|
| 371 |
+
"eval_runtime": 6.8814,
|
| 372 |
+
"eval_samples_per_second": 40.253,
|
| 373 |
+
"eval_steps_per_second": 5.086,
|
| 374 |
+
"step": 6240
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"best_epoch": 4,
|
| 378 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 379 |
+
"epoch": 20.0,
|
| 380 |
+
"step": 6240
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 20.83,
|
| 384 |
+
"learning_rate": 0.0006527777777777778,
|
| 385 |
+
"loss": 16.6825,
|
| 386 |
+
"step": 6500
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 21.0,
|
| 390 |
+
"eval_accuracy": 0.4729241877256318,
|
| 391 |
+
"eval_loss": 12.341012001037598,
|
| 392 |
+
"eval_runtime": 6.8719,
|
| 393 |
+
"eval_samples_per_second": 40.309,
|
| 394 |
+
"eval_steps_per_second": 5.093,
|
| 395 |
+
"step": 6552
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"best_epoch": 4,
|
| 399 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 400 |
+
"epoch": 21.0,
|
| 401 |
+
"step": 6552
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 22.0,
|
| 405 |
+
"eval_accuracy": 0.5270758122743683,
|
| 406 |
+
"eval_loss": 12.324116706848145,
|
| 407 |
+
"eval_runtime": 6.8995,
|
| 408 |
+
"eval_samples_per_second": 40.148,
|
| 409 |
+
"eval_steps_per_second": 5.073,
|
| 410 |
+
"step": 6864
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"best_epoch": 4,
|
| 414 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 415 |
+
"epoch": 22.0,
|
| 416 |
+
"step": 6864
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 22.44,
|
| 420 |
+
"learning_rate": 0.0006260683760683761,
|
| 421 |
+
"loss": 16.5963,
|
| 422 |
+
"step": 7000
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 23.0,
|
| 426 |
+
"eval_accuracy": 0.5270758122743683,
|
| 427 |
+
"eval_loss": 12.329614639282227,
|
| 428 |
+
"eval_runtime": 6.8549,
|
| 429 |
+
"eval_samples_per_second": 40.409,
|
| 430 |
+
"eval_steps_per_second": 5.106,
|
| 431 |
+
"step": 7176
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"best_epoch": 4,
|
| 435 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 436 |
+
"epoch": 23.0,
|
| 437 |
+
"step": 7176
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 24.0,
|
| 441 |
+
"eval_accuracy": 0.4729241877256318,
|
| 442 |
+
"eval_loss": 12.261089324951172,
|
| 443 |
+
"eval_runtime": 6.8822,
|
| 444 |
+
"eval_samples_per_second": 40.249,
|
| 445 |
+
"eval_steps_per_second": 5.086,
|
| 446 |
+
"step": 7488
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"best_epoch": 4,
|
| 450 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 451 |
+
"epoch": 24.0,
|
| 452 |
+
"step": 7488
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 24.04,
|
| 456 |
+
"learning_rate": 0.0005993589743589744,
|
| 457 |
+
"loss": 16.5513,
|
| 458 |
+
"step": 7500
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 25.0,
|
| 462 |
+
"eval_accuracy": 0.5270758122743683,
|
| 463 |
+
"eval_loss": 12.151515007019043,
|
| 464 |
+
"eval_runtime": 6.8798,
|
| 465 |
+
"eval_samples_per_second": 40.263,
|
| 466 |
+
"eval_steps_per_second": 5.087,
|
| 467 |
+
"step": 7800
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"best_epoch": 4,
|
| 471 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 472 |
+
"epoch": 25.0,
|
| 473 |
+
"step": 7800
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 25.64,
|
| 477 |
+
"learning_rate": 0.0005726495726495727,
|
| 478 |
+
"loss": 16.4926,
|
| 479 |
+
"step": 8000
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 26.0,
|
| 483 |
+
"eval_accuracy": 0.4729241877256318,
