Commit
•
3e48b92
1
Parent(s):
16ce9af
End of training
Browse files- all_results.json +11 -11
- eval_results.json +6 -6
- runs/May24_23-40-34_371088ce566e/events.out.tfevents.1716605763.371088ce566e.42.1 +3 -0
- train_results.json +6 -6
- trainer_state.json +1969 -13
all_results.json
CHANGED
|
@@ -1,13 +1,13 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_accuracy": 0.
|
| 4 |
-
"eval_loss": 0.
|
| 5 |
-
"eval_runtime":
|
| 6 |
-
"eval_samples_per_second":
|
| 7 |
-
"eval_steps_per_second": 3.
|
| 8 |
-
"total_flos":
|
| 9 |
-
"train_loss": 0.
|
| 10 |
-
"train_runtime":
|
| 11 |
-
"train_samples_per_second":
|
| 12 |
-
"train_steps_per_second": 0.
|
| 13 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"eval_accuracy": 0.8571428571428571,
|
| 4 |
+
"eval_loss": 0.4352613687515259,
|
| 5 |
+
"eval_runtime": 236.2313,
|
| 6 |
+
"eval_samples_per_second": 108.097,
|
| 7 |
+
"eval_steps_per_second": 3.378,
|
| 8 |
+
"total_flos": 7.915696500716863e+19,
|
| 9 |
+
"train_loss": 0.14746467811720712,
|
| 10 |
+
"train_runtime": 9966.943,
|
| 11 |
+
"train_samples_per_second": 102.483,
|
| 12 |
+
"train_steps_per_second": 0.801
|
| 13 |
}
|
eval_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_accuracy": 0.
|
| 4 |
-
"eval_loss": 0.
|
| 5 |
-
"eval_runtime":
|
| 6 |
-
"eval_samples_per_second":
|
| 7 |
-
"eval_steps_per_second": 3.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"eval_accuracy": 0.8571428571428571,
|
| 4 |
+
"eval_loss": 0.4352613687515259,
|
| 5 |
+
"eval_runtime": 236.2313,
|
| 6 |
+
"eval_samples_per_second": 108.097,
|
| 7 |
+
"eval_steps_per_second": 3.378
|
| 8 |
}
|
runs/May24_23-40-34_371088ce566e/events.out.tfevents.1716605763.371088ce566e.42.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:785bf90b2337263bd8b02e461767a421e64f712e790cc3ac3ae3b95edbf67a8f
|
| 3 |
+
size 411
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"total_flos":
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"total_flos": 7.915696500716863e+19,
|
| 4 |
+
"train_loss": 0.14746467811720712,
|
| 5 |
+
"train_runtime": 9966.943,
|
| 6 |
+
"train_samples_per_second": 102.483,
|
| 7 |
+
"train_steps_per_second": 0.801
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 0.
|
| 3 |
-
"best_model_checkpoint": "vit-base-patch16-224-in21k-finetuned/checkpoint-
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -2932,20 +2932,1976 @@
|
|
| 2932 |
},
|
| 2933 |
{
|
| 2934 |
"epoch": 6.0,
|
| 2935 |
-
"
|
| 2936 |
-
"
|
| 2937 |
-
"
|
| 2938 |
-
|
| 2939 |
-
|
| 2940 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2941 |
}
|
| 2942 |
],
|
| 2943 |
"logging_steps": 10,
|
| 2944 |
-
"max_steps":
|
| 2945 |
"num_input_tokens_seen": 0,
|
| 2946 |
-
"num_train_epochs":
|
| 2947 |
"save_steps": 500,
|
| 2948 |
-
"total_flos":
|
| 2949 |
"train_batch_size": 32,
|
| 2950 |
"trial_name": null,
|
| 2951 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.8571428571428571,
|
| 3 |
+
"best_model_checkpoint": "vit-base-patch16-224-in21k-finetuned/checkpoint-7980",
|
| 4 |
+
"epoch": 10.0,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 7980,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 2932 |
},
|
| 2933 |
{
|
| 2934 |
"epoch": 6.0,
|
| 2935 |
+
"learning_rate": 2.2208298524088e-05,
|
| 2936 |
+
"loss": 0.4739,
|
| 2937 |
+
"step": 4790
|
| 2938 |
+
},
|
| 2939 |
+
{
|
| 2940 |
+
"epoch": 6.02,
|
| 2941 |
+
"learning_rate": 2.2138680033416877e-05,
|
| 2942 |
+
"loss": 0.4398,
|
| 2943 |
+
"step": 4800
|
| 2944 |
+
},
|
| 2945 |
+
{
|
| 2946 |
+
"epoch": 6.03,
|
| 2947 |
+
"learning_rate": 2.2069061542745755e-05,
|
| 2948 |
+
"loss": 0.4024,
|
| 2949 |
+
"step": 4810
|
| 2950 |
+
},
|
| 2951 |
+
{
|
| 2952 |
+
"epoch": 6.04,
|
| 2953 |
+
"learning_rate": 2.199944305207463e-05,
|
| 2954 |
+
"loss": 0.4506,
|
| 2955 |
+
"step": 4820
|
| 2956 |
+
},
|
| 2957 |
+
{
|
| 2958 |
+
"epoch": 6.05,
|
| 2959 |
+
"learning_rate": 2.1929824561403507e-05,
|
| 2960 |
+
"loss": 0.5156,
|
| 2961 |
+
"step": 4830
|
| 2962 |
+
},
|
| 2963 |
+
{
|
| 2964 |
+
"epoch": 6.07,
|
| 2965 |
+
"learning_rate": 2.1860206070732385e-05,
|
| 2966 |
+
"loss": 0.5007,
|
| 2967 |
+
"step": 4840
|
| 2968 |
+
},
|
| 2969 |
+
{
|
| 2970 |
+
"epoch": 6.08,
|
| 2971 |
+
"learning_rate": 2.1790587580061263e-05,
|
| 2972 |
+
"loss": 0.4959,
|
| 2973 |
+
"step": 4850
|
| 2974 |
+
},
|
| 2975 |
+
{
|
| 2976 |
+
"epoch": 6.09,
|
| 2977 |
+
"learning_rate": 2.172096908939014e-05,
|
| 2978 |
+
"loss": 0.5087,
|
| 2979 |
+
"step": 4860
|
| 2980 |
+
},
|
| 2981 |
+
{
|
| 2982 |
+
"epoch": 6.1,
|
| 2983 |
+
"learning_rate": 2.165135059871902e-05,
|
| 2984 |
+
"loss": 0.4577,
|
| 2985 |
+
"step": 4870
|
| 2986 |
+
},
|
| 2987 |
+
{
|
| 2988 |
+
"epoch": 6.12,
|
| 2989 |
+
"learning_rate": 2.1581732108047897e-05,
|
| 2990 |
+
"loss": 0.4851,
|
| 2991 |
+
"step": 4880
|
| 2992 |
+
},
|
| 2993 |
+
{
|
| 2994 |
+
"epoch": 6.13,
|
| 2995 |
+
"learning_rate": 2.1512113617376775e-05,
|
| 2996 |
+
"loss": 0.4541,
|
| 2997 |
+
"step": 4890
|
| 2998 |
+
},
|
| 2999 |
+
{
|
| 3000 |
+
"epoch": 6.14,
|
| 3001 |
+
"learning_rate": 2.1442495126705653e-05,
|
| 3002 |
+
"loss": 0.5341,
|
| 3003 |
+
"step": 4900
|
| 3004 |
+
},
|
| 3005 |
+
{
|
| 3006 |
+
"epoch": 6.15,
|
| 3007 |
+
"learning_rate": 2.137287663603453e-05,
|
| 3008 |
+
"loss": 0.4609,
|
| 3009 |
+
"step": 4910
|
| 3010 |
+
},
|
| 3011 |
+
{
|
| 3012 |
+
"epoch": 6.17,
|
| 3013 |
+
"learning_rate": 2.130325814536341e-05,
|
| 3014 |
+
"loss": 0.5178,
|
| 3015 |
+
"step": 4920
|
| 3016 |
+
},
|
| 3017 |
+
{
|
| 3018 |
+
"epoch": 6.18,
|
| 3019 |
+
"learning_rate": 2.1233639654692286e-05,
|
| 3020 |
+
"loss": 0.4792,
|
| 3021 |
+
"step": 4930
|
| 3022 |
+
},
|
| 3023 |
+
{
|
| 3024 |
+
"epoch": 6.19,
|
| 3025 |
+
"learning_rate": 2.1164021164021164e-05,
