End of training
Browse files- all_results.json +7 -0
- train_results.json +7 -0
- trainer_state.json +1033 -0
all_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 60.0,
|
| 3 |
+
"train_loss": 0.4369181796016856,
|
| 4 |
+
"train_runtime": 3671.2013,
|
| 5 |
+
"train_samples_per_second": 40.695,
|
| 6 |
+
"train_steps_per_second": 2.55
|
| 7 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 60.0,
|
| 3 |
+
"train_loss": 0.4369181796016856,
|
| 4 |
+
"train_runtime": 3671.2013,
|
| 5 |
+
"train_samples_per_second": 40.695,
|
| 6 |
+
"train_steps_per_second": 2.55
|
| 7 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1033 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 60.0,
|
| 5 |
+
"global_step": 9360,
|
| 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": 0.4729241877256318,
|
| 13 |
+
"eval_loss": 0.8690470457077026,
|
| 14 |
+
"eval_runtime": 4.1588,
|
| 15 |
+
"eval_samples_per_second": 66.605,
|
| 16 |
+
"eval_steps_per_second": 8.416,
|
| 17 |
+
"step": 156
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"best_epoch": 0,
|
| 21 |
+
"best_eval_accuracy": 0.4729241877256318,
|
| 22 |
+
"epoch": 1.0,
|
| 23 |
+
"step": 156
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 2.0,
|
| 27 |
+
"eval_accuracy": 0.5270758122743683,
|
| 28 |
+
"eval_loss": 0.7261884212493896,
|
| 29 |
+
"eval_runtime": 4.2514,
|
| 30 |
+
"eval_samples_per_second": 65.154,
|
| 31 |
+
"eval_steps_per_second": 8.233,
|
| 32 |
+
"step": 312
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"best_epoch": 1,
|
| 36 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 37 |
+
"epoch": 2.0,
|
| 38 |
+
"step": 312
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 3.0,
|
| 42 |
+
"eval_accuracy": 0.4693140794223827,
|
| 43 |
+
"eval_loss": 0.7646142244338989,
|
| 44 |
+
"eval_runtime": 4.264,
|
| 45 |
+
"eval_samples_per_second": 64.963,
|
| 46 |
+
"eval_steps_per_second": 8.208,
|
| 47 |
+
"step": 468
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"best_epoch": 1,
|
| 51 |
+
"best_eval_accuracy": 0.5270758122743683,
|
| 52 |
+
"epoch": 3.0,
|
| 53 |
+
"step": 468
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 3.21,
|
| 57 |
+
"learning_rate": 0.0028397435897435895,
|
| 58 |
+
"loss": 0.8294,
|
| 59 |
+
"step": 500
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 4.0,
|
| 63 |
+
"eval_accuracy": 0.5884476534296029,
|
| 64 |
+
"eval_loss": 0.7043800354003906,
|
| 65 |
+
"eval_runtime": 4.2676,
|
| 66 |
+
"eval_samples_per_second": 64.907,
|
| 67 |
+
"eval_steps_per_second": 8.201,
|
| 68 |
+
"step": 624
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"best_epoch": 3,
|
| 72 |
+
"best_eval_accuracy": 0.5884476534296029,
|
| 73 |
+
"epoch": 4.0,
|
| 74 |
+
"step": 624
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 5.0,
|
| 78 |
+
"eval_accuracy": 0.5884476534296029,
|
| 79 |
+
"eval_loss": 0.7099267840385437,
|
| 80 |
+
"eval_runtime": 4.2759,
|
| 81 |
+
"eval_samples_per_second": 64.781,
|
| 82 |
+
"eval_steps_per_second": 8.185,
|
| 83 |
+
"step": 780
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"best_epoch": 3,
|
| 87 |
+
"best_eval_accuracy": 0.5884476534296029,
|
| 88 |
+
"epoch": 5.0,
|
| 89 |
+
"step": 780
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 6.0,
|
| 93 |
+
"eval_accuracy": 0.6245487364620939,
|
| 94 |
+
"eval_loss": 0.6449177265167236,
|
| 95 |
+
"eval_runtime": 4.2742,
|
| 96 |
+
"eval_samples_per_second": 64.808,
|
| 97 |
+
"eval_steps_per_second": 8.189,
|
| 98 |
+
"step": 936
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"best_epoch": 5,
|
| 102 |
+
"best_eval_accuracy": 0.6245487364620939,
|
| 103 |
+
"epoch": 6.0,
|
| 104 |
+
"step": 936
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 6.41,
|
| 108 |
+
"learning_rate": 0.00267948717948718,
|
| 109 |
+
"loss": 0.785,
|
| 110 |
+
"step": 1000
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 7.0,
|
| 114 |
+
"eval_accuracy": 0.6245487364620939,
|
| 115 |
+
"eval_loss": 0.7755226492881775,
|
| 116 |
+
"eval_runtime": 4.2724,
|
| 117 |
+
"eval_samples_per_second": 64.834,
|
| 118 |
+
"eval_steps_per_second": 8.192,
|
| 119 |
