End of training
Browse files- all_results.json +8 -0
- train_results.json +8 -0
- trainer_state.json +1117 -0
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 14.89,
|
| 3 |
+
"total_flos": 7.877470546062803e+18,
|
| 4 |
+
"train_loss": 0.5140632056590146,
|
| 5 |
+
"train_runtime": 3982.2883,
|
| 6 |
+
"train_samples_per_second": 25.704,
|
| 7 |
+
"train_steps_per_second": 0.399
|
| 8 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 14.89,
|
| 3 |
+
"total_flos": 7.877470546062803e+18,
|
| 4 |
+
"train_loss": 0.5140632056590146,
|
| 5 |
+
"train_runtime": 3982.2883,
|
| 6 |
+
"train_samples_per_second": 25.704,
|
| 7 |
+
"train_steps_per_second": 0.399
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1117 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.8398058252427184,
|
| 3 |
+
"best_model_checkpoint": "vit-base-patch16-224-brand/checkpoint-1590",
|
| 4 |
+
"epoch": 14.894613583138174,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 1590,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.09,
|
| 13 |
+
"learning_rate": 3.1446540880503146e-06,
|
| 14 |
+
"loss": 2.3942,
|
| 15 |
+
"step": 10
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.19,
|
| 19 |
+
"learning_rate": 6.289308176100629e-06,
|
| 20 |
+
"loss": 2.3923,
|
| 21 |
+
"step": 20
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.28,
|
| 25 |
+
"learning_rate": 9.433962264150944e-06,
|
| 26 |
+
"loss": 2.3106,
|
| 27 |
+
"step": 30
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.37,
|
| 31 |
+
"learning_rate": 1.2578616352201259e-05,
|
| 32 |
+
"loss": 2.2216,
|
| 33 |
+
"step": 40
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.47,
|
| 37 |
+
"learning_rate": 1.572327044025157e-05,
|
| 38 |
+
"loss": 2.0774,
|
| 39 |
+
"step": 50
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.56,
|
| 43 |
+
"learning_rate": 1.8867924528301888e-05,
|
| 44 |
+
"loss": 2.0392,
|
| 45 |
+
"step": 60
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.66,
|
| 49 |
+
"learning_rate": 2.2012578616352204e-05,
|
| 50 |
+
"loss": 1.8905,
|
| 51 |
+
"step": 70
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.75,
|
| 55 |
+
"learning_rate": 2.5157232704402517e-05,
|
| 56 |
+
"loss": 1.8387,
|
| 57 |
+
"step": 80
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.84,
|
| 61 |
+
"learning_rate": 2.830188679245283e-05,
|
| 62 |
+
"loss": 1.7255,
|
| 63 |
+
"step": 90
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.94,
|
| 67 |
+
"learning_rate": 3.144654088050314e-05,
|
| 68 |
+
"loss": 1.6631,
|
| 69 |
+
"step": 100
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.99,
|
| 73 |
+
"eval_accuracy": 0.5494263018534863,
|
| 74 |
+
"eval_loss": 1.4344048500061035,
|
| 75 |
+
"eval_runtime": 29.697,
|
| 76 |
+
"eval_samples_per_second": 76.304,
|
| 77 |
+
"eval_steps_per_second": 4.782,
|
| 78 |
+
"step": 106
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 1.03,
|
| 82 |
+
"learning_rate": 3.4591194968553456e-05,
|
| 83 |
+
"loss": 1.5273,
|
| 84 |
+
"step": 110
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.12,
|
| 88 |
+
"learning_rate": 3.7735849056603776e-05,
|
| 89 |
+
"loss": 1.4406,
|
| 90 |
+
"step": 120
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 1.22,
|
| 94 |
+
"learning_rate": 4.088050314465409e-05,
|
| 95 |
+
"loss": 1.4031,
|
| 96 |
+
"step": 130
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 1.31,
|
| 100 |
+
"learning_rate": 4.402515723270441e-05,
|
| 101 |
+
"loss": 1.2573,
|
| 102 |
+
"step": 140
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 1.41,
|
| 106 |
+
"learning_rate": 4.716981132075472e-05,
|
| 107 |
+
"loss": 1.2428,
|
| 108 |
+
"step": 150
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 1.5,
|
| 112 |
+
"learning_rate": 4.9965059399021665e-05,
|
| 113 |
+
"loss": 1.1119,
|
| 114 |
+
"step": 160
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 1.59,
|
| 118 |
+
"learning_rate": 4.9615653389238295e-05,
|
| 119 |
+
"loss": 1.2198,
|
| 120 |
+
"step": 170
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"epoch": 1.69,