|
| 3026 |
+
"loss": 0.448,
|
| 3027 |
+
"step": 4940
|
| 3028 |
+
},
|
| 3029 |
+
{
|
| 3030 |
+
"epoch": 6.2,
|
| 3031 |
+
"learning_rate": 2.1094402673350042e-05,
|
| 3032 |
+
"loss": 0.4971,
|
| 3033 |
+
"step": 4950
|
| 3034 |
+
},
|
| 3035 |
+
{
|
| 3036 |
+
"epoch": 6.22,
|
| 3037 |
+
"learning_rate": 2.102478418267892e-05,
|
| 3038 |
+
"loss": 0.4521,
|
| 3039 |
+
"step": 4960
|
| 3040 |
+
},
|
| 3041 |
+
{
|
| 3042 |
+
"epoch": 6.23,
|
| 3043 |
+
"learning_rate": 2.0955165692007798e-05,
|
| 3044 |
+
"loss": 0.4584,
|
| 3045 |
+
"step": 4970
|
| 3046 |
+
},
|
| 3047 |
+
{
|
| 3048 |
+
"epoch": 6.24,
|
| 3049 |
+
"learning_rate": 2.0885547201336676e-05,
|
| 3050 |
+
"loss": 0.4984,
|
| 3051 |
+
"step": 4980
|
| 3052 |
+
},
|
| 3053 |
+
{
|
| 3054 |
+
"epoch": 6.25,
|
| 3055 |
+
"learning_rate": 2.0815928710665554e-05,
|
| 3056 |
+
"loss": 0.4894,
|
| 3057 |
+
"step": 4990
|
| 3058 |
+
},
|
| 3059 |
+
{
|
| 3060 |
+
"epoch": 6.27,
|
| 3061 |
+
"learning_rate": 2.0746310219994432e-05,
|
| 3062 |
+
"loss": 0.4887,
|
| 3063 |
+
"step": 5000
|
| 3064 |
+
},
|
| 3065 |
+
{
|
| 3066 |
+
"epoch": 6.28,
|
| 3067 |
+
"learning_rate": 2.067669172932331e-05,
|
| 3068 |
+
"loss": 0.5063,
|
| 3069 |
+
"step": 5010
|
| 3070 |
+
},
|
| 3071 |
+
{
|
| 3072 |
+
"epoch": 6.29,
|
| 3073 |
+
"learning_rate": 2.0607073238652184e-05,
|
| 3074 |
+
"loss": 0.4963,
|
| 3075 |
+
"step": 5020
|
| 3076 |
+
},
|
| 3077 |
+
{
|
| 3078 |
+
"epoch": 6.3,
|
| 3079 |
+
"learning_rate": 2.0537454747981062e-05,
|
| 3080 |
+
"loss": 0.524,
|
| 3081 |
+
"step": 5030
|
| 3082 |
+
},
|
| 3083 |
+
{
|
| 3084 |
+
"epoch": 6.32,
|
| 3085 |
+
"learning_rate": 2.046783625730994e-05,
|
| 3086 |
+
"loss": 0.4502,
|
| 3087 |
+
"step": 5040
|
| 3088 |
+
},
|
| 3089 |
+
{
|
| 3090 |
+
"epoch": 6.33,
|
| 3091 |
+
"learning_rate": 2.0398217766638818e-05,
|
| 3092 |
+
"loss": 0.4505,
|
| 3093 |
+
"step": 5050
|
| 3094 |
+
},
|
| 3095 |
+
{
|
| 3096 |
+
"epoch": 6.34,
|
| 3097 |
+
"learning_rate": 2.0328599275967696e-05,
|
| 3098 |
+
"loss": 0.4962,
|
| 3099 |
+
"step": 5060
|
| 3100 |
+
},
|
| 3101 |
+
{
|
| 3102 |
+
"epoch": 6.35,
|
| 3103 |
+
"learning_rate": 2.0258980785296574e-05,
|
| 3104 |
+
"loss": 0.4566,
|
| 3105 |
+
"step": 5070
|
| 3106 |
+
},
|
| 3107 |
+
{
|
| 3108 |
+
"epoch": 6.37,
|
| 3109 |
+
"learning_rate": 2.018936229462545e-05,
|
| 3110 |
+
"loss": 0.4464,
|
| 3111 |
+
"step": 5080
|
| 3112 |
+
},
|
| 3113 |
+
{
|
| 3114 |
+
"epoch": 6.38,
|
| 3115 |
+
"learning_rate": 2.011974380395433e-05,
|
| 3116 |
+
"loss": 0.4872,
|
| 3117 |
+
"step": 5090
|
| 3118 |
+
},
|
| 3119 |
+
{
|
| 3120 |
+
"epoch": 6.39,
|
| 3121 |
+
"learning_rate": 2.0050125313283208e-05,
|
| 3122 |
+
"loss": 0.4396,
|
| 3123 |
+
"step": 5100
|
| 3124 |
+
},
|
| 3125 |
+
{
|
| 3126 |
+
"epoch": 6.4,
|
| 3127 |
+
"learning_rate": 1.9980506822612085e-05,
|
| 3128 |
+
"loss": 0.5302,
|
| 3129 |
+
"step": 5110
|
| 3130 |
+
},
|
| 3131 |
+
{
|
| 3132 |
+
"epoch": 6.42,
|
| 3133 |
+
"learning_rate": 1.9910888331940963e-05,
|
| 3134 |
+
"loss": 0.4927,
|
| 3135 |
+
"step": 5120
|
| 3136 |
+
},
|
| 3137 |
+
{
|
| 3138 |
+
"epoch": 6.43,
|
| 3139 |
+
"learning_rate": 1.984126984126984e-05,
|
| 3140 |
+
"loss": 0.5129,
|
| 3141 |
+
"step": 5130
|
| 3142 |
+
},
|
| 3143 |
+
{
|
| 3144 |
+
"epoch": 6.44,
|
| 3145 |
+
"learning_rate": 1.977165135059872e-05,
|
| 3146 |
+
"loss": 0.4747,
|
| 3147 |
+
"step": 5140
|
| 3148 |
+
},
|
| 3149 |
+
{
|
| 3150 |
+
"epoch": 6.45,
|
| 3151 |
+
"learning_rate": 1.9702032859927597e-05,
|
| 3152 |
+
"loss": 0.4878,
|
| 3153 |
+
"step": 5150
|
| 3154 |
+
},
|
| 3155 |
+
{
|
| 3156 |
+
"epoch": 6.47,
|
| 3157 |
+
"learning_rate": 1.9632414369256475e-05,
|
| 3158 |
+
"loss": 0.4885,
|
| 3159 |
+
"step": 5160
|
| 3160 |
+
},
|
| 3161 |
+
{
|
| 3162 |
+
"epoch": 6.48,
|
| 3163 |
+
"learning_rate": 1.9562795878585353e-05,
|
| 3164 |
+
"loss": 0.4634,
|
| 3165 |
+
"step": 5170
|
| 3166 |
+
},
|
| 3167 |
+
{
|
| 3168 |
+
"epoch": 6.49,
|
| 3169 |
+
"learning_rate": 1.949317738791423e-05,
|
| 3170 |
+
"loss": 0.4665,
|
| 3171 |
+
"step": 5180
|
| 3172 |
+
},
|
| 3173 |
+
{
|
| 3174 |
+
"epoch": 6.5,
|
| 3175 |
+
"learning_rate": 1.942355889724311e-05,
|
| 3176 |
+
"loss": 0.4392,
|
| 3177 |
+
"step": 5190
|
| 3178 |
+
},
|
| 3179 |
+
{
|
| 3180 |
+
"epoch": 6.52,
|
| 3181 |
+
"learning_rate": 1.9353940406571987e-05,
|
| 3182 |
+
"loss": 0.4754,
|
| 3183 |
+
"step": 5200
|
| 3184 |
+
},
|
| 3185 |
+
{
|
| 3186 |
+
"epoch": 6.53,
|
| 3187 |
+
"learning_rate": 1.9284321915900865e-05,
|
| 3188 |
+
"loss": 0.457,
|
| 3189 |
+
"step": 5210
|
| 3190 |
+
},
|
| 3191 |
+
{
|
| 3192 |
+
"epoch": 6.54,
|
| 3193 |
+
"learning_rate": 1.9214703425229743e-05,
|
| 3194 |
+
"loss": 0.4734,
|
| 3195 |
+
"step": 5220
|
| 3196 |
+
},
|
| 3197 |
+
{
|
| 3198 |
+
"epoch": 6.55,
|
| 3199 |
+
"learning_rate": 1.9145084934558617e-05,
|
| 3200 |
+
"loss": 0.4556,
|
| 3201 |
+
"step": 5230
|
| 3202 |
+
},
|
| 3203 |
+
{
|
| 3204 |
+
"epoch": 6.57,
|
| 3205 |
+
"learning_rate": 1.9075466443887495e-05,
|
| 3206 |
+
"loss": 0.4844,
|
| 3207 |
+
"step": 5240
|
| 3208 |
+
},
|
| 3209 |
+
{
|
| 3210 |
+
"epoch": 6.58,
|
| 3211 |
+
"learning_rate": 1.9005847953216373e-05,
|
| 3212 |
+
"loss": 0.4161,
|
| 3213 |
+
"step": 5250
|
| 3214 |
+
},
|
| 3215 |
+
{
|
| 3216 |
+
"epoch": 6.59,
|
| 3217 |
+
"learning_rate": 1.893622946254525e-05,
|
| 3218 |
+
"loss": 0.4722,
|
| 3219 |
+
"step": 5260
|
| 3220 |
+
},
|
| 3221 |
+
{
|
| 3222 |
+
"epoch": 6.6,
|
| 3223 |
+
"learning_rate": 1.886661097187413e-05,
|
| 3224 |
+
"loss": 0.517,
|
| 3225 |
+
"step": 5270
|
| 3226 |
+
},
|
| 3227 |
+
{
|
| 3228 |
+
"epoch": 6.62,
|
| 3229 |
+
"learning_rate": 1.8796992481203007e-05,
|
| 3230 |
+
"loss": 0.4612,
|
| 3231 |
+