|
| 484 |
+
"eval_loss": 12.119367599487305,
|
| 485 |
+
"eval_runtime": 6.8242,
|
| 486 |
+
"eval_samples_per_second": 40.591,
|
| 487 |
+
"eval_steps_per_second": 5.129,
|
| 488 |
+
"step": 8112
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"best_epoch": 4,
|
| 492 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 493 |
+
"epoch": 26.0,
|
| 494 |
+
"step": 8112
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 27.0,
|
| 498 |
+
"eval_accuracy": 0.4729241877256318,
|
| 499 |
+
"eval_loss": 12.105155944824219,
|
| 500 |
+
"eval_runtime": 6.9189,
|
| 501 |
+
"eval_samples_per_second": 40.035,
|
| 502 |
+
"eval_steps_per_second": 5.059,
|
| 503 |
+
"step": 8424
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"best_epoch": 4,
|
| 507 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 508 |
+
"epoch": 27.0,
|
| 509 |
+
"step": 8424
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 27.24,
|
| 513 |
+
"learning_rate": 0.000545940170940171,
|
| 514 |
+
"loss": 16.4398,
|
| 515 |
+
"step": 8500
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 28.0,
|
| 519 |
+
"eval_accuracy": 0.5270758122743683,
|
| 520 |
+
"eval_loss": 12.051624298095703,
|
| 521 |
+
"eval_runtime": 6.8467,
|
| 522 |
+
"eval_samples_per_second": 40.457,
|
| 523 |
+
"eval_steps_per_second": 5.112,
|
| 524 |
+
"step": 8736
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"best_epoch": 4,
|
| 528 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 529 |
+
"epoch": 28.0,
|
| 530 |
+
"step": 8736
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 28.85,
|
| 534 |
+
"learning_rate": 0.0005192307692307693,
|
| 535 |
+
"loss": 16.399,
|
| 536 |
+
"step": 9000
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 29.0,
|
| 540 |
+
"eval_accuracy": 0.49458483754512633,
|
| 541 |
+
"eval_loss": 12.021004676818848,
|
| 542 |
+
"eval_runtime": 6.9157,
|
| 543 |
+
"eval_samples_per_second": 40.054,
|
| 544 |
+
"eval_steps_per_second": 5.061,
|
| 545 |
+
"step": 9048
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"best_epoch": 4,
|
| 549 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 550 |
+
"epoch": 29.0,
|
| 551 |
+
"step": 9048
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 30.0,
|
| 555 |
+
"eval_accuracy": 0.4729241877256318,
|
| 556 |
+
"eval_loss": 12.005374908447266,
|
| 557 |
+
"eval_runtime": 6.9233,
|
| 558 |
+
"eval_samples_per_second": 40.01,
|
| 559 |
+
"eval_steps_per_second": 5.055,
|
| 560 |
+
"step": 9360
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"best_epoch": 4,
|
| 564 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 565 |
+
"epoch": 30.0,
|
| 566 |
+
"step": 9360
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 30.45,
|
| 570 |
+
"learning_rate": 0.0004925213675213676,
|
| 571 |
+
"loss": 16.3657,
|
| 572 |
+
"step": 9500
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 31.0,
|
| 576 |
+
"eval_accuracy": 0.5270758122743683,
|
| 577 |
+
"eval_loss": 11.996041297912598,
|
| 578 |
+
"eval_runtime": 6.8696,
|
| 579 |
+
"eval_samples_per_second": 40.323,
|
| 580 |
+
"eval_steps_per_second": 5.095,
|
| 581 |
+
"step": 9672
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"best_epoch": 4,
|
| 585 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 586 |
+
"epoch": 31.0,
|
| 587 |
+
"step": 9672
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 32.0,
|
| 591 |
+
"eval_accuracy": 0.5270758122743683,
|
| 592 |
+
"eval_loss": 11.954824447631836,
|
| 593 |
+
"eval_runtime": 6.8336,
|
| 594 |