+
"step": 1092
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"best_epoch": 5,
|
| 123 |
+
"best_eval_accuracy": 0.6245487364620939,
|
| 124 |
+
"epoch": 7.0,
|
| 125 |
+
"step": 1092
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 8.0,
|
| 129 |
+
"eval_accuracy": 0.6606498194945848,
|
| 130 |
+
"eval_loss": 0.6442985534667969,
|
| 131 |
+
"eval_runtime": 4.2721,
|
| 132 |
+
"eval_samples_per_second": 64.84,
|
| 133 |
+
"eval_steps_per_second": 8.193,
|
| 134 |
+
"step": 1248
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"best_epoch": 7,
|
| 138 |
+
"best_eval_accuracy": 0.6606498194945848,
|
| 139 |
+
"epoch": 8.0,
|
| 140 |
+
"step": 1248
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 9.0,
|
| 144 |
+
"eval_accuracy": 0.6859205776173285,
|
| 145 |
+
"eval_loss": 0.6348857879638672,
|
| 146 |
+
"eval_runtime": 4.2715,
|
| 147 |
+
"eval_samples_per_second": 64.849,
|
| 148 |
+
"eval_steps_per_second": 8.194,
|
| 149 |
+
"step": 1404
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"best_epoch": 8,
|
| 153 |
+
"best_eval_accuracy": 0.6859205776173285,
|
| 154 |
+
"epoch": 9.0,
|
| 155 |
+
"step": 1404
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 9.62,
|
| 159 |
+
"learning_rate": 0.0025192307692307693,
|
| 160 |
+
"loss": 0.6665,
|
| 161 |
+
"step": 1500
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 10.0,
|
| 165 |
+
"eval_accuracy": 0.6462093862815884,
|
| 166 |
+
"eval_loss": 0.9543846845626831,
|
| 167 |
+
"eval_runtime": 4.2724,
|
| 168 |
+
"eval_samples_per_second": 64.835,
|
| 169 |
+
"eval_steps_per_second": 8.192,
|
| 170 |
+
"step": 1560
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"best_epoch": 8,
|
| 174 |
+
"best_eval_accuracy": 0.6859205776173285,
|
| 175 |
+
"epoch": 10.0,
|
| 176 |
+
"step": 1560
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 11.0,
|
| 180 |
+
"eval_accuracy": 0.7184115523465704,
|
| 181 |
+
"eval_loss": 0.6008003354072571,
|
| 182 |
+
"eval_runtime": 4.2739,
|
| 183 |
+
"eval_samples_per_second": 64.811,
|
| 184 |
+
"eval_steps_per_second": 8.189,
|
| 185 |
+
"step": 1716
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"best_epoch": 10,
|
| 189 |
+
"best_eval_accuracy": 0.7184115523465704,
|
| 190 |
+
"epoch": 11.0,
|
| 191 |
+
"step": 1716
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 12.0,
|
| 195 |
+
"eval_accuracy": 0.7075812274368231,
|
| 196 |
+
"eval_loss": 0.6502608060836792,
|
| 197 |
+
"eval_runtime": 4.2753,
|
| 198 |
+
"eval_samples_per_second": 64.791,
|
| 199 |
+
"eval_steps_per_second": 8.187,
|
| 200 |
+
"step": 1872
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"best_epoch": 10,
|
| 204 |
+
"best_eval_accuracy": 0.7184115523465704,
|
| 205 |
+
"epoch": 12.0,
|
| 206 |
+
"step": 1872
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 12.82,
|
| 210 |
+
"learning_rate": 0.002358974358974359,
|
| 211 |
+
"loss": 0.6276,
|
| 212 |
+
"step": 2000
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 13.0,
|
| 216 |
+
"eval_accuracy": 0.7075812274368231,
|
| 217 |
+
"eval_loss": 0.626946210861206,
|
| 218 |
+
"eval_runtime": 4.2756,
|
| 219 |
+
"eval_samples_per_second": 64.787,
|
| 220 |
+
"eval_steps_per_second": 8.186,
|
| 221 |
+
"step": 2028
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"best_epoch": 10,
|
| 225 |
+
"best_eval_accuracy": 0.7184115523465704,
|
| 226 |
+
"epoch": 13.0,
|
| 227 |
+
"step": 2028
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 14.0,
|
| 231 |
+
"eval_accuracy": 0.7148014440433214,
|
| 232 |
+
"eval_loss": 0.5787616968154907,
|
| 233 |
+
"eval_runtime": 4.2726,
|
| 234 |
+
"eval_samples_per_second": 64.832,
|
| 235 |
+
"eval_steps_per_second": 8.192,
|
| 236 |
+
"step": 2184
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"best_epoch": 10,
|
| 240 |
+
"best_eval_accuracy": 0.7184115523465704,
|
| 241 |
+
"epoch": 14.0,
|
| 242 |
+
"step": 2184
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 15.0,
|
| 246 |
+
"eval_accuracy": 0.7075812274368231,
|
| 247 |
+
"eval_loss": 0.664523720741272,
|
| 248 |
+
"eval_runtime": 4.2707,
|
| 249 |
+
"eval_samples_per_second": 64.861,
|
| 250 |