"step": 5280
|
| 3232 |
+
},
|
| 3233 |
+
{
|
| 3234 |
+
"epoch": 6.63,
|
| 3235 |
+
"learning_rate": 1.8727373990531885e-05,
|
| 3236 |
+
"loss": 0.5017,
|
| 3237 |
+
"step": 5290
|
| 3238 |
+
},
|
| 3239 |
+
{
|
| 3240 |
+
"epoch": 6.64,
|
| 3241 |
+
"learning_rate": 1.8657755499860762e-05,
|
| 3242 |
+
"loss": 0.4882,
|
| 3243 |
+
"step": 5300
|
| 3244 |
+
},
|
| 3245 |
+
{
|
| 3246 |
+
"epoch": 6.65,
|
| 3247 |
+
"learning_rate": 1.858813700918964e-05,
|
| 3248 |
+
"loss": 0.4722,
|
| 3249 |
+
"step": 5310
|
| 3250 |
+
},
|
| 3251 |
+
{
|
| 3252 |
+
"epoch": 6.67,
|
| 3253 |
+
"learning_rate": 1.8518518518518518e-05,
|
| 3254 |
+
"loss": 0.4674,
|
| 3255 |
+
"step": 5320
|
| 3256 |
+
},
|
| 3257 |
+
{
|
| 3258 |
+
"epoch": 6.68,
|
| 3259 |
+
"learning_rate": 1.8448900027847396e-05,
|
| 3260 |
+
"loss": 0.4928,
|
| 3261 |
+
"step": 5330
|
| 3262 |
+
},
|
| 3263 |
+
{
|
| 3264 |
+
"epoch": 6.69,
|
| 3265 |
+
"learning_rate": 1.8379281537176274e-05,
|
| 3266 |
+
"loss": 0.4426,
|
| 3267 |
+
"step": 5340
|
| 3268 |
+
},
|
| 3269 |
+
{
|
| 3270 |
+
"epoch": 6.7,
|
| 3271 |
+
"learning_rate": 1.8309663046505152e-05,
|
| 3272 |
+
"loss": 0.4778,
|
| 3273 |
+
"step": 5350
|
| 3274 |
+
},
|
| 3275 |
+
{
|
| 3276 |
+
"epoch": 6.72,
|
| 3277 |
+
"learning_rate": 1.824004455583403e-05,
|
| 3278 |
+
"loss": 0.4555,
|
| 3279 |
+
"step": 5360
|
| 3280 |
+
},
|
| 3281 |
+
{
|
| 3282 |
+
"epoch": 6.73,
|
| 3283 |
+
"learning_rate": 1.8170426065162908e-05,
|
| 3284 |
+
"loss": 0.4707,
|
| 3285 |
+
"step": 5370
|
| 3286 |
+
},
|
| 3287 |
+
{
|
| 3288 |
+
"epoch": 6.74,
|
| 3289 |
+
"learning_rate": 1.8100807574491786e-05,
|
| 3290 |
+
"loss": 0.4729,
|
| 3291 |
+
"step": 5380
|
| 3292 |
+
},
|
| 3293 |
+
{
|
| 3294 |
+
"epoch": 6.75,
|
| 3295 |
+
"learning_rate": 1.8031189083820664e-05,
|
| 3296 |
+
"loss": 0.5236,
|
| 3297 |
+
"step": 5390
|
| 3298 |
+
},
|
| 3299 |
+
{
|
| 3300 |
+
"epoch": 6.77,
|
| 3301 |
+
"learning_rate": 1.796157059314954e-05,
|
| 3302 |
+
"loss": 0.4635,
|
| 3303 |
+
"step": 5400
|
| 3304 |
+
},
|
| 3305 |
+
{
|
| 3306 |
+
"epoch": 6.78,
|
| 3307 |
+
"learning_rate": 1.789195210247842e-05,
|
| 3308 |
+
"loss": 0.4731,
|
| 3309 |
+
"step": 5410
|
| 3310 |
+
},
|
| 3311 |
+
{
|
| 3312 |
+
"epoch": 6.79,
|
| 3313 |
+
"learning_rate": 1.7822333611807297e-05,
|
| 3314 |
+
"loss": 0.4784,
|
| 3315 |
+
"step": 5420
|
| 3316 |
+
},
|
| 3317 |
+
{
|
| 3318 |
+
"epoch": 6.8,
|
| 3319 |
+
"learning_rate": 1.7752715121136172e-05,
|
| 3320 |
+
"loss": 0.4913,
|
| 3321 |
+
"step": 5430
|
| 3322 |
+
},
|
| 3323 |
+
{
|
| 3324 |
+
"epoch": 6.82,
|
| 3325 |
+
"learning_rate": 1.768309663046505e-05,
|
| 3326 |
+
"loss": 0.4617,
|
| 3327 |
+
"step": 5440
|
| 3328 |
+
},
|
| 3329 |
+
{
|
| 3330 |
+
"epoch": 6.83,
|
| 3331 |
+
"learning_rate": 1.7613478139793928e-05,
|
| 3332 |
+
"loss": 0.4888,
|
| 3333 |
+
"step": 5450
|
| 3334 |
+
},
|
| 3335 |
+
{
|
| 3336 |
+
"epoch": 6.84,
|
| 3337 |
+
"learning_rate": 1.7543859649122806e-05,
|
| 3338 |
+
"loss": 0.4976,
|
| 3339 |
+
"step": 5460
|
| 3340 |
+
},
|
| 3341 |
+
{
|
| 3342 |
+
"epoch": 6.85,
|
| 3343 |
+
"learning_rate": 1.7474241158451684e-05,
|
| 3344 |
+
"loss": 0.4778,
|
| 3345 |
+
"step": 5470
|
| 3346 |
+
},
|
| 3347 |
+
{
|
| 3348 |
+
"epoch": 6.87,
|
| 3349 |
+
"learning_rate": 1.740462266778056e-05,
|
| 3350 |
+
"loss": 0.4818,
|
| 3351 |
+
"step": 5480
|
| 3352 |
+
},
|
| 3353 |
+
{
|
| 3354 |
+
"epoch": 6.88,
|
| 3355 |
+
"learning_rate": 1.733500417710944e-05,
|
| 3356 |
+
"loss": 0.5,
|
| 3357 |
+
"step": 5490
|
| 3358 |
+
},
|
| 3359 |
+
{
|
| 3360 |
+
"epoch": 6.89,
|
| 3361 |
+
"learning_rate": 1.7265385686438317e-05,
|
| 3362 |
+
"loss": 0.4529,
|
| 3363 |
+
"step": 5500
|
| 3364 |
+
},
|
| 3365 |
+
{
|
| 3366 |
+
"epoch": 6.9,
|
| 3367 |
+
"learning_rate": 1.7195767195767195e-05,
|
| 3368 |
+
"loss": 0.4674,
|
| 3369 |
+
"step": 5510
|
| 3370 |
+
},
|
| 3371 |
+
{
|
| 3372 |
+
"epoch": 6.92,
|
| 3373 |
+
"learning_rate": 1.7126148705096073e-05,
|
| 3374 |
+
"loss": 0.4733,
|
| 3375 |
+
"step": 5520
|
| 3376 |
+
},
|
| 3377 |
+
{
|
| 3378 |
+
"epoch": 6.93,
|
| 3379 |
+
"learning_rate": 1.705653021442495e-05,
|
| 3380 |
+
"loss": 0.4702,
|
| 3381 |
+
"step": 5530
|
| 3382 |
+
},
|
| 3383 |
+
{
|
| 3384 |
+
"epoch": 6.94,
|
| 3385 |
+
"learning_rate": 1.698691172375383e-05,
|
| 3386 |
+
"loss": 0.4809,
|
| 3387 |
+
"step": 5540
|
| 3388 |
+
},
|
| 3389 |
+
{
|
| 3390 |
+
"epoch": 6.95,
|
| 3391 |
+
"learning_rate": 1.6917293233082707e-05,
|
| 3392 |
+
"loss": 0.509,
|
| 3393 |
+
"step": 5550
|
| 3394 |
+
},
|
| 3395 |
+
{
|
| 3396 |
+
"epoch": 6.97,
|
| 3397 |
+
"learning_rate": 1.6847674742411585e-05,
|
| 3398 |
+
"loss": 0.4692,
|
| 3399 |
+
"step": 5560
|
| 3400 |
+
},
|
| 3401 |
+
{
|
| 3402 |
+
"epoch": 6.98,
|
| 3403 |
+
"learning_rate": 1.6778056251740463e-05,
|
| 3404 |
+
"loss": 0.5013,
|
| 3405 |
+
"step": 5570
|
| 3406 |
+
},
|
| 3407 |
+
{
|
| 3408 |
+
"epoch": 6.99,
|
| 3409 |
+
"learning_rate": 1.670843776106934e-05,
|
| 3410 |
+
"loss": 0.4245,
|
| 3411 |
+
"step": 5580
|
| 3412 |
+
},
|
| 3413 |
+
{
|
| 3414 |
+
"epoch": 7.0,
|
| 3415 |
+
"eval_accuracy": 0.8541666666666666,
|
| 3416 |
+
"eval_loss": 0.43816250562667847,
|
| 3417 |
+
"eval_runtime": 233.7655,
|
| 3418 |
+
"eval_samples_per_second": 109.238,
|
| 3419 |
+
"eval_steps_per_second": 3.414,
|
| 3420 |
+
"step": 5586
|
| 3421 |
+
},
|
| 3422 |
+
{
|
| 3423 |
+
"epoch": 7.01,
|
| 3424 |
+
"learning_rate": 1.663881927039822e-05,
|
| 3425 |
+
"loss": 0.4232,
|
| 3426 |
+
"step": 5590
|
| 3427 |
+
},
|
| 3428 |
+
{
|
| 3429 |
+
"epoch": 7.02,
|
| 3430 |
+
"learning_rate": 1.6569200779727097e-05,
|
| 3431 |
+
"loss": 0.4133,
|
| 3432 |
+
"step": 5600
|
| 3433 |
+
},
|
| 3434 |
+
{
|
| 3435 |
+
"epoch": 7.03,
|
| 3436 |
+