|
| 124 |
+
"learning_rate": 4.9266247379454926e-05,
|
| 125 |
+
"loss": 1.0555,
|
| 126 |
+
"step": 180
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 1.78,
|
| 130 |
+
"learning_rate": 4.891684136967156e-05,
|
| 131 |
+
"loss": 1.0781,
|
| 132 |
+
"step": 190
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"epoch": 1.87,
|
| 136 |
+
"learning_rate": 4.856743535988819e-05,
|
| 137 |
+
"loss": 1.0989,
|
| 138 |
+
"step": 200
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 1.97,
|
| 142 |
+
"learning_rate": 4.8218029350104823e-05,
|
| 143 |
+
"loss": 1.069,
|
| 144 |
+
"step": 210
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 2.0,
|
| 148 |
+
"eval_accuracy": 0.7281553398058253,
|
| 149 |
+
"eval_loss": 0.8916863799095154,
|
| 150 |
+
"eval_runtime": 29.9218,
|
| 151 |
+
"eval_samples_per_second": 75.731,
|
| 152 |
+
"eval_steps_per_second": 4.746,
|
| 153 |
+
"step": 213
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 2.06,
|
| 157 |
+
"learning_rate": 4.7868623340321454e-05,
|
| 158 |
+
"loss": 0.9655,
|
| 159 |
+
"step": 220
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 2.15,
|
| 163 |
+
"learning_rate": 4.7519217330538084e-05,
|
| 164 |
+
"loss": 0.8571,
|
| 165 |
+
"step": 230
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 2.25,
|
| 169 |
+
"learning_rate": 4.716981132075472e-05,
|
| 170 |
+
"loss": 0.8375,
|
| 171 |
+
"step": 240
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 2.34,
|
| 175 |
+
"learning_rate": 4.682040531097135e-05,
|
| 176 |
+
"loss": 0.8631,
|
| 177 |
+
"step": 250
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 2.44,
|
| 181 |
+
"learning_rate": 4.647099930118798e-05,
|
| 182 |
+
"loss": 0.8991,
|
| 183 |
+
"step": 260
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 2.53,
|
| 187 |
+
"learning_rate": 4.612159329140461e-05,
|
| 188 |
+
"loss": 0.9349,
|
| 189 |
+
"step": 270
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 2.62,
|
| 193 |
+
"learning_rate": 4.577218728162125e-05,
|
| 194 |
+
"loss": 0.8766,
|
| 195 |
+
"step": 280
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 2.72,
|
| 199 |
+
"learning_rate": 4.542278127183788e-05,
|
| 200 |
+
"loss": 0.8294,
|
| 201 |
+
"step": 290
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 2.81,
|
| 205 |
+
"learning_rate": 4.5073375262054504e-05,
|
| 206 |
+
"loss": 0.8507,
|
| 207 |
+
"step": 300
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 2.9,
|
| 211 |
+
"learning_rate": 4.472396925227114e-05,
|
| 212 |
+
"loss": 0.7529,
|
| 213 |
+
"step": 310
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 3.0,
|
| 217 |
+
"learning_rate": 4.437456324248777e-05,
|
| 218 |
+
"loss": 0.801,
|
| 219 |
+
"step": 320
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 3.0,
|
| 223 |
+
"eval_accuracy": 0.7533097969991174,
|
| 224 |
+
"eval_loss": 0.7611907720565796,
|
| 225 |
+
"eval_runtime": 29.9179,
|
| 226 |
+
"eval_samples_per_second": 75.741,
|
| 227 |
+
"eval_steps_per_second": 4.746,
|
| 228 |
+
"step": 320
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 3.09,
|
| 232 |
+
"learning_rate": 4.402515723270441e-05,
|
| 233 |
+
"loss": 0.7598,
|
| 234 |
+
"step": 330
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 3.19,
|
| 238 |
+
"learning_rate": 4.367575122292103e-05,
|
| 239 |
+
"loss": 0.6375,
|
| 240 |
+
"step": 340
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 3.28,
|
| 244 |
+
"learning_rate": 4.332634521313767e-05,
|
| 245 |
+
"loss": 0.6634,
|
| 246 |
+
"step": 350
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 3.37,
|
| 250 |
+
"learning_rate": 4.29769392033543e-05,
|
| 251 |
+
"loss": 0.6474,
|
| 252 |
+
"step": 360
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 3.47,
|
| 256 |
+
"learning_rate": 4.262753319357093e-05,
|
| 257 |
+
"loss": 0.6328,
|
| 258 |
+
"step": 370
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 3.56,
|
| 262 |
+
"learning_rate": 4.227812718378756e-05,
|
| 263 |
+
"loss": 0.6945,
|
| 264 |
+
"step": 380