+
"eval_steps_per_second": 8.195,
|
| 251 |
+
"step": 2340
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"best_epoch": 10,
|
| 255 |
+
"best_eval_accuracy": 0.7184115523465704,
|
| 256 |
+
"epoch": 15.0,
|
| 257 |
+
"step": 2340
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 16.0,
|
| 261 |
+
"eval_accuracy": 0.6425992779783394,
|
| 262 |
+
"eval_loss": 0.9684011936187744,
|
| 263 |
+
"eval_runtime": 4.273,
|
| 264 |
+
"eval_samples_per_second": 64.825,
|
| 265 |
+
"eval_steps_per_second": 8.191,
|
| 266 |
+
"step": 2496
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"best_epoch": 10,
|
| 270 |
+
"best_eval_accuracy": 0.7184115523465704,
|
| 271 |
+
"epoch": 16.0,
|
| 272 |
+
"step": 2496
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 16.03,
|
| 276 |
+
"learning_rate": 0.0021987179487179486,
|
| 277 |
+
"loss": 0.587,
|
| 278 |
+
"step": 2500
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 17.0,
|
| 282 |
+
"eval_accuracy": 0.7184115523465704,
|
| 283 |
+
"eval_loss": 0.6227187514305115,
|
| 284 |
+
"eval_runtime": 4.2714,
|
| 285 |
+
"eval_samples_per_second": 64.85,
|
| 286 |
+
"eval_steps_per_second": 8.194,
|
| 287 |
+
"step": 2652
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"best_epoch": 10,
|
| 291 |
+
"best_eval_accuracy": 0.7184115523465704,
|
| 292 |
+
"epoch": 17.0,
|
| 293 |
+
"step": 2652
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 18.0,
|
| 297 |
+
"eval_accuracy": 0.7075812274368231,
|
| 298 |
+
"eval_loss": 0.6448877453804016,
|
| 299 |
+
"eval_runtime": 4.2744,
|
| 300 |
+
"eval_samples_per_second": 64.804,
|
| 301 |
+
"eval_steps_per_second": 8.188,
|
| 302 |
+
"step": 2808
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"best_epoch": 10,
|
| 306 |
+
"best_eval_accuracy": 0.7184115523465704,
|
| 307 |
+
"epoch": 18.0,
|
| 308 |
+
"step": 2808
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 19.0,
|
| 312 |
+
"eval_accuracy": 0.7364620938628159,
|
| 313 |
+
"eval_loss": 0.6650940179824829,
|
| 314 |
+
"eval_runtime": 4.271,
|
| 315 |
+
"eval_samples_per_second": 64.857,
|
| 316 |
+
"eval_steps_per_second": 8.195,
|
| 317 |
+
"step": 2964
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"best_epoch": 18,
|
| 321 |
+
"best_eval_accuracy": 0.7364620938628159,
|
| 322 |
+
"epoch": 19.0,
|
| 323 |
+
"step": 2964
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 19.23,
|
| 327 |
+
"learning_rate": 0.0020384615384615385,
|
| 328 |
+
"loss": 0.5287,
|
| 329 |
+
"step": 3000
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 20.0,
|
| 333 |
+
"eval_accuracy": 0.6498194945848376,
|
| 334 |
+
"eval_loss": 1.132426142692566,
|
| 335 |
+
"eval_runtime": 4.2727,
|
| 336 |
+
"eval_samples_per_second": 64.83,
|
| 337 |
+
"eval_steps_per_second": 8.191,
|
| 338 |
+
"step": 3120
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"best_epoch": 18,
|
| 342 |
+
"best_eval_accuracy": 0.7364620938628159,
|
| 343 |
+
"epoch": 20.0,
|
| 344 |
+
"step": 3120
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 21.0,
|
| 348 |
+
"eval_accuracy": 0.6895306859205776,
|
| 349 |
+
"eval_loss": 0.7391165494918823,
|
| 350 |
+
"eval_runtime": 4.2755,
|
| 351 |
+
"eval_samples_per_second": 64.787,
|
| 352 |
+
"eval_steps_per_second": 8.186,
|
| 353 |
+
"step": 3276
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"best_epoch": 18,
|
| 357 |
+
"best_eval_accuracy": 0.7364620938628159,
|
| 358 |
+
"epoch": 21.0,
|
| 359 |
+
"step": 3276
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 22.0,
|
| 363 |
+
"eval_accuracy": 0.6642599277978339,
|
| 364 |
+
"eval_loss": 1.0194385051727295,
|
| 365 |
+
"eval_runtime": 4.2717,
|
| 366 |
+
"eval_samples_per_second": 64.845,
|
| 367 |
+
"eval_steps_per_second": 8.193,
|
| 368 |
+
"step": 3432
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"best_epoch": 18,
|
| 372 |
+
"best_eval_accuracy": 0.7364620938628159,
|
| 373 |
+
"epoch": 22.0,
|
| 374 |
+
"step": 3432
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 22.44,
|
| 378 |
+
"learning_rate": 0.0018782051282051281,
|
| 379 |
+
"loss": 0.5035,
|
| 380 |
+
"step": 3500