"learning_rate": 1.6499582289055974e-05,
|
| 3437 |
+
"loss": 0.4066,
|
| 3438 |
+
"step": 5610
|
| 3439 |
+
},
|
| 3440 |
+
{
|
| 3441 |
+
"epoch": 7.04,
|
| 3442 |
+
"learning_rate": 1.6429963798384852e-05,
|
| 3443 |
+
"loss": 0.4381,
|
| 3444 |
+
"step": 5620
|
| 3445 |
+
},
|
| 3446 |
+
{
|
| 3447 |
+
"epoch": 7.06,
|
| 3448 |
+
"learning_rate": 1.636034530771373e-05,
|
| 3449 |
+
"loss": 0.4063,
|
| 3450 |
+
"step": 5630
|
| 3451 |
+
},
|
| 3452 |
+
{
|
| 3453 |
+
"epoch": 7.07,
|
| 3454 |
+
"learning_rate": 1.6290726817042605e-05,
|
| 3455 |
+
"loss": 0.3721,
|
| 3456 |
+
"step": 5640
|
| 3457 |
+
},
|
| 3458 |
+
{
|
| 3459 |
+
"epoch": 7.08,
|
| 3460 |
+
"learning_rate": 1.6221108326371483e-05,
|
| 3461 |
+
"loss": 0.3574,
|
| 3462 |
+
"step": 5650
|
| 3463 |
+
},
|
| 3464 |
+
{
|
| 3465 |
+
"epoch": 7.09,
|
| 3466 |
+
"learning_rate": 1.615148983570036e-05,
|
| 3467 |
+
"loss": 0.4156,
|
| 3468 |
+
"step": 5660
|
| 3469 |
+
},
|
| 3470 |
+
{
|
| 3471 |
+
"epoch": 7.11,
|
| 3472 |
+
"learning_rate": 1.608187134502924e-05,
|
| 3473 |
+
"loss": 0.4311,
|
| 3474 |
+
"step": 5670
|
| 3475 |
+
},
|
| 3476 |
+
{
|
| 3477 |
+
"epoch": 7.12,
|
| 3478 |
+
"learning_rate": 1.6012252854358116e-05,
|
| 3479 |
+
"loss": 0.3922,
|
| 3480 |
+
"step": 5680
|
| 3481 |
+
},
|
| 3482 |
+
{
|
| 3483 |
+
"epoch": 7.13,
|
| 3484 |
+
"learning_rate": 1.5942634363686994e-05,
|
| 3485 |
+
"loss": 0.3568,
|
| 3486 |
+
"step": 5690
|
| 3487 |
+
},
|
| 3488 |
+
{
|
| 3489 |
+
"epoch": 7.14,
|
| 3490 |
+
"learning_rate": 1.5873015873015872e-05,
|
| 3491 |
+
"loss": 0.3943,
|
| 3492 |
+
"step": 5700
|
| 3493 |
+
},
|
| 3494 |
+
{
|
| 3495 |
+
"epoch": 7.16,
|
| 3496 |
+
"learning_rate": 1.580339738234475e-05,
|
| 3497 |
+
"loss": 0.3995,
|
| 3498 |
+
"step": 5710
|
| 3499 |
+
},
|
| 3500 |
+
{
|
| 3501 |
+
"epoch": 7.17,
|
| 3502 |
+
"learning_rate": 1.5733778891673628e-05,
|
| 3503 |
+
"loss": 0.3948,
|
| 3504 |
+
"step": 5720
|
| 3505 |
+
},
|
| 3506 |
+
{
|
| 3507 |
+
"epoch": 7.18,
|
| 3508 |
+
"learning_rate": 1.5664160401002506e-05,
|
| 3509 |
+
"loss": 0.3912,
|
| 3510 |
+
"step": 5730
|
| 3511 |
+
},
|
| 3512 |
+
{
|
| 3513 |
+
"epoch": 7.19,
|
| 3514 |
+
"learning_rate": 1.5594541910331384e-05,
|
| 3515 |
+
"loss": 0.3791,
|
| 3516 |
+
"step": 5740
|
| 3517 |
+
},
|
| 3518 |
+
{
|
| 3519 |
+
"epoch": 7.21,
|
| 3520 |
+
"learning_rate": 1.5524923419660262e-05,
|
| 3521 |
+
"loss": 0.4085,
|
| 3522 |
+
"step": 5750
|
| 3523 |
+
},
|
| 3524 |
+
{
|
| 3525 |
+
"epoch": 7.22,
|
| 3526 |
+
"learning_rate": 1.545530492898914e-05,
|
| 3527 |
+
"loss": 0.4175,
|
| 3528 |
+
"step": 5760
|
| 3529 |
+
},
|
| 3530 |
+
{
|
| 3531 |
+
"epoch": 7.23,
|
| 3532 |
+
"learning_rate": 1.5385686438318018e-05,
|
| 3533 |
+
"loss": 0.4018,
|
| 3534 |
+
"step": 5770
|
| 3535 |
+
},
|
| 3536 |
+
{
|
| 3537 |
+
"epoch": 7.24,
|
| 3538 |
+
"learning_rate": 1.5316067947646896e-05,
|
| 3539 |
+
"loss": 0.4203,
|
| 3540 |
+
"step": 5780
|
| 3541 |
+
},
|
| 3542 |
+
{
|
| 3543 |
+
"epoch": 7.26,
|
| 3544 |
+
"learning_rate": 1.5246449456975772e-05,
|
| 3545 |
+
"loss": 0.3939,
|
| 3546 |
+
"step": 5790
|
| 3547 |
+
},
|
| 3548 |
+
{
|
| 3549 |
+
"epoch": 7.27,
|
| 3550 |
+
"learning_rate": 1.517683096630465e-05,
|
| 3551 |
+
"loss": 0.4417,
|
| 3552 |
+
"step": 5800
|
| 3553 |
+
},
|
| 3554 |
+
{
|
| 3555 |
+
"epoch": 7.28,
|
| 3556 |
+
"learning_rate": 1.5107212475633528e-05,
|
| 3557 |
+
"loss": 0.3599,
|
| 3558 |
+
"step": 5810
|
| 3559 |
+
},
|
| 3560 |
+
{
|
| 3561 |
+
"epoch": 7.29,
|
| 3562 |
+
"learning_rate": 1.5037593984962406e-05,
|
| 3563 |
+
"loss": 0.3877,
|
| 3564 |
+
"step": 5820
|
| 3565 |
+
},
|
| 3566 |
+
{
|
| 3567 |
+
"epoch": 7.31,
|
| 3568 |
+
"learning_rate": 1.4967975494291284e-05,
|
| 3569 |
+
"loss": 0.4054,
|
| 3570 |
+
"step": 5830
|
| 3571 |
+
},
|
| 3572 |
+
{
|
| 3573 |
+
"epoch": 7.32,
|
| 3574 |
+
"learning_rate": 1.4898357003620161e-05,
|
| 3575 |
+
"loss": 0.3935,
|
| 3576 |
+
"step": 5840
|
| 3577 |
+
},
|
| 3578 |
+
{
|
| 3579 |
+
"epoch": 7.33,
|
| 3580 |
+
"learning_rate": 1.482873851294904e-05,
|
| 3581 |
+
"loss": 0.3589,
|
| 3582 |
+
"step": 5850
|
| 3583 |
+
},
|
| 3584 |
+
{
|
| 3585 |
+
"epoch": 7.34,
|
| 3586 |
+
"learning_rate": 1.4759120022277917e-05,
|
| 3587 |
+
"loss": 0.3815,
|
| 3588 |
+
"step": 5860
|
| 3589 |
+
},
|
| 3590 |
+
{
|
| 3591 |
+
"epoch": 7.36,
|
| 3592 |
+
"learning_rate": 1.4689501531606795e-05,
|
| 3593 |
+
"loss": 0.4185,
|
| 3594 |
+
"step": 5870
|
| 3595 |
+
},
|
| 3596 |
+
{
|
| 3597 |
+
"epoch": 7.37,
|
| 3598 |
+
"learning_rate": 1.4619883040935673e-05,
|
| 3599 |
+
"loss": 0.4318,
|
| 3600 |
+
"step": 5880
|
| 3601 |
+
},
|
| 3602 |
+
{
|
| 3603 |
+
"epoch": 7.38,
|
| 3604 |
+
"learning_rate": 1.455026455026455e-05,
|
| 3605 |
+
"loss": 0.3831,
|
| 3606 |
+
"step": 5890
|
| 3607 |
+
},
|
| 3608 |
+
{
|
| 3609 |
+
"epoch": 7.39,
|
| 3610 |
+
"learning_rate": 1.4480646059593427e-05,
|
| 3611 |
+
"loss": 0.3999,
|
| 3612 |
+
"step": 5900
|
| 3613 |
+
},
|
| 3614 |
+
{
|
| 3615 |
+
"epoch": 7.41,
|
| 3616 |
+
"learning_rate": 1.4411027568922305e-05,
|
| 3617 |
+
"loss": 0.4112,
|
| 3618 |
+
"step": 5910
|
| 3619 |
+
},
|
| 3620 |
+
{
|
| 3621 |
+
"epoch": 7.42,
|
| 3622 |
+
"learning_rate": 1.4341409078251183e-05,
|
| 3623 |
+
"loss": 0.4429,
|
| 3624 |
+
"step": 5920
|
| 3625 |
+
},
|
| 3626 |
+
{
|
| 3627 |
+
"epoch": 7.43,
|
| 3628 |
+
"learning_rate": 1.4271790587580061e-05,
|
| 3629 |
+
"loss": 0.4023,
|
| 3630 |
+
"step": 5930
|
| 3631 |
+
},
|
| 3632 |
+
{
|
| 3633 |
+
"epoch": 7.44,
|
| 3634 |
+
"learning_rate": 1.4202172096908939e-05,
|
| 3635 |
+
"loss": 0.4087,
|
| 3636 |
+
"step": 5940
|
| 3637 |
+
},
|
| 3638 |
+
{
|
| 3639 |
+
"epoch": 7.46,
|
| 3640 |
+
"learning_rate": 1.4132553606237817e-05,
|
| 3641 |
+