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 3.65,
|
| 268 |
+
"learning_rate": 4.192872117400419e-05,
|
| 269 |
+
"loss": 0.6194,
|
| 270 |
+
"step": 390
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 3.75,
|
| 274 |
+
"learning_rate": 4.157931516422083e-05,
|
| 275 |
+
"loss": 0.6398,
|
| 276 |
+
"step": 400
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 3.84,
|
| 280 |
+
"learning_rate": 4.122990915443746e-05,
|
| 281 |
+
"loss": 0.6756,
|
| 282 |
+
"step": 410
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 3.93,
|
| 286 |
+
"learning_rate": 4.088050314465409e-05,
|
| 287 |
+
"loss": 0.6075,
|
| 288 |
+
"step": 420
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 4.0,
|
| 292 |
+
"eval_accuracy": 0.792144748455428,
|
| 293 |
+
"eval_loss": 0.6522440314292908,
|
| 294 |
+
"eval_runtime": 29.5461,
|
| 295 |
+
"eval_samples_per_second": 76.694,
|
| 296 |
+
"eval_steps_per_second": 4.806,
|
| 297 |
+
"step": 427
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 4.03,
|
| 301 |
+
"learning_rate": 4.053109713487072e-05,
|
| 302 |
+
"loss": 0.5489,
|
| 303 |
+
"step": 430
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 4.12,
|
| 307 |
+
"learning_rate": 4.018169112508735e-05,
|
| 308 |
+
"loss": 0.5316,
|
| 309 |
+
"step": 440
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 4.22,
|
| 313 |
+
"learning_rate": 3.983228511530399e-05,
|
| 314 |
+
"loss": 0.523,
|
| 315 |
+
"step": 450
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 4.31,
|
| 319 |
+
"learning_rate": 3.948287910552062e-05,
|
| 320 |
+
"loss": 0.5247,
|
| 321 |
+
"step": 460
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 4.4,
|
| 325 |
+
"learning_rate": 3.913347309573725e-05,
|
| 326 |
+
"loss": 0.5356,
|
| 327 |
+
"step": 470
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 4.5,
|
| 331 |
+
"learning_rate": 3.878406708595388e-05,
|
| 332 |
+
"loss": 0.5128,
|
| 333 |
+
"step": 480
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 4.59,
|
| 337 |
+
"learning_rate": 3.8434661076170515e-05,
|
| 338 |
+
"loss": 0.5752,
|
| 339 |
+
"step": 490
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 4.68,
|
| 343 |
+
"learning_rate": 3.8085255066387145e-05,
|
| 344 |
+
"loss": 0.5445,
|
| 345 |
+
"step": 500
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 4.78,
|
| 349 |
+
"learning_rate": 3.7735849056603776e-05,
|
| 350 |
+
"loss": 0.5064,
|
| 351 |
+
"step": 510
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 4.87,
|
| 355 |
+
"learning_rate": 3.7386443046820406e-05,
|
| 356 |
+
"loss": 0.508,
|
| 357 |
+
"step": 520
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 4.96,
|
| 361 |
+
"learning_rate": 3.7037037037037037e-05,
|
| 362 |
+
"loss": 0.5046,
|
| 363 |
+
"step": 530
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 4.99,
|
| 367 |
+
"eval_accuracy": 0.8005295675198588,
|
| 368 |
+
"eval_loss": 0.6084854006767273,
|
| 369 |
+
"eval_runtime": 29.8921,
|
| 370 |
+
"eval_samples_per_second": 75.806,
|
| 371 |
+
"eval_steps_per_second": 4.75,
|
| 372 |
+
"step": 533
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 5.06,
|
| 376 |
+
"learning_rate": 3.6687631027253674e-05,
|
| 377 |
+
"loss": 0.4403,
|
| 378 |
+
"step": 540
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 5.15,
|
| 382 |
+
"learning_rate": 3.6338225017470304e-05,
|
| 383 |
+
"loss": 0.3919,
|
| 384 |
+
"step": 550
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 5.25,
|
| 388 |
+
"learning_rate": 3.5988819007686934e-05,
|
| 389 |
+
"loss": 0.4224,
|
| 390 |
+
"step": 560
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 5.34,
|
| 394 |
+
"learning_rate": 3.5639412997903565e-05,
|
| 395 |
+
"loss": 0.4708,
|
| 396 |
+
"step": 570
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 5.43,
|
| 400 |
+
"learning_rate": 3.52900069881202e-05,
|
| 401 |
+
"loss": 0.4206,
|
| 402 |
+
"step": 580
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 5.53,
|
| 406 |
+
"learning_rate": 3.494060097833683e-05,
|
| 407 |
+
"loss": 0.4416,
|
| 408 |
+
"step": 590