+
"eval_samples_per_second": 40.535,
|
| 595 |
+
"eval_steps_per_second": 5.122,
|
| 596 |
+
"step": 9984
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"best_epoch": 4,
|
| 600 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 601 |
+
"epoch": 32.0,
|
| 602 |
+
"step": 9984
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 32.05,
|
| 606 |
+
"learning_rate": 0.00046581196581196583,
|
| 607 |
+
"loss": 16.3306,
|
| 608 |
+
"step": 10000
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 33.0,
|
| 612 |
+
"eval_accuracy": 0.5270758122743683,
|
| 613 |
+
"eval_loss": 11.933246612548828,
|
| 614 |
+
"eval_runtime": 6.8702,
|
| 615 |
+
"eval_samples_per_second": 40.319,
|
| 616 |
+
"eval_steps_per_second": 5.094,
|
| 617 |
+
"step": 10296
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"best_epoch": 4,
|
| 621 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 622 |
+
"epoch": 33.0,
|
| 623 |
+
"step": 10296
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 33.65,
|
| 627 |
+
"learning_rate": 0.0004391025641025641,
|
| 628 |
+
"loss": 16.294,
|
| 629 |
+
"step": 10500
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 34.0,
|
| 633 |
+
"eval_accuracy": 0.4729241877256318,
|
| 634 |
+
"eval_loss": 11.914789199829102,
|
| 635 |
+
"eval_runtime": 6.8807,
|
| 636 |
+
"eval_samples_per_second": 40.258,
|
| 637 |
+
"eval_steps_per_second": 5.087,
|
| 638 |
+
"step": 10608
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"best_epoch": 4,
|
| 642 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 643 |
+
"epoch": 34.0,
|
| 644 |
+
"step": 10608
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 35.0,
|
| 648 |
+
"eval_accuracy": 0.4729241877256318,
|
| 649 |
+
"eval_loss": 11.92249870300293,
|
| 650 |
+
"eval_runtime": 6.8751,
|
| 651 |
+
"eval_samples_per_second": 40.29,
|
| 652 |
+
"eval_steps_per_second": 5.091,
|
| 653 |
+
"step": 10920
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"best_epoch": 4,
|
| 657 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 658 |
+
"epoch": 35.0,
|
| 659 |
+
"step": 10920
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 35.26,
|
| 663 |
+
"learning_rate": 0.0004123931623931624,
|
| 664 |
+
"loss": 16.2657,
|
| 665 |
+
"step": 11000
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 36.0,
|
| 669 |
+
"eval_accuracy": 0.47653429602888087,
|
| 670 |
+
"eval_loss": 11.87264633178711,
|
| 671 |
+
"eval_runtime": 6.849,
|
| 672 |
+
"eval_samples_per_second": 40.444,
|
| 673 |
+
"eval_steps_per_second": 5.11,
|
| 674 |
+
"step": 11232
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"best_epoch": 4,
|
| 678 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 679 |
+
"epoch": 36.0,
|
| 680 |
+
"step": 11232
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 36.86,
|
| 684 |
+
"learning_rate": 0.0003856837606837607,
|
| 685 |
+
"loss": 16.2465,
|
| 686 |
+
"step": 11500
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 37.0,
|
| 690 |
+
"eval_accuracy": 0.4729241877256318,
|
| 691 |
+
"eval_loss": 11.845185279846191,
|
| 692 |
+
"eval_runtime": 6.8882,
|
| 693 |
+
"eval_samples_per_second": 40.214,
|
| 694 |
+
"eval_steps_per_second": 5.081,
|
| 695 |
+
"step": 11544
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"best_epoch": 4,
|
| 699 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 700 |
+
"epoch": 37.0,
|
| 701 |
+
"step": 11544
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 38.0,
|
| 705 |
+
"eval_accuracy": 0.5270758122743683,
|
| 706 |
+