"loss": 0.3838,
|
| 3642 |
+
"step": 5950
|
| 3643 |
+
},
|
| 3644 |
+
{
|
| 3645 |
+
"epoch": 7.47,
|
| 3646 |
+
"learning_rate": 1.4062935115566695e-05,
|
| 3647 |
+
"loss": 0.4084,
|
| 3648 |
+
"step": 5960
|
| 3649 |
+
},
|
| 3650 |
+
{
|
| 3651 |
+
"epoch": 7.48,
|
| 3652 |
+
"learning_rate": 1.3993316624895573e-05,
|
| 3653 |
+
"loss": 0.3851,
|
| 3654 |
+
"step": 5970
|
| 3655 |
+
},
|
| 3656 |
+
{
|
| 3657 |
+
"epoch": 7.49,
|
| 3658 |
+
"learning_rate": 1.392369813422445e-05,
|
| 3659 |
+
"loss": 0.4234,
|
| 3660 |
+
"step": 5980
|
| 3661 |
+
},
|
| 3662 |
+
{
|
| 3663 |
+
"epoch": 7.51,
|
| 3664 |
+
"learning_rate": 1.3854079643553327e-05,
|
| 3665 |
+
"loss": 0.4183,
|
| 3666 |
+
"step": 5990
|
| 3667 |
+
},
|
| 3668 |
+
{
|
| 3669 |
+
"epoch": 7.52,
|
| 3670 |
+
"learning_rate": 1.3784461152882205e-05,
|
| 3671 |
+
"loss": 0.4108,
|
| 3672 |
+
"step": 6000
|
| 3673 |
+
},
|
| 3674 |
+
{
|
| 3675 |
+
"epoch": 7.53,
|
| 3676 |
+
"learning_rate": 1.3714842662211083e-05,
|
| 3677 |
+
"loss": 0.4314,
|
| 3678 |
+
"step": 6010
|
| 3679 |
+
},
|
| 3680 |
+
{
|
| 3681 |
+
"epoch": 7.54,
|
| 3682 |
+
"learning_rate": 1.364522417153996e-05,
|
| 3683 |
+
"loss": 0.4252,
|
| 3684 |
+
"step": 6020
|
| 3685 |
+
},
|
| 3686 |
+
{
|
| 3687 |
+
"epoch": 7.56,
|
| 3688 |
+
"learning_rate": 1.3575605680868838e-05,
|
| 3689 |
+
"loss": 0.4272,
|
| 3690 |
+
"step": 6030
|
| 3691 |
+
},
|
| 3692 |
+
{
|
| 3693 |
+
"epoch": 7.57,
|
| 3694 |
+
"learning_rate": 1.3505987190197716e-05,
|
| 3695 |
+
"loss": 0.3799,
|
| 3696 |
+
"step": 6040
|
| 3697 |
+
},
|
| 3698 |
+
{
|
| 3699 |
+
"epoch": 7.58,
|
| 3700 |
+
"learning_rate": 1.3436368699526594e-05,
|
| 3701 |
+
"loss": 0.3705,
|
| 3702 |
+
"step": 6050
|
| 3703 |
+
},
|
| 3704 |
+
{
|
| 3705 |
+
"epoch": 7.59,
|
| 3706 |
+
"learning_rate": 1.3366750208855472e-05,
|
| 3707 |
+
"loss": 0.4124,
|
| 3708 |
+
"step": 6060
|
| 3709 |
+
},
|
| 3710 |
+
{
|
| 3711 |
+
"epoch": 7.61,
|
| 3712 |
+
"learning_rate": 1.329713171818435e-05,
|
| 3713 |
+
"loss": 0.3976,
|
| 3714 |
+
"step": 6070
|
| 3715 |
+
},
|
| 3716 |
+
{
|
| 3717 |
+
"epoch": 7.62,
|
| 3718 |
+
"learning_rate": 1.3227513227513228e-05,
|
| 3719 |
+
"loss": 0.4187,
|
| 3720 |
+
"step": 6080
|
| 3721 |
+
},
|
| 3722 |
+
{
|
| 3723 |
+
"epoch": 7.63,
|
| 3724 |
+
"learning_rate": 1.3157894736842106e-05,
|
| 3725 |
+
"loss": 0.386,
|
| 3726 |
+
"step": 6090
|
| 3727 |
+
},
|
| 3728 |
+
{
|
| 3729 |
+
"epoch": 7.64,
|
| 3730 |
+
"learning_rate": 1.3088276246170982e-05,
|
| 3731 |
+
"loss": 0.4153,
|
| 3732 |
+
"step": 6100
|
| 3733 |
+
},
|
| 3734 |
+
{
|
| 3735 |
+
"epoch": 7.66,
|
| 3736 |
+
"learning_rate": 1.301865775549986e-05,
|
| 3737 |
+
"loss": 0.4067,
|
| 3738 |
+
"step": 6110
|
| 3739 |
+
},
|
| 3740 |
+
{
|
| 3741 |
+
"epoch": 7.67,
|
| 3742 |
+
"learning_rate": 1.2949039264828738e-05,
|
| 3743 |
+
"loss": 0.3874,
|
| 3744 |
+
"step": 6120
|
| 3745 |
+
},
|
| 3746 |
+
{
|
| 3747 |
+
"epoch": 7.68,
|
| 3748 |
+
"learning_rate": 1.2879420774157616e-05,
|
| 3749 |
+
"loss": 0.3921,
|
| 3750 |
+
"step": 6130
|
| 3751 |
+
},
|
| 3752 |
+
{
|
| 3753 |
+
"epoch": 7.69,
|
| 3754 |
+
"learning_rate": 1.2809802283486494e-05,
|
| 3755 |
+
"loss": 0.369,
|
| 3756 |
+
"step": 6140
|
| 3757 |
+
},
|
| 3758 |
+
{
|
| 3759 |
+
"epoch": 7.71,
|
| 3760 |
+
"learning_rate": 1.2740183792815372e-05,
|
| 3761 |
+
"loss": 0.4082,
|
| 3762 |
+
"step": 6150
|
| 3763 |
+
},
|
| 3764 |
+
{
|
| 3765 |
+
"epoch": 7.72,
|
| 3766 |
+
"learning_rate": 1.267056530214425e-05,
|
| 3767 |
+
"loss": 0.4238,
|
| 3768 |
+
"step": 6160
|
| 3769 |
+
},
|
| 3770 |
+
{
|
| 3771 |
+
"epoch": 7.73,
|
| 3772 |
+
"learning_rate": 1.2600946811473128e-05,
|
| 3773 |
+
"loss": 0.3899,
|
| 3774 |
+
"step": 6170
|
| 3775 |
+
},
|
| 3776 |
+
{
|
| 3777 |
+
"epoch": 7.74,
|
| 3778 |
+
"learning_rate": 1.2531328320802006e-05,
|
| 3779 |
+
"loss": 0.3806,
|
| 3780 |
+
"step": 6180
|
| 3781 |
+
},
|
| 3782 |
+
{
|
| 3783 |
+
"epoch": 7.76,
|
| 3784 |
+
"learning_rate": 1.2461709830130883e-05,
|
| 3785 |
+
"loss": 0.3617,
|
| 3786 |
+
"step": 6190
|
| 3787 |
+
},
|
| 3788 |
+
{
|
| 3789 |
+
"epoch": 7.77,
|
| 3790 |
+
"learning_rate": 1.239209133945976e-05,
|
| 3791 |
+
"loss": 0.4171,
|
| 3792 |
+
"step": 6200
|
| 3793 |
+
},
|
| 3794 |
+
{
|
| 3795 |
+
"epoch": 7.78,
|
| 3796 |
+
"learning_rate": 1.2322472848788638e-05,
|
| 3797 |
+
"loss": 0.3941,
|
| 3798 |
+
"step": 6210
|
| 3799 |
+
},
|
| 3800 |
+
{
|
| 3801 |
+
"epoch": 7.79,
|
| 3802 |
+
"learning_rate": 1.2252854358117516e-05,
|
| 3803 |
+
"loss": 0.3891,
|
| 3804 |
+
"step": 6220
|
| 3805 |
+
},
|
| 3806 |
+
{
|
| 3807 |
+
"epoch": 7.81,
|
| 3808 |
+
"learning_rate": 1.2183235867446393e-05,
|
| 3809 |
+
"loss": 0.3987,
|
| 3810 |
+
"step": 6230
|
| 3811 |
+
},
|
| 3812 |
+
{
|
| 3813 |
+
"epoch": 7.82,
|
| 3814 |
+
"learning_rate": 1.2113617376775271e-05,
|
| 3815 |
+
"loss": 0.3707,
|
| 3816 |
+
"step": 6240
|
| 3817 |
+
},
|
| 3818 |
+
{
|
| 3819 |
+
"epoch": 7.83,
|
| 3820 |
+
"learning_rate": 1.204399888610415e-05,
|
| 3821 |
+
"loss": 0.3835,
|
| 3822 |
+
"step": 6250
|
| 3823 |
+
},
|
| 3824 |
+
{
|
| 3825 |
+
"epoch": 7.84,
|
| 3826 |
+
"learning_rate": 1.1974380395433027e-05,
|
| 3827 |
+
"loss": 0.3925,
|
| 3828 |
+
"step": 6260
|
| 3829 |
+
},
|
| 3830 |
+
{
|
| 3831 |
+
"epoch": 7.86,
|
| 3832 |
+
"learning_rate": 1.1904761904761905e-05,
|
| 3833 |
+
"loss": 0.3693,
|
| 3834 |
+
"step": 6270
|
| 3835 |
+
},
|
| 3836 |
+
{
|
| 3837 |
+
"epoch": 7.87,
|
| 3838 |
+
"learning_rate": 1.1835143414090783e-05,
|
| 3839 |
+
"loss": 0.3809,
|
| 3840 |
+
"step": 6280
|
| 3841 |
+
},
|
| 3842 |
+
{
|
| 3843 |
+
"epoch": 7.88,
|
| 3844 |
+
"learning_rate": 1.1765524923419661e-05,
|
| 3845 |
+
"loss": 0.3872,
|
| 3846 |
+
"step": 6290
|
| 3847 |
+
},