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 5.62,
|
| 412 |
+
"learning_rate": 3.4591194968553456e-05,
|
| 413 |
+
"loss": 0.4478,
|
| 414 |
+
"step": 600
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 5.71,
|
| 418 |
+
"learning_rate": 3.424178895877009e-05,
|
| 419 |
+
"loss": 0.4403,
|
| 420 |
+
"step": 610
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 5.81,
|
| 424 |
+
"learning_rate": 3.3892382948986724e-05,
|
| 425 |
+
"loss": 0.4428,
|
| 426 |
+
"step": 620
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 5.9,
|
| 430 |
+
"learning_rate": 3.354297693920336e-05,
|
| 431 |
+
"loss": 0.3941,
|
| 432 |
+
"step": 630
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 6.0,
|
| 436 |
+
"learning_rate": 3.3193570929419984e-05,
|
| 437 |
+
"loss": 0.4018,
|
| 438 |
+
"step": 640
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 6.0,
|
| 442 |
+
"eval_accuracy": 0.8022947925860547,
|
| 443 |
+
"eval_loss": 0.6132453680038452,
|
| 444 |
+
"eval_runtime": 30.1158,
|
| 445 |
+
"eval_samples_per_second": 75.243,
|
| 446 |
+
"eval_steps_per_second": 4.715,
|
| 447 |
+
"step": 640
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 6.09,
|
| 451 |
+
"learning_rate": 3.284416491963662e-05,
|
| 452 |
+
"loss": 0.3719,
|
| 453 |
+
"step": 650
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 6.18,
|
| 457 |
+
"learning_rate": 3.249475890985325e-05,
|
| 458 |
+
"loss": 0.3549,
|
| 459 |
+
"step": 660
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 6.28,
|
| 463 |
+
"learning_rate": 3.214535290006988e-05,
|
| 464 |
+
"loss": 0.3866,
|
| 465 |
+
"step": 670
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 6.37,
|
| 469 |
+
"learning_rate": 3.179594689028651e-05,
|
| 470 |
+
"loss": 0.4172,
|
| 471 |
+
"step": 680
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 6.46,
|
| 475 |
+
"learning_rate": 3.144654088050314e-05,
|
| 476 |
+
"loss": 0.3798,
|
| 477 |
+
"step": 690
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 6.56,
|
| 481 |
+
"learning_rate": 3.109713487071978e-05,
|
| 482 |
+
"loss": 0.3288,
|
| 483 |
+
"step": 700
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 6.65,
|
| 487 |
+
"learning_rate": 3.074772886093641e-05,
|
| 488 |
+
"loss": 0.3829,
|
| 489 |
+
"step": 710
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 6.74,
|
| 493 |
+
"learning_rate": 3.0398322851153044e-05,
|
| 494 |
+
"loss": 0.3771,
|
| 495 |
+
"step": 720
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 6.84,
|
| 499 |
+
"learning_rate": 3.004891684136967e-05,
|
| 500 |
+
"loss": 0.3637,
|
| 501 |
+
"step": 730
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 6.93,
|
| 505 |
+
"learning_rate": 2.9699510831586302e-05,
|
| 506 |
+
"loss": 0.3641,
|
| 507 |
+
"step": 740
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 7.0,
|
| 511 |
+
"eval_accuracy": 0.8292144748455428,
|
| 512 |
+
"eval_loss": 0.5509597659111023,
|
| 513 |
+
"eval_runtime": 29.6593,
|
| 514 |
+
"eval_samples_per_second": 76.401,
|
| 515 |
+
"eval_steps_per_second": 4.788,
|
| 516 |
+
"step": 747
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 7.03,
|
| 520 |
+
"learning_rate": 2.935010482180294e-05,
|
| 521 |
+
"loss": 0.373,
|
| 522 |
+
"step": 750
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 7.12,
|
| 526 |
+
"learning_rate": 2.9000698812019566e-05,
|
| 527 |
+
"loss": 0.3128,
|
| 528 |
+
"step": 760
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 7.21,
|
| 532 |
+
"learning_rate": 2.8651292802236203e-05,
|
| 533 |
+
"loss": 0.3177,
|
| 534 |
+
"step": 770
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 7.31,
|
| 538 |
+
"learning_rate": 2.830188679245283e-05,
|
| 539 |
+
"loss": 0.2853,
|
| 540 |
+
"step": 780
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 7.4,
|
| 544 |
+
"learning_rate": 2.7952480782669467e-05,
|
| 545 |
+
"loss": 0.2969,
|
| 546 |
+
"step": 790
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 7.49,
|
| 550 |
+
"learning_rate": 2.7603074772886094e-05,
|
| 551 |
+
"loss": 0.2786,
|
| 552 |
+
"step": 800