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 23.0,
|
| 384 |
+
"eval_accuracy": 0.703971119133574,
|
| 385 |
+
"eval_loss": 0.7837710380554199,
|
| 386 |
+
"eval_runtime": 4.2729,
|
| 387 |
+
"eval_samples_per_second": 64.826,
|
| 388 |
+
"eval_steps_per_second": 8.191,
|
| 389 |
+
"step": 3588
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"best_epoch": 18,
|
| 393 |
+
"best_eval_accuracy": 0.7364620938628159,
|
| 394 |
+
"epoch": 23.0,
|
| 395 |
+
"step": 3588
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 24.0,
|
| 399 |
+
"eval_accuracy": 0.7184115523465704,
|
| 400 |
+
"eval_loss": 0.8647497296333313,
|
| 401 |
+
"eval_runtime": 4.2733,
|
| 402 |
+
"eval_samples_per_second": 64.821,
|
| 403 |
+
"eval_steps_per_second": 8.19,
|
| 404 |
+
"step": 3744
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"best_epoch": 18,
|
| 408 |
+
"best_eval_accuracy": 0.7364620938628159,
|
| 409 |
+
"epoch": 24.0,
|
| 410 |
+
"step": 3744
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 25.0,
|
| 414 |
+
"eval_accuracy": 0.6714801444043321,
|
| 415 |
+
"eval_loss": 1.0973684787750244,
|
| 416 |
+
"eval_runtime": 4.2735,
|
| 417 |
+
"eval_samples_per_second": 64.817,
|
| 418 |
+
"eval_steps_per_second": 8.19,
|
| 419 |
+
"step": 3900
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"best_epoch": 18,
|
| 423 |
+
"best_eval_accuracy": 0.7364620938628159,
|
| 424 |
+
"epoch": 25.0,
|
| 425 |
+
"step": 3900
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 25.64,
|
| 429 |
+
"learning_rate": 0.0017179487179487178,
|
| 430 |
+
"loss": 0.4533,
|
| 431 |
+
"step": 4000
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 26.0,
|
| 435 |
+
"eval_accuracy": 0.7292418772563177,
|
| 436 |
+
"eval_loss": 0.5861077308654785,
|
| 437 |
+
"eval_runtime": 4.2708,
|
| 438 |
+
"eval_samples_per_second": 64.86,
|
| 439 |
+
"eval_steps_per_second": 8.195,
|
| 440 |
+
"step": 4056
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"best_epoch": 18,
|
| 444 |
+
"best_eval_accuracy": 0.7364620938628159,
|
| 445 |
+
"epoch": 26.0,
|
| 446 |
+
"step": 4056
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 27.0,
|
| 450 |
+
"eval_accuracy": 0.7436823104693141,
|
| 451 |
+
"eval_loss": 0.6684616804122925,
|
| 452 |
+
"eval_runtime": 4.2664,
|
| 453 |
+
"eval_samples_per_second": 64.925,
|
| 454 |
+
"eval_steps_per_second": 8.204,
|
| 455 |
+
"step": 4212
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"best_epoch": 26,
|
| 459 |
+
"best_eval_accuracy": 0.7436823104693141,
|
| 460 |
+
"epoch": 27.0,
|
| 461 |
+
"step": 4212
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 28.0,
|
| 465 |
+
"eval_accuracy": 0.7256317689530686,
|
| 466 |
+
"eval_loss": 0.6998275518417358,
|
| 467 |
+
"eval_runtime": 4.2768,
|
| 468 |
+
"eval_samples_per_second": 64.768,
|
| 469 |
+
"eval_steps_per_second": 8.184,
|
| 470 |
+
"step": 4368
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"best_epoch": 26,
|
| 474 |
+
"best_eval_accuracy": 0.7436823104693141,
|
| 475 |
+
"epoch": 28.0,
|
| 476 |
+
"step": 4368
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 28.85,
|
| 480 |
+
"learning_rate": 0.0015576923076923079,
|
| 481 |
+
"loss": 0.4398,
|
| 482 |
+
"step": 4500
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 29.0,
|
| 486 |
+
"eval_accuracy": 0.7328519855595668,
|
| 487 |
+
"eval_loss": 0.7596197128295898,
|
| 488 |
+
"eval_runtime": 4.2737,
|
| 489 |
+
"eval_samples_per_second": 64.815,
|
| 490 |
+
"eval_steps_per_second": 8.19,
|
| 491 |
+
"step": 4524
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"best_epoch": 26,
|
| 495 |
+
"best_eval_accuracy": 0.7436823104693141,
|
| 496 |
+
"epoch": 29.0,
|
| 497 |
+
"step": 4524
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 30.0,
|
| 501 |
+
"eval_accuracy": 0.7436823104693141,
|
| 502 |
+
"eval_loss": 0.696681797504425,
|
| 503 |
+
"eval_runtime": 4.2817,
|
| 504 |
+
"eval_samples_per_second": 64.695,
|
| 505 |
+
"eval_steps_per_second": 8.174,
|
| 506 |
+
"step": 4680
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"best_epoch": 26,
|
| 510 |
+
"best_eval_accuracy": 0.7436823104693141,
|
| 511 |
+
"epoch": 30.0,