"eval_loss": 11.834094047546387,
|
| 707 |
+
"eval_runtime": 6.8409,
|
| 708 |
+
"eval_samples_per_second": 40.492,
|
| 709 |
+
"eval_steps_per_second": 5.116,
|
| 710 |
+
"step": 11856
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"best_epoch": 4,
|
| 714 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 715 |
+
"epoch": 38.0,
|
| 716 |
+
"step": 11856
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 38.46,
|
| 720 |
+
"learning_rate": 0.000358974358974359,
|
| 721 |
+
"loss": 16.208,
|
| 722 |
+
"step": 12000
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 39.0,
|
| 726 |
+
"eval_accuracy": 0.4729241877256318,
|
| 727 |
+
"eval_loss": 11.823224067687988,
|
| 728 |
+
"eval_runtime": 6.842,
|
| 729 |
+
"eval_samples_per_second": 40.485,
|
| 730 |
+
"eval_steps_per_second": 5.115,
|
| 731 |
+
"step": 12168
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"best_epoch": 4,
|
| 735 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 736 |
+
"epoch": 39.0,
|
| 737 |
+
"step": 12168
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 40.0,
|
| 741 |
+
"eval_accuracy": 0.4729241877256318,
|
| 742 |
+
"eval_loss": 11.797883033752441,
|
| 743 |
+
"eval_runtime": 6.8695,
|
| 744 |
+
"eval_samples_per_second": 40.323,
|
| 745 |
+
"eval_steps_per_second": 5.095,
|
| 746 |
+
"step": 12480
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"best_epoch": 4,
|
| 750 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 751 |
+
"epoch": 40.0,
|
| 752 |
+
"step": 12480
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 40.06,
|
| 756 |
+
"learning_rate": 0.00033226495726495727,
|
| 757 |
+
"loss": 16.191,
|
| 758 |
+
"step": 12500
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 41.0,
|
| 762 |
+
"eval_accuracy": 0.4729241877256318,
|
| 763 |
+
"eval_loss": 11.789505004882812,
|
| 764 |
+
"eval_runtime": 6.908,
|
| 765 |
+
"eval_samples_per_second": 40.098,
|
| 766 |
+
"eval_steps_per_second": 5.067,
|
| 767 |
+
"step": 12792
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"best_epoch": 4,
|
| 771 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 772 |
+
"epoch": 41.0,
|
| 773 |
+
"step": 12792
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 41.67,
|
| 777 |
+
"learning_rate": 0.0003055555555555556,
|
| 778 |
+
"loss": 16.1729,
|
| 779 |
+
"step": 13000
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 42.0,
|
| 783 |
+
"eval_accuracy": 0.4729241877256318,
|
| 784 |
+
"eval_loss": 11.839137077331543,
|
| 785 |
+
"eval_runtime": 6.9254,
|
| 786 |
+
"eval_samples_per_second": 39.998,
|
| 787 |
+
"eval_steps_per_second": 5.054,
|
| 788 |
+
"step": 13104
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"best_epoch": 4,
|
| 792 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 793 |
+
"epoch": 42.0,
|
| 794 |
+
"step": 13104
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 43.0,
|
| 798 |
+
"eval_accuracy": 0.5270758122743683,
|
| 799 |
+
"eval_loss": 11.761892318725586,
|
| 800 |
+
"eval_runtime": 3.298,
|
| 801 |
+
"eval_samples_per_second": 83.991,
|
| 802 |
+
"eval_steps_per_second": 10.613,
|
| 803 |
+
"step": 13416
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"best_epoch": 4,
|
| 807 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 808 |
+
"epoch": 43.0,
|
| 809 |
+
"step": 13416
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 43.27,
|
| 813 |
+
"learning_rate": 0.0002788461538461539,
|
| 814 |
+
"loss": 16.1571,
|
| 815 |
+
"step": 13500
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 44.0,