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 7.59,
|
| 556 |
+
"learning_rate": 2.7253668763102725e-05,
|
| 557 |
+
"loss": 0.2653,
|
| 558 |
+
"step": 810
|
| 559 |
+
},
|
| 560 |
+
{
|
| 561 |
+
"epoch": 7.68,
|
| 562 |
+
"learning_rate": 2.690426275331936e-05,
|
| 563 |
+
"loss": 0.2388,
|
| 564 |
+
"step": 820
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 7.78,
|
| 568 |
+
"learning_rate": 2.655485674353599e-05,
|
| 569 |
+
"loss": 0.2941,
|
| 570 |
+
"step": 830
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 7.87,
|
| 574 |
+
"learning_rate": 2.6205450733752623e-05,
|
| 575 |
+
"loss": 0.3248,
|
| 576 |
+
"step": 840
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 7.96,
|
| 580 |
+
"learning_rate": 2.5856044723969253e-05,
|
| 581 |
+
"loss": 0.4003,
|
| 582 |
+
"step": 850
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 8.0,
|
| 586 |
+
"eval_accuracy": 0.824801412180053,
|
| 587 |
+
"eval_loss": 0.530396580696106,
|
| 588 |
+
"eval_runtime": 29.7498,
|
| 589 |
+
"eval_samples_per_second": 76.169,
|
| 590 |
+
"eval_steps_per_second": 4.773,
|
| 591 |
+
"step": 854
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 8.06,
|
| 595 |
+
"learning_rate": 2.5506638714185887e-05,
|
| 596 |
+
"loss": 0.2842,
|
| 597 |
+
"step": 860
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 8.15,
|
| 601 |
+
"learning_rate": 2.5157232704402517e-05,
|
| 602 |
+
"loss": 0.2404,
|
| 603 |
+
"step": 870
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 8.24,
|
| 607 |
+
"learning_rate": 2.4807826694619148e-05,
|
| 608 |
+
"loss": 0.2525,
|
| 609 |
+
"step": 880
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 8.34,
|
| 613 |
+
"learning_rate": 2.445842068483578e-05,
|
| 614 |
+
"loss": 0.2415,
|
| 615 |
+
"step": 890
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 8.43,
|
| 619 |
+
"learning_rate": 2.4109014675052412e-05,
|
| 620 |
+
"loss": 0.222,
|
| 621 |
+
"step": 900
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 8.52,
|
| 625 |
+
"learning_rate": 2.3759608665269042e-05,
|
| 626 |
+
"loss": 0.2461,
|
| 627 |
+
"step": 910
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 8.62,
|
| 631 |
+
"learning_rate": 2.3410202655485676e-05,
|
| 632 |
+
"loss": 0.2533,
|
| 633 |
+
"step": 920
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 8.71,
|
| 637 |
+
"learning_rate": 2.3060796645702306e-05,
|
| 638 |
+
"loss": 0.267,
|
| 639 |
+
"step": 930
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 8.81,
|
| 643 |
+
"learning_rate": 2.271139063591894e-05,
|
| 644 |
+
"loss": 0.2572,
|
| 645 |
+
"step": 940
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 8.9,
|
| 649 |
+
"learning_rate": 2.236198462613557e-05,
|
| 650 |
+
"loss": 0.2263,
|
| 651 |
+
"step": 950
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 8.99,
|
| 655 |
+
"learning_rate": 2.2012578616352204e-05,
|
| 656 |
+
"loss": 0.3142,
|
| 657 |
+
"step": 960
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 8.99,
|
| 661 |
+
"eval_accuracy": 0.8349514563106796,
|
| 662 |
+
"eval_loss": 0.5270902514457703,
|
| 663 |
+
"eval_runtime": 30.327,
|
| 664 |
+
"eval_samples_per_second": 74.719,
|
| 665 |
+
"eval_steps_per_second": 4.682,
|
| 666 |
+
"step": 960
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 9.09,
|
| 670 |
+
"learning_rate": 2.1663172606568835e-05,
|
| 671 |
+
"loss": 0.2163,
|
| 672 |
+
"step": 970
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 9.18,
|
| 676 |
+
"learning_rate": 2.1313766596785465e-05,
|
| 677 |
+
"loss": 0.21,
|
| 678 |
+
"step": 980
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 9.27,
|
| 682 |
+
"learning_rate": 2.0964360587002095e-05,
|
| 683 |
+
"loss": 0.2354,
|
| 684 |
+
"step": 990
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 9.37,
|
| 688 |
+
"learning_rate": 2.061495457721873e-05,
|
| 689 |
+
"loss": 0.2394,
|
| 690 |
+
"step": 1000
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 9.46,
|
| 694 |
+
"learning_rate": 2.026554856743536e-05,
|
| 695 |
+
"loss": 0.2225,
|
| 696 |
+