|
| 512 |
+
"step": 4680
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 31.0,
|
| 516 |
+
"eval_accuracy": 0.7472924187725631,
|
| 517 |
+
"eval_loss": 0.7040970325469971,
|
| 518 |
+
"eval_runtime": 4.2717,
|
| 519 |
+
"eval_samples_per_second": 64.845,
|
| 520 |
+
"eval_steps_per_second": 8.193,
|
| 521 |
+
"step": 4836
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"best_epoch": 30,
|
| 525 |
+
"best_eval_accuracy": 0.7472924187725631,
|
| 526 |
+
"epoch": 31.0,
|
| 527 |
+
"step": 4836
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 32.0,
|
| 531 |
+
"eval_accuracy": 0.7328519855595668,
|
| 532 |
+
"eval_loss": 0.7616851925849915,
|
| 533 |
+
"eval_runtime": 4.2802,
|
| 534 |
+
"eval_samples_per_second": 64.716,
|
| 535 |
+
"eval_steps_per_second": 8.177,
|
| 536 |
+
"step": 4992
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"best_epoch": 30,
|
| 540 |
+
"best_eval_accuracy": 0.7472924187725631,
|
| 541 |
+
"epoch": 32.0,
|
| 542 |
+
"step": 4992
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 32.05,
|
| 546 |
+
"learning_rate": 0.0013974358974358976,
|
| 547 |
+
"loss": 0.3837,
|
| 548 |
+
"step": 5000
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 33.0,
|
| 552 |
+
"eval_accuracy": 0.7328519855595668,
|
| 553 |
+
"eval_loss": 0.7990546822547913,
|
| 554 |
+
"eval_runtime": 4.2772,
|
| 555 |
+
"eval_samples_per_second": 64.762,
|
| 556 |
+
"eval_steps_per_second": 8.183,
|
| 557 |
+
"step": 5148
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"best_epoch": 30,
|
| 561 |
+
"best_eval_accuracy": 0.7472924187725631,
|
| 562 |
+
"epoch": 33.0,
|
| 563 |
+
"step": 5148
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 34.0,
|
| 567 |
+
"eval_accuracy": 0.7472924187725631,
|
| 568 |
+
"eval_loss": 0.8228901028633118,
|
| 569 |
+
"eval_runtime": 4.2769,
|
| 570 |
+
"eval_samples_per_second": 64.766,
|
| 571 |
+
"eval_steps_per_second": 8.183,
|
| 572 |
+
"step": 5304
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"best_epoch": 30,
|
| 576 |
+
"best_eval_accuracy": 0.7472924187725631,
|
| 577 |
+
"epoch": 34.0,
|
| 578 |
+
"step": 5304
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 35.0,
|
| 582 |
+
"eval_accuracy": 0.740072202166065,
|
| 583 |
+
"eval_loss": 0.7745354175567627,
|
| 584 |
+
"eval_runtime": 4.2809,
|
| 585 |
+
"eval_samples_per_second": 64.706,
|
| 586 |
+
"eval_steps_per_second": 8.176,
|
| 587 |
+
"step": 5460
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"best_epoch": 30,
|
| 591 |
+
"best_eval_accuracy": 0.7472924187725631,
|
| 592 |
+
"epoch": 35.0,
|
| 593 |
+
"step": 5460
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 35.26,
|
| 597 |
+
"learning_rate": 0.0012371794871794872,
|
| 598 |
+
"loss": 0.3471,
|
| 599 |
+
"step": 5500
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 36.0,
|
| 603 |
+
"eval_accuracy": 0.7436823104693141,
|
| 604 |
+
"eval_loss": 0.778685450553894,
|
| 605 |
+
"eval_runtime": 4.2824,
|
| 606 |
+
"eval_samples_per_second": 64.684,
|
| 607 |
+
"eval_steps_per_second": 8.173,
|
| 608 |
+
"step": 5616
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"best_epoch": 30,
|
| 612 |
+
"best_eval_accuracy": 0.7472924187725631,
|
| 613 |
+
"epoch": 36.0,
|
| 614 |
+
"step": 5616
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 37.0,
|
| 618 |
+
"eval_accuracy": 0.7364620938628159,
|
| 619 |
+
"eval_loss": 0.7990731596946716,
|
| 620 |
+
"eval_runtime": 4.2833,
|
| 621 |
+
"eval_samples_per_second": 64.669,
|
| 622 |
+
"eval_steps_per_second": 8.171,
|
| 623 |
+
"step": 5772
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"best_epoch": 30,
|
| 627 |
+
"best_eval_accuracy": 0.7472924187725631,
|
| 628 |
+
"epoch": 37.0,
|
| 629 |
+
"step": 5772
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 38.0,
|
| 633 |
+
"eval_accuracy": 0.7256317689530686,
|
| 634 |
+
"eval_loss": 1.0205601453781128,
|
| 635 |
+
"eval_runtime": 4.2801,
|
| 636 |
+
"eval_samples_per_second": 64.718,
|
| 637 |
+
"eval_steps_per_second": 8.177,
|
| 638 |
+
"step": 5928
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"best_epoch": 30,
|
| 642 |
+