|
| 3848 |
+
{
|
| 3849 |
+
"epoch": 7.89,
|
| 3850 |
+
"learning_rate": 1.1695906432748537e-05,
|
| 3851 |
+
"loss": 0.3964,
|
| 3852 |
+
"step": 6300
|
| 3853 |
+
},
|
| 3854 |
+
{
|
| 3855 |
+
"epoch": 7.91,
|
| 3856 |
+
"learning_rate": 1.1626287942077415e-05,
|
| 3857 |
+
"loss": 0.3767,
|
| 3858 |
+
"step": 6310
|
| 3859 |
+
},
|
| 3860 |
+
{
|
| 3861 |
+
"epoch": 7.92,
|
| 3862 |
+
"learning_rate": 1.1556669451406293e-05,
|
| 3863 |
+
"loss": 0.3902,
|
| 3864 |
+
"step": 6320
|
| 3865 |
+
},
|
| 3866 |
+
{
|
| 3867 |
+
"epoch": 7.93,
|
| 3868 |
+
"learning_rate": 1.1487050960735171e-05,
|
| 3869 |
+
"loss": 0.3551,
|
| 3870 |
+
"step": 6330
|
| 3871 |
+
},
|
| 3872 |
+
{
|
| 3873 |
+
"epoch": 7.94,
|
| 3874 |
+
"learning_rate": 1.1417432470064049e-05,
|
| 3875 |
+
"loss": 0.4097,
|
| 3876 |
+
"step": 6340
|
| 3877 |
+
},
|
| 3878 |
+
{
|
| 3879 |
+
"epoch": 7.96,
|
| 3880 |
+
"learning_rate": 1.1347813979392927e-05,
|
| 3881 |
+
"loss": 0.4374,
|
| 3882 |
+
"step": 6350
|
| 3883 |
+
},
|
| 3884 |
+
{
|
| 3885 |
+
"epoch": 7.97,
|
| 3886 |
+
"learning_rate": 1.1278195488721805e-05,
|
| 3887 |
+
"loss": 0.3815,
|
| 3888 |
+
"step": 6360
|
| 3889 |
+
},
|
| 3890 |
+
{
|
| 3891 |
+
"epoch": 7.98,
|
| 3892 |
+
"learning_rate": 1.1208576998050683e-05,
|
| 3893 |
+
"loss": 0.3829,
|
| 3894 |
+
"step": 6370
|
| 3895 |
+
},
|
| 3896 |
+
{
|
| 3897 |
+
"epoch": 7.99,
|
| 3898 |
+
"learning_rate": 1.113895850737956e-05,
|
| 3899 |
+
"loss": 0.3806,
|
| 3900 |
+
"step": 6380
|
| 3901 |
+
},
|
| 3902 |
+
{
|
| 3903 |
+
"epoch": 8.0,
|
| 3904 |
+
"eval_accuracy": 0.853109335839599,
|
| 3905 |
+
"eval_loss": 0.4375392496585846,
|
| 3906 |
+
"eval_runtime": 234.5395,
|
| 3907 |
+
"eval_samples_per_second": 108.877,
|
| 3908 |
+
"eval_steps_per_second": 3.402,
|
| 3909 |
+
"step": 6384
|
| 3910 |
+
},
|
| 3911 |
+
{
|
| 3912 |
+
"epoch": 8.01,
|
| 3913 |
+
"learning_rate": 1.1069340016708438e-05,
|
| 3914 |
+
"loss": 0.3573,
|
| 3915 |
+
"step": 6390
|
| 3916 |
+
},
|
| 3917 |
+
{
|
| 3918 |
+
"epoch": 8.02,
|
| 3919 |
+
"learning_rate": 1.0999721526037315e-05,
|
| 3920 |
+
"loss": 0.3303,
|
| 3921 |
+
"step": 6400
|
| 3922 |
+
},
|
| 3923 |
+
{
|
| 3924 |
+
"epoch": 8.03,
|
| 3925 |
+
"learning_rate": 1.0930103035366193e-05,
|
| 3926 |
+
"loss": 0.3206,
|
| 3927 |
+
"step": 6410
|
| 3928 |
+
},
|
| 3929 |
+
{
|
| 3930 |
+
"epoch": 8.05,
|
| 3931 |
+
"learning_rate": 1.086048454469507e-05,
|
| 3932 |
+
"loss": 0.3416,
|
| 3933 |
+
"step": 6420
|
| 3934 |
+
},
|
| 3935 |
+
{
|
| 3936 |
+
"epoch": 8.06,
|
| 3937 |
+
"learning_rate": 1.0790866054023948e-05,
|
| 3938 |
+
"loss": 0.3379,
|
| 3939 |
+
"step": 6430
|
| 3940 |
+
},
|
| 3941 |
+
{
|
| 3942 |
+
"epoch": 8.07,
|
| 3943 |
+
"learning_rate": 1.0721247563352826e-05,
|
| 3944 |
+
"loss": 0.3255,
|
| 3945 |
+
"step": 6440
|
| 3946 |
+
},
|
| 3947 |
+
{
|
| 3948 |
+
"epoch": 8.08,
|
| 3949 |
+
"learning_rate": 1.0651629072681704e-05,
|
| 3950 |
+
"loss": 0.3278,
|
| 3951 |
+
"step": 6450
|
| 3952 |
+
},
|
| 3953 |
+
{
|
| 3954 |
+
"epoch": 8.1,
|
| 3955 |
+
"learning_rate": 1.0582010582010582e-05,
|
| 3956 |
+
"loss": 0.3136,
|
| 3957 |
+
"step": 6460
|
| 3958 |
+
},
|
| 3959 |
+
{
|
| 3960 |
+
"epoch": 8.11,
|
| 3961 |
+
"learning_rate": 1.051239209133946e-05,
|
| 3962 |
+
"loss": 0.2932,
|
| 3963 |
+
"step": 6470
|
| 3964 |
+
},
|
| 3965 |
+
{
|
| 3966 |
+
"epoch": 8.12,
|
| 3967 |
+
"learning_rate": 1.0442773600668338e-05,
|
| 3968 |
+
"loss": 0.3199,
|
| 3969 |
+
"step": 6480
|
| 3970 |
+
},
|
| 3971 |
+
{
|
| 3972 |
+
"epoch": 8.13,
|
| 3973 |
+
"learning_rate": 1.0373155109997216e-05,
|
| 3974 |
+
"loss": 0.3212,
|
| 3975 |
+
"step": 6490
|
| 3976 |
+
},
|
| 3977 |
+
{
|
| 3978 |
+
"epoch": 8.15,
|
| 3979 |
+
"learning_rate": 1.0303536619326092e-05,
|
| 3980 |
+
"loss": 0.3487,
|
| 3981 |
+
"step": 6500
|
| 3982 |
+
},
|
| 3983 |
+
{
|
| 3984 |
+
"epoch": 8.16,
|
| 3985 |
+
"learning_rate": 1.023391812865497e-05,
|
| 3986 |
+
"loss": 0.3364,
|
| 3987 |
+
"step": 6510
|
| 3988 |
+
},
|
| 3989 |
+
{
|
| 3990 |
+
"epoch": 8.17,
|
| 3991 |
+
"learning_rate": 1.0164299637983848e-05,
|
| 3992 |
+
"loss": 0.3261,
|
| 3993 |
+
"step": 6520
|
| 3994 |
+
},
|
| 3995 |
+
{
|
| 3996 |
+
"epoch": 8.18,
|
| 3997 |
+
"learning_rate": 1.0094681147312726e-05,
|
| 3998 |
+
"loss": 0.3398,
|
| 3999 |
+
"step": 6530
|
| 4000 |
+
},
|
| 4001 |
+
{
|
| 4002 |
+
"epoch": 8.2,
|
| 4003 |
+
"learning_rate": 1.0025062656641604e-05,
|
| 4004 |
+
"loss": 0.3111,
|
| 4005 |
+
"step": 6540
|
| 4006 |
+
},
|
| 4007 |
+
{
|
| 4008 |
+
"epoch": 8.21,
|
| 4009 |
+
"learning_rate": 9.955444165970482e-06,
|
| 4010 |
+
"loss": 0.3523,
|
| 4011 |
+
"step": 6550
|
| 4012 |
+
},
|
| 4013 |
+
{
|
| 4014 |
+
"epoch": 8.22,
|
| 4015 |
+
"learning_rate": 9.88582567529936e-06,
|
| 4016 |
+
"loss": 0.3325,
|
| 4017 |
+
"step": 6560
|
| 4018 |
+
},
|
| 4019 |
+
{
|
| 4020 |
+
"epoch": 8.23,
|
| 4021 |
+
"learning_rate": 9.816207184628238e-06,
|
| 4022 |
+
"loss": 0.329,
|
| 4023 |
+
"step": 6570
|
| 4024 |
+
},
|
| 4025 |
+
{
|
| 4026 |
+
"epoch": 8.25,
|
| 4027 |
+
"learning_rate": 9.746588693957115e-06,
|
| 4028 |
+
"loss": 0.3558,
|
| 4029 |
+
"step": 6580
|
| 4030 |
+
},
|
| 4031 |
+
{
|
| 4032 |
+
"epoch": 8.26,
|
| 4033 |
+
"learning_rate": 9.676970203285993e-06,
|
| 4034 |
+
"loss": 0.3286,
|
| 4035 |
+
"step": 6590
|
| 4036 |
+
},
|
| 4037 |
+
{
|
| 4038 |
+
"epoch": 8.27,
|
| 4039 |
+
"learning_rate": 9.607351712614871e-06,
|
| 4040 |
+
"loss": 0.3429,
|
| 4041 |
+
"step": 6600
|
| 4042 |
+
},
|
| 4043 |
+
{
|
| 4044 |
+
"epoch": 8.28,
|
| 4045 |
+
"learning_rate": 9.537733221943747e-06,
|
| 4046 |
+
"loss": 0.3146,
|
| 4047 |
+
"step": 6610
|
| 4048 |
+
},
|
| 4049 |
+
{
|
| 4050 |
+
"epoch": 8.3,
|
| 4051 |
+
"learning_rate": 9.468114731272625e-06,