"step": 1010
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 9.56,
|
| 700 |
+
"learning_rate": 1.9916142557651993e-05,
|
| 701 |
+
"loss": 0.2392,
|
| 702 |
+
"step": 1020
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 9.65,
|
| 706 |
+
"learning_rate": 1.9566736547868624e-05,
|
| 707 |
+
"loss": 0.2237,
|
| 708 |
+
"step": 1030
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 9.74,
|
| 712 |
+
"learning_rate": 1.9217330538085258e-05,
|
| 713 |
+
"loss": 0.251,
|
| 714 |
+
"step": 1040
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 9.84,
|
| 718 |
+
"learning_rate": 1.8867924528301888e-05,
|
| 719 |
+
"loss": 0.2033,
|
| 720 |
+
"step": 1050
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 9.93,
|
| 724 |
+
"learning_rate": 1.8518518518518518e-05,
|
| 725 |
+
"loss": 0.2266,
|
| 726 |
+
"step": 1060
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 10.0,
|
| 730 |
+
"eval_accuracy": 0.8362753751103266,
|
| 731 |
+
"eval_loss": 0.5450473427772522,
|
| 732 |
+
"eval_runtime": 29.6596,
|
| 733 |
+
"eval_samples_per_second": 76.4,
|
| 734 |
+
"eval_steps_per_second": 4.788,
|
| 735 |
+
"step": 1067
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 10.02,
|
| 739 |
+
"learning_rate": 1.8169112508735152e-05,
|
| 740 |
+
"loss": 0.211,
|
| 741 |
+
"step": 1070
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 10.12,
|
| 745 |
+
"learning_rate": 1.7819706498951782e-05,
|
| 746 |
+
"loss": 0.2449,
|
| 747 |
+
"step": 1080
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 10.21,
|
| 751 |
+
"learning_rate": 1.7470300489168416e-05,
|
| 752 |
+
"loss": 0.1846,
|
| 753 |
+
"step": 1090
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 10.3,
|
| 757 |
+
"learning_rate": 1.7120894479385047e-05,
|
| 758 |
+
"loss": 0.1763,
|
| 759 |
+
"step": 1100
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 10.4,
|
| 763 |
+
"learning_rate": 1.677148846960168e-05,
|
| 764 |
+
"loss": 0.2067,
|
| 765 |
+
"step": 1110
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 10.49,
|
| 769 |
+
"learning_rate": 1.642208245981831e-05,
|
| 770 |
+
"loss": 0.188,
|
| 771 |
+
"step": 1120
|
| 772 |
+
},
|
| 773 |
+
{
|
| 774 |
+
"epoch": 10.59,
|
| 775 |
+
"learning_rate": 1.607267645003494e-05,
|
| 776 |
+
"loss": 0.2359,
|
| 777 |
+
"step": 1130
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 10.68,
|
| 781 |
+
"learning_rate": 1.572327044025157e-05,
|
| 782 |
+
"loss": 0.187,
|
| 783 |
+
"step": 1140
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 10.77,
|
| 787 |
+
"learning_rate": 1.5373864430468205e-05,
|
| 788 |
+
"loss": 0.2369,
|
| 789 |
+
"step": 1150
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 10.87,
|
| 793 |
+
"learning_rate": 1.5024458420684836e-05,
|
| 794 |
+
"loss": 0.1856,
|
| 795 |
+
"step": 1160
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 10.96,
|
| 799 |
+
"learning_rate": 1.467505241090147e-05,
|
| 800 |
+
"loss": 0.1572,
|
| 801 |
+
"step": 1170
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 11.0,
|
| 805 |
+
"eval_accuracy": 0.8323036187113857,
|
| 806 |
+
"eval_loss": 0.5316830277442932,
|
| 807 |
+
"eval_runtime": 29.5637,
|
| 808 |
+
"eval_samples_per_second": 76.648,
|
| 809 |
+
"eval_steps_per_second": 4.803,
|
| 810 |
+
"step": 1174
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 11.05,
|
| 814 |
+
"learning_rate": 1.4325646401118102e-05,
|
| 815 |
+
"loss": 0.1944,
|
| 816 |
+
"step": 1180
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 11.15,
|
| 820 |
+
"learning_rate": 1.3976240391334734e-05,
|
| 821 |
+
"loss": 0.1841,
|
| 822 |
+
"step": 1190
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 11.24,
|
| 826 |
+
"learning_rate": 1.3626834381551362e-05,
|
| 827 |
+
"loss": 0.1675,
|
| 828 |
+
"step": 1200
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 11.33,
|
| 832 |
+
"learning_rate": 1.3277428371767994e-05,
|
| 833 |
+
"loss": 0.1881,
|
| 834 |
+
"step": 1210
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 11.43,
|
| 838 |
+
"learning_rate": 1.2928022361984627e-05,