|
| 819 |
+
"eval_accuracy": 0.4729241877256318,
|
| 820 |
+
"eval_loss": 11.750166893005371,
|
| 821 |
+
"eval_runtime": 3.2851,
|
| 822 |
+
"eval_samples_per_second": 84.319,
|
| 823 |
+
"eval_steps_per_second": 10.654,
|
| 824 |
+
"step": 13728
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"best_epoch": 4,
|
| 828 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 829 |
+
"epoch": 44.0,
|
| 830 |
+
"step": 13728
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 44.87,
|
| 834 |
+
"learning_rate": 0.00025213675213675216,
|
| 835 |
+
"loss": 16.1268,
|
| 836 |
+
"step": 14000
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 45.0,
|
| 840 |
+
"eval_accuracy": 0.4729241877256318,
|
| 841 |
+
"eval_loss": 11.751955032348633,
|
| 842 |
+
"eval_runtime": 3.2892,
|
| 843 |
+
"eval_samples_per_second": 84.215,
|
| 844 |
+
"eval_steps_per_second": 10.641,
|
| 845 |
+
"step": 14040
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"best_epoch": 4,
|
| 849 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 850 |
+
"epoch": 45.0,
|
| 851 |
+
"step": 14040
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 46.0,
|
| 855 |
+
"eval_accuracy": 0.4729241877256318,
|
| 856 |
+
"eval_loss": 11.75387954711914,
|
| 857 |
+
"eval_runtime": 3.3027,
|
| 858 |
+
"eval_samples_per_second": 83.87,
|
| 859 |
+
"eval_steps_per_second": 10.597,
|
| 860 |
+
"step": 14352
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"best_epoch": 4,
|
| 864 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 865 |
+
"epoch": 46.0,
|
| 866 |
+
"step": 14352
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 46.47,
|
| 870 |
+
"learning_rate": 0.00022542735042735044,
|
| 871 |
+
"loss": 16.1194,
|
| 872 |
+
"step": 14500
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 47.0,
|
| 876 |
+
"eval_accuracy": 0.4729241877256318,
|
| 877 |
+
"eval_loss": 11.754090309143066,
|
| 878 |
+
"eval_runtime": 3.2928,
|
| 879 |
+
"eval_samples_per_second": 84.122,
|
| 880 |
+
"eval_steps_per_second": 10.629,
|
| 881 |
+
"step": 14664
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"best_epoch": 4,
|
| 885 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 886 |
+
"epoch": 47.0,
|
| 887 |
+
"step": 14664
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 48.0,
|
| 891 |
+
"eval_accuracy": 0.5270758122743683,
|
| 892 |
+
"eval_loss": 11.712992668151855,
|
| 893 |
+
"eval_runtime": 3.2869,
|
| 894 |
+
"eval_samples_per_second": 84.274,
|
| 895 |
+
"eval_steps_per_second": 10.648,
|
| 896 |
+
"step": 14976
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"best_epoch": 4,
|
| 900 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 901 |
+
"epoch": 48.0,
|
| 902 |
+
"step": 14976
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 48.08,
|
| 906 |
+
"learning_rate": 0.0001987179487179487,
|
| 907 |
+
"loss": 16.11,
|
| 908 |
+
"step": 15000
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 49.0,
|
| 912 |
+
"eval_accuracy": 0.5270758122743683,
|
| 913 |
+
"eval_loss": 11.701953887939453,
|
| 914 |
+
"eval_runtime": 3.2907,
|
| 915 |
+
"eval_samples_per_second": 84.176,
|
| 916 |
+
"eval_steps_per_second": 10.636,
|
| 917 |
+
"step": 15288
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"best_epoch": 4,
|
| 921 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 922 |
+
"epoch": 49.0,
|
| 923 |
+
"step": 15288
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 49.68,
|
| 927 |
+
"learning_rate": 0.00017200854700854702,
|
| 928 |
+
"loss": 16.0989,