|
| 4052 |
+
"loss": 0.3075,
|
| 4053 |
+
"step": 6620
|
| 4054 |
+
},
|
| 4055 |
+
{
|
| 4056 |
+
"epoch": 8.31,
|
| 4057 |
+
"learning_rate": 9.398496240601503e-06,
|
| 4058 |
+
"loss": 0.3053,
|
| 4059 |
+
"step": 6630
|
| 4060 |
+
},
|
| 4061 |
+
{
|
| 4062 |
+
"epoch": 8.32,
|
| 4063 |
+
"learning_rate": 9.328877749930381e-06,
|
| 4064 |
+
"loss": 0.3262,
|
| 4065 |
+
"step": 6640
|
| 4066 |
+
},
|
| 4067 |
+
{
|
| 4068 |
+
"epoch": 8.33,
|
| 4069 |
+
"learning_rate": 9.259259259259259e-06,
|
| 4070 |
+
"loss": 0.3315,
|
| 4071 |
+
"step": 6650
|
| 4072 |
+
},
|
| 4073 |
+
{
|
| 4074 |
+
"epoch": 8.35,
|
| 4075 |
+
"learning_rate": 9.189640768588137e-06,
|
| 4076 |
+
"loss": 0.3446,
|
| 4077 |
+
"step": 6660
|
| 4078 |
+
},
|
| 4079 |
+
{
|
| 4080 |
+
"epoch": 8.36,
|
| 4081 |
+
"learning_rate": 9.120022277917015e-06,
|
| 4082 |
+
"loss": 0.3406,
|
| 4083 |
+
"step": 6670
|
| 4084 |
+
},
|
| 4085 |
+
{
|
| 4086 |
+
"epoch": 8.37,
|
| 4087 |
+
"learning_rate": 9.050403787245893e-06,
|
| 4088 |
+
"loss": 0.3119,
|
| 4089 |
+
"step": 6680
|
| 4090 |
+
},
|
| 4091 |
+
{
|
| 4092 |
+
"epoch": 8.38,
|
| 4093 |
+
"learning_rate": 8.98078529657477e-06,
|
| 4094 |
+
"loss": 0.3102,
|
| 4095 |
+
"step": 6690
|
| 4096 |
+
},
|
| 4097 |
+
{
|
| 4098 |
+
"epoch": 8.4,
|
| 4099 |
+
"learning_rate": 8.911166805903649e-06,
|
| 4100 |
+
"loss": 0.3051,
|
| 4101 |
+
"step": 6700
|
| 4102 |
+
},
|
| 4103 |
+
{
|
| 4104 |
+
"epoch": 8.41,
|
| 4105 |
+
"learning_rate": 8.841548315232525e-06,
|
| 4106 |
+
"loss": 0.3509,
|
| 4107 |
+
"step": 6710
|
| 4108 |
+
},
|
| 4109 |
+
{
|
| 4110 |
+
"epoch": 8.42,
|
| 4111 |
+
"learning_rate": 8.771929824561403e-06,
|
| 4112 |
+
"loss": 0.3265,
|
| 4113 |
+
"step": 6720
|
| 4114 |
+
},
|
| 4115 |
+
{
|
| 4116 |
+
"epoch": 8.43,
|
| 4117 |
+
"learning_rate": 8.70231133389028e-06,
|
| 4118 |
+
"loss": 0.3522,
|
| 4119 |
+
"step": 6730
|
| 4120 |
+
},
|
| 4121 |
+
{
|
| 4122 |
+
"epoch": 8.45,
|
| 4123 |
+
"learning_rate": 8.632692843219159e-06,
|
| 4124 |
+
"loss": 0.3029,
|
| 4125 |
+
"step": 6740
|
| 4126 |
+
},
|
| 4127 |
+
{
|
| 4128 |
+
"epoch": 8.46,
|
| 4129 |
+
"learning_rate": 8.563074352548037e-06,
|
| 4130 |
+
"loss": 0.317,
|
| 4131 |
+
"step": 6750
|
| 4132 |
+
},
|
| 4133 |
+
{
|
| 4134 |
+
"epoch": 8.47,
|
| 4135 |
+
"learning_rate": 8.493455861876915e-06,
|
| 4136 |
+
"loss": 0.3446,
|
| 4137 |
+
"step": 6760
|
| 4138 |
+
},
|
| 4139 |
+
{
|
| 4140 |
+
"epoch": 8.48,
|
| 4141 |
+
"learning_rate": 8.423837371205792e-06,
|
| 4142 |
+
"loss": 0.3627,
|
| 4143 |
+
"step": 6770
|
| 4144 |
+
},
|
| 4145 |
+
{
|
| 4146 |
+
"epoch": 8.5,
|
| 4147 |
+
"learning_rate": 8.35421888053467e-06,
|
| 4148 |
+
"loss": 0.3583,
|
| 4149 |
+
"step": 6780
|
| 4150 |
+
},
|
| 4151 |
+
{
|
| 4152 |
+
"epoch": 8.51,
|
| 4153 |
+
"learning_rate": 8.284600389863548e-06,
|
| 4154 |
+
"loss": 0.3106,
|
| 4155 |
+
"step": 6790
|
| 4156 |
+
},
|
| 4157 |
+
{
|
| 4158 |
+
"epoch": 8.52,
|
| 4159 |
+
"learning_rate": 8.214981899192426e-06,
|
| 4160 |
+
"loss": 0.3142,
|
| 4161 |
+
"step": 6800
|
| 4162 |
+
},
|
| 4163 |
+
{
|
| 4164 |
+
"epoch": 8.53,
|
| 4165 |
+
"learning_rate": 8.145363408521302e-06,
|
| 4166 |
+
"loss": 0.3031,
|
| 4167 |
+
"step": 6810
|
| 4168 |
+
},
|
| 4169 |
+
{
|
| 4170 |
+
"epoch": 8.55,
|
| 4171 |
+
"learning_rate": 8.07574491785018e-06,
|
| 4172 |
+
"loss": 0.3235,
|
| 4173 |
+
"step": 6820
|
| 4174 |
+
},
|
| 4175 |
+
{
|
| 4176 |
+
"epoch": 8.56,
|
| 4177 |
+
"learning_rate": 8.006126427179058e-06,
|
| 4178 |
+
"loss": 0.3186,
|
| 4179 |
+
"step": 6830
|
| 4180 |
+
},
|
| 4181 |
+
{
|
| 4182 |
+
"epoch": 8.57,
|
| 4183 |
+
"learning_rate": 7.936507936507936e-06,
|
| 4184 |
+
"loss": 0.2935,
|
| 4185 |
+
"step": 6840
|
| 4186 |
+
},
|
| 4187 |
+
{
|
| 4188 |
+
"epoch": 8.58,
|
| 4189 |
+
"learning_rate": 7.866889445836814e-06,
|
| 4190 |
+
"loss": 0.2895,
|
| 4191 |
+
"step": 6850
|
| 4192 |
+
},
|
| 4193 |
+
{
|
| 4194 |
+
"epoch": 8.6,
|
| 4195 |
+
"learning_rate": 7.797270955165692e-06,
|
| 4196 |
+
"loss": 0.3252,
|
| 4197 |
+
"step": 6860
|
| 4198 |
+
},
|
| 4199 |
+
{
|
| 4200 |
+
"epoch": 8.61,
|
| 4201 |
+
"learning_rate": 7.72765246449457e-06,
|
| 4202 |
+
"loss": 0.3128,
|
| 4203 |
+
"step": 6870
|
| 4204 |
+
},
|
| 4205 |
+
{
|
| 4206 |
+
"epoch": 8.62,
|
| 4207 |
+
"learning_rate": 7.658033973823448e-06,
|
| 4208 |
+
"loss": 0.3176,
|
| 4209 |
+
"step": 6880
|
| 4210 |
+
},
|
| 4211 |
+
{
|
| 4212 |
+
"epoch": 8.63,
|
| 4213 |
+
"learning_rate": 7.588415483152325e-06,
|
| 4214 |
+
"loss": 0.3343,
|
| 4215 |
+
"step": 6890
|
| 4216 |
+
},
|
| 4217 |
+
{
|
| 4218 |
+
"epoch": 8.65,
|
| 4219 |
+
"learning_rate": 7.518796992481203e-06,
|
| 4220 |
+
"loss": 0.3738,
|
| 4221 |
+
"step": 6900
|
| 4222 |
+
},
|
| 4223 |
+
{
|
| 4224 |
+
"epoch": 8.66,
|
| 4225 |
+
"learning_rate": 7.449178501810081e-06,
|
| 4226 |
+
"loss": 0.3144,
|
| 4227 |
+
"step": 6910
|
| 4228 |
+
},
|
| 4229 |
+
{
|
| 4230 |
+
"epoch": 8.67,
|
| 4231 |
+
"learning_rate": 7.379560011138959e-06,
|
| 4232 |
+
"loss": 0.3808,
|
| 4233 |
+
"step": 6920
|
| 4234 |
+
},
|
| 4235 |
+
{
|
| 4236 |
+
"epoch": 8.68,
|
| 4237 |
+
"learning_rate": 7.3099415204678366e-06,
|
| 4238 |
+
"loss": 0.3089,
|
| 4239 |
+
"step": 6930
|
| 4240 |
+
},
|
| 4241 |
+
{
|
| 4242 |
+
"epoch": 8.7,
|
| 4243 |
+
"learning_rate": 7.240323029796714e-06,
|
| 4244 |
+
"loss": 0.3327,
|
| 4245 |
+
"step": 6940
|
| 4246 |
+
},
|
| 4247 |
+
{
|
| 4248 |
+
"epoch": 8.71,
|
| 4249 |
+
"learning_rate": 7.1707045391255915e-06,
|
| 4250 |
+
"loss": 0.3289,
|
| 4251 |
+
"step": 6950
|
| 4252 |
+
},
|
| 4253 |
+
{
|
| 4254 |
+
"epoch": 8.72,
|
| 4255 |
+
"learning_rate": 7.1010860484544695e-06,
|
| 4256 |
+
"loss": 0.3699,
|
| 4257 |
+
"step": 6960
|
| 4258 |
+
},
|