"best_eval_accuracy": 0.7472924187725631,
|
| 643 |
+
"epoch": 38.0,
|
| 644 |
+
"step": 5928
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 38.46,
|
| 648 |
+
"learning_rate": 0.0010769230769230769,
|
| 649 |
+
"loss": 0.3303,
|
| 650 |
+
"step": 6000
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 39.0,
|
| 654 |
+
"eval_accuracy": 0.7292418772563177,
|
| 655 |
+
"eval_loss": 0.8976525664329529,
|
| 656 |
+
"eval_runtime": 4.2827,
|
| 657 |
+
"eval_samples_per_second": 64.679,
|
| 658 |
+
"eval_steps_per_second": 8.172,
|
| 659 |
+
"step": 6084
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"best_epoch": 30,
|
| 663 |
+
"best_eval_accuracy": 0.7472924187725631,
|
| 664 |
+
"epoch": 39.0,
|
| 665 |
+
"step": 6084
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 40.0,
|
| 669 |
+
"eval_accuracy": 0.7220216606498195,
|
| 670 |
+
"eval_loss": 0.7327412366867065,
|
| 671 |
+
"eval_runtime": 4.2744,
|
| 672 |
+
"eval_samples_per_second": 64.805,
|
| 673 |
+
"eval_steps_per_second": 8.188,
|
| 674 |
+
"step": 6240
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"best_epoch": 30,
|
| 678 |
+
"best_eval_accuracy": 0.7472924187725631,
|
| 679 |
+
"epoch": 40.0,
|
| 680 |
+
"step": 6240
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 41.0,
|
| 684 |
+
"eval_accuracy": 0.7292418772563177,
|
| 685 |
+
"eval_loss": 0.810184895992279,
|
| 686 |
+
"eval_runtime": 4.2816,
|
| 687 |
+
"eval_samples_per_second": 64.696,
|
| 688 |
+
"eval_steps_per_second": 8.175,
|
| 689 |
+
"step": 6396
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"best_epoch": 30,
|
| 693 |
+
"best_eval_accuracy": 0.7472924187725631,
|
| 694 |
+
"epoch": 41.0,
|
| 695 |
+
"step": 6396
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 41.67,
|
| 699 |
+
"learning_rate": 0.0009166666666666668,
|
| 700 |
+
"loss": 0.2991,
|
| 701 |
+
"step": 6500
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 42.0,
|
| 705 |
+
"eval_accuracy": 0.7472924187725631,
|
| 706 |
+
"eval_loss": 0.7347400784492493,
|
| 707 |
+
"eval_runtime": 4.2774,
|
| 708 |
+
"eval_samples_per_second": 64.759,
|
| 709 |
+
"eval_steps_per_second": 8.182,
|
| 710 |
+
"step": 6552
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"best_epoch": 30,
|
| 714 |
+
"best_eval_accuracy": 0.7472924187725631,
|
| 715 |
+
"epoch": 42.0,
|
| 716 |
+
"step": 6552
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 43.0,
|
| 720 |
+
"eval_accuracy": 0.7472924187725631,
|
| 721 |
+
"eval_loss": 0.867670476436615,
|
| 722 |
+
"eval_runtime": 4.2777,
|
| 723 |
+
"eval_samples_per_second": 64.754,
|
| 724 |
+
"eval_steps_per_second": 8.182,
|
| 725 |
+
"step": 6708
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"best_epoch": 30,
|
| 729 |
+
"best_eval_accuracy": 0.7472924187725631,
|
| 730 |
+
"epoch": 43.0,
|
| 731 |
+
"step": 6708
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 44.0,
|
| 735 |
+
"eval_accuracy": 0.7364620938628159,
|
| 736 |
+
"eval_loss": 0.9773887991905212,
|
| 737 |
+
"eval_runtime": 4.2767,
|
| 738 |
+
"eval_samples_per_second": 64.77,
|
| 739 |
+
"eval_steps_per_second": 8.184,
|
| 740 |
+
"step": 6864
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"best_epoch": 30,
|
| 744 |
+
"best_eval_accuracy": 0.7472924187725631,
|
| 745 |
+
"epoch": 44.0,
|
| 746 |
+
"step": 6864
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 44.87,
|
| 750 |
+
"learning_rate": 0.0007564102564102564,
|
| 751 |
+
"loss": 0.275,
|
| 752 |
+
"step": 7000
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 45.0,
|
| 756 |
+
"eval_accuracy": 0.7581227436823105,
|
| 757 |
+
"eval_loss": 0.8556901216506958,
|
| 758 |
+
"eval_runtime": 4.277,
|
| 759 |
+
"eval_samples_per_second": 64.765,
|
| 760 |
+
"eval_steps_per_second": 8.183,
|
| 761 |
+
"step": 7020
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"best_epoch": 44,
|
| 765 |
+
"best_eval_accuracy": 0.7581227436823105,
|
| 766 |
+
"epoch": 45.0,
|
| 767 |
+
"step": 7020
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 46.0,
|
| 771 |
+
"eval_accuracy": 0.7436823104693141,
|
| 772 |
+
"eval_loss": 0.9789180755615234,
|
| 773 |
+
"eval_runtime": 4.2782,