|
| 929 |
+
"step": 15500
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 50.0,
|
| 933 |
+
"eval_accuracy": 0.4729241877256318,
|
| 934 |
+
"eval_loss": 11.694924354553223,
|
| 935 |
+
"eval_runtime": 3.2959,
|
| 936 |
+
"eval_samples_per_second": 84.044,
|
| 937 |
+
"eval_steps_per_second": 10.619,
|
| 938 |
+
"step": 15600
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"best_epoch": 4,
|
| 942 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 943 |
+
"epoch": 50.0,
|
| 944 |
+
"step": 15600
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 51.0,
|
| 948 |
+
"eval_accuracy": 0.4729241877256318,
|
| 949 |
+
"eval_loss": 11.702590942382812,
|
| 950 |
+
"eval_runtime": 3.3086,
|
| 951 |
+
"eval_samples_per_second": 83.721,
|
| 952 |
+
"eval_steps_per_second": 10.579,
|
| 953 |
+
"step": 15912
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"best_epoch": 4,
|
| 957 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 958 |
+
"epoch": 51.0,
|
| 959 |
+
"step": 15912
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 51.28,
|
| 963 |
+
"learning_rate": 0.00014529914529914532,
|
| 964 |
+
"loss": 16.0802,
|
| 965 |
+
"step": 16000
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 52.0,
|
| 969 |
+
"eval_accuracy": 0.4729241877256318,
|
| 970 |
+
"eval_loss": 11.70562744140625,
|
| 971 |
+
"eval_runtime": 3.3021,
|
| 972 |
+
"eval_samples_per_second": 83.885,
|
| 973 |
+
"eval_steps_per_second": 10.599,
|
| 974 |
+
"step": 16224
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"best_epoch": 4,
|
| 978 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 979 |
+
"epoch": 52.0,
|
| 980 |
+
"step": 16224
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 52.88,
|
| 984 |
+
"learning_rate": 0.00011858974358974358,
|
| 985 |
+
"loss": 16.0765,
|
| 986 |
+
"step": 16500
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 53.0,
|
| 990 |
+
"eval_accuracy": 0.5270758122743683,
|
| 991 |
+
"eval_loss": 11.679302215576172,
|
| 992 |
+
"eval_runtime": 3.3016,
|
| 993 |
+
"eval_samples_per_second": 83.899,
|
| 994 |
+
"eval_steps_per_second": 10.601,
|
| 995 |
+
"step": 16536
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"best_epoch": 4,
|
| 999 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 1000 |
+
"epoch": 53.0,
|
| 1001 |
+
"step": 16536
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 54.0,
|
| 1005 |
+
"eval_accuracy": 0.5270758122743683,
|
| 1006 |
+
"eval_loss": 11.675891876220703,
|
| 1007 |
+
"eval_runtime": 3.3127,
|
| 1008 |
+
"eval_samples_per_second": 83.618,
|
| 1009 |
+
"eval_steps_per_second": 10.566,
|
| 1010 |
+
"step": 16848
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"best_epoch": 4,
|
| 1014 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 1015 |
+
"epoch": 54.0,
|
| 1016 |
+
"step": 16848
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 54.49,
|
| 1020 |
+
"learning_rate": 9.188034188034189e-05,
|
| 1021 |
+
"loss": 16.0629,
|
| 1022 |
+
"step": 17000
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 55.0,
|
| 1026 |
+
"eval_accuracy": 0.4729241877256318,
|
| 1027 |
+
"eval_loss": 11.671220779418945,
|
| 1028 |
+
"eval_runtime": 3.2982,
|
| 1029 |
+
"eval_samples_per_second": 83.984,
|
| 1030 |
+
"eval_steps_per_second": 10.612,
|
| 1031 |
+
"step": 17160
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"best_epoch": 4,
|
| 1035 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 1036 |
+
"epoch": 55.0,
|
| 1037 |
+
"step": 17160
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 56.0,