|
| 839 |
+
"loss": 0.2145,
|
| 840 |
+
"step": 1220
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 11.52,
|
| 844 |
+
"learning_rate": 1.2578616352201259e-05,
|
| 845 |
+
"loss": 0.1618,
|
| 846 |
+
"step": 1230
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 11.62,
|
| 850 |
+
"learning_rate": 1.222921034241789e-05,
|
| 851 |
+
"loss": 0.189,
|
| 852 |
+
"step": 1240
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 11.71,
|
| 856 |
+
"learning_rate": 1.1879804332634521e-05,
|
| 857 |
+
"loss": 0.1915,
|
| 858 |
+
"step": 1250
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 11.8,
|
| 862 |
+
"learning_rate": 1.1530398322851153e-05,
|
| 863 |
+
"loss": 0.1697,
|
| 864 |
+
"step": 1260
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 11.9,
|
| 868 |
+
"learning_rate": 1.1180992313067785e-05,
|
| 869 |
+
"loss": 0.1793,
|
| 870 |
+
"step": 1270
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 11.99,
|
| 874 |
+
"learning_rate": 1.0831586303284417e-05,
|
| 875 |
+
"loss": 0.1891,
|
| 876 |
+
"step": 1280
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 12.0,
|
| 880 |
+
"eval_accuracy": 0.8362753751103266,
|
| 881 |
+
"eval_loss": 0.5269371867179871,
|
| 882 |
+
"eval_runtime": 29.7573,
|
| 883 |
+
"eval_samples_per_second": 76.149,
|
| 884 |
+
"eval_steps_per_second": 4.772,
|
| 885 |
+
"step": 1281
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 12.08,
|
| 889 |
+
"learning_rate": 1.0482180293501048e-05,
|
| 890 |
+
"loss": 0.1265,
|
| 891 |
+
"step": 1290
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 12.18,
|
| 895 |
+
"learning_rate": 1.013277428371768e-05,
|
| 896 |
+
"loss": 0.1668,
|
| 897 |
+
"step": 1300
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 12.27,
|
| 901 |
+
"learning_rate": 9.783368273934312e-06,
|
| 902 |
+
"loss": 0.179,
|
| 903 |
+
"step": 1310
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 12.37,
|
| 907 |
+
"learning_rate": 9.433962264150944e-06,
|
| 908 |
+
"loss": 0.176,
|
| 909 |
+
"step": 1320
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 12.46,
|
| 913 |
+
"learning_rate": 9.084556254367576e-06,
|
| 914 |
+
"loss": 0.1698,
|
| 915 |
+
"step": 1330
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 12.55,
|
| 919 |
+
"learning_rate": 8.735150244584208e-06,
|
| 920 |
+
"loss": 0.1563,
|
| 921 |
+
"step": 1340
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 12.65,
|
| 925 |
+
"learning_rate": 8.38574423480084e-06,
|
| 926 |
+
"loss": 0.1583,
|
| 927 |
+
"step": 1350
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 12.74,
|
| 931 |
+
"learning_rate": 8.03633822501747e-06,
|
| 932 |
+
"loss": 0.1185,
|
| 933 |
+
"step": 1360
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 12.83,
|
| 937 |
+
"learning_rate": 7.686932215234103e-06,
|
| 938 |
+
"loss": 0.1229,
|
| 939 |
+
"step": 1370
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 12.93,
|
| 943 |
+
"learning_rate": 7.337526205450735e-06,
|
| 944 |
+
"loss": 0.1551,
|
| 945 |
+
"step": 1380
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 12.99,
|
| 949 |
+
"eval_accuracy": 0.8375992939099736,
|
| 950 |
+
"eval_loss": 0.5339873433113098,
|
| 951 |
+
"eval_runtime": 29.8516,
|
| 952 |
+
"eval_samples_per_second": 75.909,
|
| 953 |
+
"eval_steps_per_second": 4.757,
|
| 954 |
+
"step": 1387
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 13.02,
|
| 958 |
+
"learning_rate": 6.988120195667367e-06,
|
| 959 |
+
"loss": 0.1566,
|
| 960 |
+
"step": 1390
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 13.11,
|
| 964 |
+
"learning_rate": 6.638714185883997e-06,
|
| 965 |
+
"loss": 0.1394,
|
| 966 |
+
"step": 1400
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 13.21,
|
| 970 |
+
"learning_rate": 6.289308176100629e-06,
|
| 971 |
+
"loss": 0.1642,
|
| 972 |
+
"step": 1410
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"epoch": 13.3,
|
| 976 |
+
"learning_rate": 5.9399021663172605e-06,
|
| 977 |
+
"loss": 0.1334,
|
| 978 |
+
"step": 1420
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 13.4,