|
| 774 |
+
"eval_samples_per_second": 64.747,
|
| 775 |
+
"eval_steps_per_second": 8.181,
|
| 776 |
+
"step": 7176
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"best_epoch": 44,
|
| 780 |
+
"best_eval_accuracy": 0.7581227436823105,
|
| 781 |
+
"epoch": 46.0,
|
| 782 |
+
"step": 7176
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 47.0,
|
| 786 |
+
"eval_accuracy": 0.7436823104693141,
|
| 787 |
+
"eval_loss": 1.001485824584961,
|
| 788 |
+
"eval_runtime": 4.2769,
|
| 789 |
+
"eval_samples_per_second": 64.767,
|
| 790 |
+
"eval_steps_per_second": 8.184,
|
| 791 |
+
"step": 7332
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"best_epoch": 44,
|
| 795 |
+
"best_eval_accuracy": 0.7581227436823105,
|
| 796 |
+
"epoch": 47.0,
|
| 797 |
+
"step": 7332
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 48.0,
|
| 801 |
+
"eval_accuracy": 0.740072202166065,
|
| 802 |
+
"eval_loss": 0.8449525833129883,
|
| 803 |
+
"eval_runtime": 4.2736,
|
| 804 |
+
"eval_samples_per_second": 64.816,
|
| 805 |
+
"eval_steps_per_second": 8.19,
|
| 806 |
+
"step": 7488
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"best_epoch": 44,
|
| 810 |
+
"best_eval_accuracy": 0.7581227436823105,
|
| 811 |
+
"epoch": 48.0,
|
| 812 |
+
"step": 7488
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 48.08,
|
| 816 |
+
"learning_rate": 0.0005961538461538461,
|
| 817 |
+
"loss": 0.2596,
|
| 818 |
+
"step": 7500
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 49.0,
|
| 822 |
+
"eval_accuracy": 0.7581227436823105,
|
| 823 |
+
"eval_loss": 0.8222180008888245,
|
| 824 |
+
"eval_runtime": 4.278,
|
| 825 |
+
"eval_samples_per_second": 64.75,
|
| 826 |
+
"eval_steps_per_second": 8.181,
|
| 827 |
+
"step": 7644
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"best_epoch": 44,
|
| 831 |
+
"best_eval_accuracy": 0.7581227436823105,
|
| 832 |
+
"epoch": 49.0,
|
| 833 |
+
"step": 7644
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 50.0,
|
| 837 |
+
"eval_accuracy": 0.740072202166065,
|
| 838 |
+
"eval_loss": 0.8967640995979309,
|
| 839 |
+
"eval_runtime": 4.286,
|
| 840 |
+
"eval_samples_per_second": 64.629,
|
| 841 |
+
"eval_steps_per_second": 8.166,
|
| 842 |
+
"step": 7800
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"best_epoch": 44,
|
| 846 |
+
"best_eval_accuracy": 0.7581227436823105,
|
| 847 |
+
"epoch": 50.0,
|
| 848 |
+
"step": 7800
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 51.0,
|
| 852 |
+
"eval_accuracy": 0.7436823104693141,
|
| 853 |
+
"eval_loss": 0.8584460616111755,
|
| 854 |
+
"eval_runtime": 4.2772,
|
| 855 |
+
"eval_samples_per_second": 64.762,
|
| 856 |
+
"eval_steps_per_second": 8.183,
|
| 857 |
+
"step": 7956
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"best_epoch": 44,
|
| 861 |
+
"best_eval_accuracy": 0.7581227436823105,
|
| 862 |
+
"epoch": 51.0,
|
| 863 |
+
"step": 7956
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 51.28,
|
| 867 |
+
"learning_rate": 0.00043589743589743596,
|
| 868 |
+
"loss": 0.2469,
|
| 869 |
+
"step": 8000
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 52.0,
|
| 873 |
+
"eval_accuracy": 0.740072202166065,
|
| 874 |
+
"eval_loss": 0.9157448410987854,
|
| 875 |
+
"eval_runtime": 4.2727,
|
| 876 |
+
"eval_samples_per_second": 64.83,
|
| 877 |
+
"eval_steps_per_second": 8.191,
|
| 878 |
+
"step": 8112
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"best_epoch": 44,
|
| 882 |
+
"best_eval_accuracy": 0.7581227436823105,
|
| 883 |
+
"epoch": 52.0,
|
| 884 |
+
"step": 8112
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 53.0,
|
| 888 |
+
"eval_accuracy": 0.7364620938628159,
|
| 889 |
+
"eval_loss": 0.9731793403625488,
|
| 890 |
+
"eval_runtime": 4.2736,
|
| 891 |
+
"eval_samples_per_second": 64.817,
|
| 892 |
+
"eval_steps_per_second": 8.19,
|
| 893 |
+
"step": 8268
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"best_epoch": 44,
|
| 897 |
+
"best_eval_accuracy": 0.7581227436823105,
|
| 898 |
+
"epoch": 53.0,
|
| 899 |
+
"step": 8268
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 54.0,
|
| 903 |
+
"eval_accuracy": 0.740072202166065,
|
| 904 |
+