|
| 1041 |
+
"eval_accuracy": 0.49458483754512633,
|
| 1042 |
+
"eval_loss": 11.665974617004395,
|
| 1043 |
+
"eval_runtime": 3.3072,
|
| 1044 |
+
"eval_samples_per_second": 83.758,
|
| 1045 |
+
"eval_steps_per_second": 10.583,
|
| 1046 |
+
"step": 17472
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"best_epoch": 4,
|
| 1050 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 1051 |
+
"epoch": 56.0,
|
| 1052 |
+
"step": 17472
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 56.09,
|
| 1056 |
+
"learning_rate": 6.517094017094018e-05,
|
| 1057 |
+
"loss": 16.0619,
|
| 1058 |
+
"step": 17500
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 57.0,
|
| 1062 |
+
"eval_accuracy": 0.4729241877256318,
|
| 1063 |
+
"eval_loss": 11.666152000427246,
|
| 1064 |
+
"eval_runtime": 3.2973,
|
| 1065 |
+
"eval_samples_per_second": 84.009,
|
| 1066 |
+
"eval_steps_per_second": 10.615,
|
| 1067 |
+
"step": 17784
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"best_epoch": 4,
|
| 1071 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 1072 |
+
"epoch": 57.0,
|
| 1073 |
+
"step": 17784
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 57.69,
|
| 1077 |
+
"learning_rate": 3.846153846153846e-05,
|
| 1078 |
+
"loss": 16.0566,
|
| 1079 |
+
"step": 18000
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 58.0,
|
| 1083 |
+
"eval_accuracy": 0.4729241877256318,
|
| 1084 |
+
"eval_loss": 11.664301872253418,
|
| 1085 |
+
"eval_runtime": 3.2855,
|
| 1086 |
+
"eval_samples_per_second": 84.311,
|
| 1087 |
+
"eval_steps_per_second": 10.653,
|
| 1088 |
+
"step": 18096
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"best_epoch": 4,
|
| 1092 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 1093 |
+
"epoch": 58.0,
|
| 1094 |
+
"step": 18096
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 59.0,
|
| 1098 |
+
"eval_accuracy": 0.4729241877256318,
|
| 1099 |
+
"eval_loss": 11.661619186401367,
|
| 1100 |
+
"eval_runtime": 3.3182,
|
| 1101 |
+
"eval_samples_per_second": 83.479,
|
| 1102 |
+
"eval_steps_per_second": 10.548,
|
| 1103 |
+
"step": 18408
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"best_epoch": 4,
|
| 1107 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 1108 |
+
"epoch": 59.0,
|
| 1109 |
+
"step": 18408
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 59.29,
|
| 1113 |
+
"learning_rate": 1.1752136752136752e-05,
|
| 1114 |
+
"loss": 16.0547,
|
| 1115 |
+
"step": 18500
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 60.0,
|
| 1119 |
+
"eval_accuracy": 0.4729241877256318,
|
| 1120 |
+
"eval_loss": 11.661417961120605,
|
| 1121 |
+
"eval_runtime": 3.2798,
|
| 1122 |
+
"eval_samples_per_second": 84.456,
|
| 1123 |
+
"eval_steps_per_second": 10.671,
|
| 1124 |
+
"step": 18720
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"best_epoch": 4,
|
| 1128 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 1129 |
+
"epoch": 60.0,
|
| 1130 |
+
"step": 18720
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 60.0,
|
| 1134 |
+
"step": 18720,
|
| 1135 |
+
"total_flos": 6.96152728406016e+16,
|
| 1136 |
+
"train_loss": 17.346778074085204,
|
| 1137 |
+
"train_runtime": 5573.2397,
|
| 1138 |
+
"train_samples_per_second": 26.807,
|
| 1139 |
+
"train_steps_per_second": 3.359
|
| 1140 |
+
}
|
| 1141 |
+
],
|
| 1142 |
+
"max_steps": 18720,
|
| 1143 |
+
"num_train_epochs": 60,
|
| 1144 |
+
"total_flos": 6.96152728406016e+16,
|
| 1145 |
+
"trial_name": null,
|
| 1146 |
+
"trial_params": null
|
| 1147 |
+
}
|