|
| 982 |
+
"learning_rate": 5.590496156533893e-06,
|
| 983 |
+
"loss": 0.157,
|
| 984 |
+
"step": 1430
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 13.49,
|
| 988 |
+
"learning_rate": 5.241090146750524e-06,
|
| 989 |
+
"loss": 0.1658,
|
| 990 |
+
"step": 1440
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 13.58,
|
| 994 |
+
"learning_rate": 4.891684136967156e-06,
|
| 995 |
+
"loss": 0.1692,
|
| 996 |
+
"step": 1450
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 13.68,
|
| 1000 |
+
"learning_rate": 4.542278127183788e-06,
|
| 1001 |
+
"loss": 0.1393,
|
| 1002 |
+
"step": 1460
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 13.77,
|
| 1006 |
+
"learning_rate": 4.19287211740042e-06,
|
| 1007 |
+
"loss": 0.1492,
|
| 1008 |
+
"step": 1470
|
| 1009 |
+
},
|
| 1010 |
+
{
|
| 1011 |
+
"epoch": 13.86,
|
| 1012 |
+
"learning_rate": 3.843466107617051e-06,
|
| 1013 |
+
"loss": 0.1503,
|
| 1014 |
+
"step": 1480
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 13.96,
|
| 1018 |
+
"learning_rate": 3.4940600978336834e-06,
|
| 1019 |
+
"loss": 0.1548,
|
| 1020 |
+
"step": 1490
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 14.0,
|
| 1024 |
+
"eval_accuracy": 0.8367166813768756,
|
| 1025 |
+
"eval_loss": 0.5312865376472473,
|
| 1026 |
+
"eval_runtime": 29.7732,
|
| 1027 |
+
"eval_samples_per_second": 76.109,
|
| 1028 |
+
"eval_steps_per_second": 4.769,
|
| 1029 |
+
"step": 1494
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 14.05,
|
| 1033 |
+
"learning_rate": 3.1446540880503146e-06,
|
| 1034 |
+
"loss": 0.1186,
|
| 1035 |
+
"step": 1500
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 14.15,
|
| 1039 |
+
"learning_rate": 2.7952480782669463e-06,
|
| 1040 |
+
"loss": 0.1213,
|
| 1041 |
+
"step": 1510
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 14.24,
|
| 1045 |
+
"learning_rate": 2.445842068483578e-06,
|
| 1046 |
+
"loss": 0.1601,
|
| 1047 |
+
"step": 1520
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 14.33,
|
| 1051 |
+
"learning_rate": 2.09643605870021e-06,
|
| 1052 |
+
"loss": 0.1916,
|
| 1053 |
+
"step": 1530
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 14.43,
|
| 1057 |
+
"learning_rate": 1.7470300489168417e-06,
|
| 1058 |
+
"loss": 0.1243,
|
| 1059 |
+
"step": 1540
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 14.52,
|
| 1063 |
+
"learning_rate": 1.3976240391334732e-06,
|
| 1064 |
+
"loss": 0.1323,
|
| 1065 |
+
"step": 1550
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 14.61,
|
| 1069 |
+
"learning_rate": 1.048218029350105e-06,
|
| 1070 |
+
"loss": 0.1535,
|
| 1071 |
+
"step": 1560
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 14.71,
|
| 1075 |
+
"learning_rate": 6.988120195667366e-07,
|
| 1076 |
+
"loss": 0.1385,
|
| 1077 |
+
"step": 1570
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 14.8,
|
| 1081 |
+
"learning_rate": 3.494060097833683e-07,
|
| 1082 |
+
"loss": 0.1466,
|
| 1083 |
+
"step": 1580
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 14.89,
|
| 1087 |
+
"learning_rate": 0.0,
|
| 1088 |
+
"loss": 0.1781,
|
| 1089 |
+
"step": 1590
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 14.89,
|
| 1093 |
+
"eval_accuracy": 0.8398058252427184,
|
| 1094 |
+
"eval_loss": 0.5298482775688171,
|
| 1095 |
+
"eval_runtime": 29.7814,
|
| 1096 |
+
"eval_samples_per_second": 76.088,
|
| 1097 |
+
"eval_steps_per_second": 4.768,
|
| 1098 |
+
"step": 1590
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 14.89,
|
| 1102 |
+
"step": 1590,
|
| 1103 |
+
"total_flos": 7.877470546062803e+18,
|
| 1104 |
+
"train_loss": 0.5140632056590146,
|
| 1105 |
+
"train_runtime": 3982.2883,
|
| 1106 |
+
"train_samples_per_second": 25.704,
|
| 1107 |
+
"train_steps_per_second": 0.399
|
| 1108 |
+
}
|
| 1109 |
+
],
|
| 1110 |
+
"logging_steps": 10,
|
| 1111 |
+
"max_steps": 1590,
|
| 1112 |
+
"num_train_epochs": 15,
|
| 1113 |
+
"save_steps": 500,
|
| 1114 |
+
"total_flos": 7.877470546062803e+18,
|
| 1115 |
+
"trial_name": null,
|
| 1116 |
+
"trial_params": null
|
| 1117 |
+
}
|