"eval_loss": 1.0671247243881226,
|
| 905 |
+
"eval_runtime": 4.2701,
|
| 906 |
+
"eval_samples_per_second": 64.87,
|
| 907 |
+
"eval_steps_per_second": 8.197,
|
| 908 |
+
"step": 8424
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"best_epoch": 44,
|
| 912 |
+
"best_eval_accuracy": 0.7581227436823105,
|
| 913 |
+
"epoch": 54.0,
|
| 914 |
+
"step": 8424
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 54.49,
|
| 918 |
+
"learning_rate": 0.0002756410256410257,
|
| 919 |
+
"loss": 0.2303,
|
| 920 |
+
"step": 8500
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 55.0,
|
| 924 |
+
"eval_accuracy": 0.7472924187725631,
|
| 925 |
+
"eval_loss": 0.9512353539466858,
|
| 926 |
+
"eval_runtime": 4.2682,
|
| 927 |
+
"eval_samples_per_second": 64.899,
|
| 928 |
+
"eval_steps_per_second": 8.2,
|
| 929 |
+
"step": 8580
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"best_epoch": 44,
|
| 933 |
+
"best_eval_accuracy": 0.7581227436823105,
|
| 934 |
+
"epoch": 55.0,
|
| 935 |
+
"step": 8580
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 56.0,
|
| 939 |
+
"eval_accuracy": 0.7472924187725631,
|
| 940 |
+
"eval_loss": 0.8708024621009827,
|
| 941 |
+
"eval_runtime": 4.2721,
|
| 942 |
+
"eval_samples_per_second": 64.839,
|
| 943 |
+
"eval_steps_per_second": 8.193,
|
| 944 |
+
"step": 8736
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"best_epoch": 44,
|
| 948 |
+
"best_eval_accuracy": 0.7581227436823105,
|
| 949 |
+
"epoch": 56.0,
|
| 950 |
+
"step": 8736
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 57.0,
|
| 954 |
+
"eval_accuracy": 0.7436823104693141,
|
| 955 |
+
"eval_loss": 0.9289500117301941,
|
| 956 |
+
"eval_runtime": 4.2714,
|
| 957 |
+
"eval_samples_per_second": 64.849,
|
| 958 |
+
"eval_steps_per_second": 8.194,
|
| 959 |
+
"step": 8892
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"best_epoch": 44,
|
| 963 |
+
"best_eval_accuracy": 0.7581227436823105,
|
| 964 |
+
"epoch": 57.0,
|
| 965 |
+
"step": 8892
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 57.69,
|
| 969 |
+
"learning_rate": 0.0001153846153846154,
|
| 970 |
+
"loss": 0.2275,
|
| 971 |
+
"step": 9000
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 58.0,
|
| 975 |
+
"eval_accuracy": 0.740072202166065,
|
| 976 |
+
"eval_loss": 0.8866466283798218,
|
| 977 |
+
"eval_runtime": 4.2702,
|
| 978 |
+
"eval_samples_per_second": 64.868,
|
| 979 |
+
"eval_steps_per_second": 8.196,
|
| 980 |
+
"step": 9048
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"best_epoch": 44,
|
| 984 |
+
"best_eval_accuracy": 0.7581227436823105,
|
| 985 |
+
"epoch": 58.0,
|
| 986 |
+
"step": 9048
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 59.0,
|
| 990 |
+
"eval_accuracy": 0.7364620938628159,
|
| 991 |
+
"eval_loss": 0.936638593673706,
|
| 992 |
+
"eval_runtime": 4.2724,
|
| 993 |
+
"eval_samples_per_second": 64.835,
|
| 994 |
+
"eval_steps_per_second": 8.192,
|
| 995 |
+
"step": 9204
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"best_epoch": 44,
|
| 999 |
+
"best_eval_accuracy": 0.7581227436823105,
|
| 1000 |
+
"epoch": 59.0,
|
| 1001 |
+
"step": 9204
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 60.0,
|
| 1005 |
+
"eval_accuracy": 0.7436823104693141,
|
| 1006 |
+
"eval_loss": 0.9555269479751587,
|
| 1007 |
+
"eval_runtime": 4.279,
|
| 1008 |
+
"eval_samples_per_second": 64.735,
|
| 1009 |
+
"eval_steps_per_second": 8.179,
|
| 1010 |
+
"step": 9360
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"best_epoch": 44,
|
| 1014 |
+
"best_eval_accuracy": 0.7581227436823105,
|
| 1015 |
+
"epoch": 60.0,
|
| 1016 |
+
"step": 9360
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 60.0,
|
| 1020 |
+
"step": 9360,
|
| 1021 |
+
"total_flos": 6.96152728406016e+16,
|
| 1022 |
+
"train_loss": 0.4369181796016856,
|
| 1023 |
+
"train_runtime": 3671.2013,
|
| 1024 |
+
"train_samples_per_second": 40.695,
|
| 1025 |
+
"train_steps_per_second": 2.55
|
| 1026 |
+
}
|
| 1027 |
+
],
|
| 1028 |
+
"max_steps": 9360,
|
| 1029 |
+
"num_train_epochs": 60,
|
| 1030 |
+
"total_flos": 6.96152728406016e+16,
|
| 1031 |
+
"trial_name": null,
|
| 1032 |
+
"trial_params": null
|
| 1033 |
+
}
|