| 4259 |
+
{
|
| 4260 |
+
"epoch": 8.73,
|
| 4261 |
+
"learning_rate": 7.031467557783347e-06,
|
| 4262 |
+
"loss": 0.3061,
|
| 4263 |
+
"step": 6970
|
| 4264 |
+
},
|
| 4265 |
+
{
|
| 4266 |
+
"epoch": 8.75,
|
| 4267 |
+
"learning_rate": 6.961849067112225e-06,
|
| 4268 |
+
"loss": 0.3026,
|
| 4269 |
+
"step": 6980
|
| 4270 |
+
},
|
| 4271 |
+
{
|
| 4272 |
+
"epoch": 8.76,
|
| 4273 |
+
"learning_rate": 6.892230576441102e-06,
|
| 4274 |
+
"loss": 0.3233,
|
| 4275 |
+
"step": 6990
|
| 4276 |
+
},
|
| 4277 |
+
{
|
| 4278 |
+
"epoch": 8.77,
|
| 4279 |
+
"learning_rate": 6.82261208576998e-06,
|
| 4280 |
+
"loss": 0.3106,
|
| 4281 |
+
"step": 7000
|
| 4282 |
+
},
|
| 4283 |
+
{
|
| 4284 |
+
"epoch": 8.78,
|
| 4285 |
+
"learning_rate": 6.752993595098858e-06,
|
| 4286 |
+
"loss": 0.3429,
|
| 4287 |
+
"step": 7010
|
| 4288 |
+
},
|
| 4289 |
+
{
|
| 4290 |
+
"epoch": 8.8,
|
| 4291 |
+
"learning_rate": 6.683375104427736e-06,
|
| 4292 |
+
"loss": 0.3397,
|
| 4293 |
+
"step": 7020
|
| 4294 |
+
},
|
| 4295 |
+
{
|
| 4296 |
+
"epoch": 8.81,
|
| 4297 |
+
"learning_rate": 6.613756613756614e-06,
|
| 4298 |
+
"loss": 0.352,
|
| 4299 |
+
"step": 7030
|
| 4300 |
+
},
|
| 4301 |
+
{
|
| 4302 |
+
"epoch": 8.82,
|
| 4303 |
+
"learning_rate": 6.544138123085491e-06,
|
| 4304 |
+
"loss": 0.2803,
|
| 4305 |
+
"step": 7040
|
| 4306 |
+
},
|
| 4307 |
+
{
|
| 4308 |
+
"epoch": 8.83,
|
| 4309 |
+
"learning_rate": 6.474519632414369e-06,
|
| 4310 |
+
"loss": 0.3335,
|
| 4311 |
+
"step": 7050
|
| 4312 |
+
},
|
| 4313 |
+
{
|
| 4314 |
+
"epoch": 8.85,
|
| 4315 |
+
"learning_rate": 6.404901141743247e-06,
|
| 4316 |
+
"loss": 0.3054,
|
| 4317 |
+
"step": 7060
|
| 4318 |
+
},
|
| 4319 |
+
{
|
| 4320 |
+
"epoch": 8.86,
|
| 4321 |
+
"learning_rate": 6.335282651072125e-06,
|
| 4322 |
+
"loss": 0.3383,
|
| 4323 |
+
"step": 7070
|
| 4324 |
+
},
|
| 4325 |
+
{
|
| 4326 |
+
"epoch": 8.87,
|
| 4327 |
+
"learning_rate": 6.265664160401003e-06,
|
| 4328 |
+
"loss": 0.3237,
|
| 4329 |
+
"step": 7080
|
| 4330 |
+
},
|
| 4331 |
+
{
|
| 4332 |
+
"epoch": 8.88,
|
| 4333 |
+
"learning_rate": 6.19604566972988e-06,
|
| 4334 |
+
"loss": 0.3227,
|
| 4335 |
+
"step": 7090
|
| 4336 |
+
},
|
| 4337 |
+
{
|
| 4338 |
+
"epoch": 8.9,
|
| 4339 |
+
"learning_rate": 6.126427179058758e-06,
|
| 4340 |
+
"loss": 0.3375,
|
| 4341 |
+
"step": 7100
|
| 4342 |
+
},
|
| 4343 |
+
{
|
| 4344 |
+
"epoch": 8.91,
|
| 4345 |
+
"learning_rate": 6.056808688387636e-06,
|
| 4346 |
+
"loss": 0.2785,
|
| 4347 |
+
"step": 7110
|
| 4348 |
+
},
|
| 4349 |
+
{
|
| 4350 |
+
"epoch": 8.92,
|
| 4351 |
+
"learning_rate": 5.987190197716514e-06,
|
| 4352 |
+
"loss": 0.31,
|
| 4353 |
+
"step": 7120
|
| 4354 |
+
},
|
| 4355 |
+
{
|
| 4356 |
+
"epoch": 8.93,
|
| 4357 |
+
"learning_rate": 5.9175717070453915e-06,
|
| 4358 |
+
"loss": 0.3472,
|
| 4359 |
+
"step": 7130
|
| 4360 |
+
},
|
| 4361 |
+
{
|
| 4362 |
+
"epoch": 8.95,
|
| 4363 |
+
"learning_rate": 5.8479532163742686e-06,
|
| 4364 |
+
"loss": 0.3577,
|
| 4365 |
+
"step": 7140
|
| 4366 |
+
},
|
| 4367 |
+
{
|
| 4368 |
+
"epoch": 8.96,
|
| 4369 |
+
"learning_rate": 5.7783347257031465e-06,
|
| 4370 |
+
"loss": 0.3299,
|
| 4371 |
+
"step": 7150
|
| 4372 |
+
},
|
| 4373 |
+
{
|
| 4374 |
+
"epoch": 8.97,
|
| 4375 |
+
"learning_rate": 5.708716235032024e-06,
|
| 4376 |
+
"loss": 0.2989,
|
| 4377 |
+
"step": 7160
|
| 4378 |
+
},
|
| 4379 |
+
{
|
| 4380 |
+
"epoch": 8.98,
|
| 4381 |
+
"learning_rate": 5.639097744360902e-06,
|
| 4382 |
+
"loss": 0.2973,
|
| 4383 |
+
"step": 7170
|
| 4384 |
+
},
|
| 4385 |
+
{
|
| 4386 |
+
"epoch": 9.0,
|
| 4387 |
+
"learning_rate": 5.56947925368978e-06,
|
| 4388 |
+
"loss": 0.3112,
|
| 4389 |
+
"step": 7180
|
| 4390 |
+
},
|
| 4391 |
+
{
|
| 4392 |
+
"epoch": 9.0,
|
| 4393 |
+
"eval_accuracy": 0.8556547619047619,
|
| 4394 |
+
"eval_loss": 0.4372400641441345,
|
| 4395 |
+
"eval_runtime": 234.8854,
|
| 4396 |
+
"eval_samples_per_second": 108.717,
|
| 4397 |
+
"eval_steps_per_second": 3.397,
|
| 4398 |
+
"step": 7182
|
| 4399 |
+
},
|
| 4400 |
+
{
|
| 4401 |
+
"epoch": 9.01,
|
| 4402 |
+
"learning_rate": 5.499860763018657e-06,
|
| 4403 |
+
"loss": 0.2907,
|
| 4404 |
+
"step": 7190
|
| 4405 |
+
},
|
| 4406 |
+
{
|
| 4407 |
+
"epoch": 9.02,
|
| 4408 |
+
"learning_rate": 5.430242272347535e-06,
|
| 4409 |
+
"loss": 0.3028,
|
| 4410 |
+
"step": 7200
|
| 4411 |
+
},
|
| 4412 |
+
{
|
| 4413 |
+
"epoch": 9.04,
|
| 4414 |
+
"learning_rate": 5.360623781676413e-06,
|
| 4415 |
+
"loss": 0.2706,
|
| 4416 |
+
"step": 7210
|
| 4417 |
+
},
|
| 4418 |
+
{
|
| 4419 |
+
"epoch": 9.05,
|
| 4420 |
+
"learning_rate": 5.291005291005291e-06,
|
| 4421 |
+
"loss": 0.2695,
|
| 4422 |
+
"step": 7220
|
| 4423 |
+
},
|
| 4424 |
+
{
|
| 4425 |
+
"epoch": 9.06,
|
| 4426 |
+
"learning_rate": 5.221386800334169e-06,
|
| 4427 |
+
"loss": 0.3298,
|
| 4428 |
+
"step": 7230
|
| 4429 |
+
},
|
| 4430 |
+
{
|
| 4431 |
+
"epoch": 9.07,
|
| 4432 |
+
"learning_rate": 5.151768309663046e-06,
|
| 4433 |
+
"loss": 0.2717,
|
| 4434 |
+
"step": 7240
|
| 4435 |
+
},
|
| 4436 |
+
{
|
| 4437 |
+
"epoch": 9.09,
|
| 4438 |
+
"learning_rate": 5.082149818991924e-06,
|
| 4439 |
+
"loss": 0.2785,
|
| 4440 |
+
"step": 7250
|
| 4441 |
+
},
|
| 4442 |
+
{
|
| 4443 |
+
"epoch": 9.1,
|
| 4444 |
+
"learning_rate": 5.012531328320802e-06,
|
| 4445 |
+
"loss": 0.2635,
|
| 4446 |
+
"step": 7260
|
| 4447 |
+
},
|
| 4448 |
+
{
|
| 4449 |
+
"epoch": 9.11,
|
| 4450 |
+
"learning_rate": 4.94291283764968e-06,
|
| 4451 |
+
"loss": 0.305,
|
| 4452 |
+
"step": 7270
|
| 4453 |
+
},
|
| 4454 |
+
{
|
| 4455 |
+
"epoch": 9.12,
|
| 4456 |
+
"learning_rate": 4.873294346978558e-06,
|
| 4457 |
+
"loss": 0.2786,
|
| 4458 |
+
"step": 7280
|
| 4459 |
+
},
|
| 4460 |
+
{
|
| 4461 |
+
"epoch": 9.14,
|
| 4462 |
+
"learning_rate": 4.803675856307436e-06,
|
| 4463 |
+
"loss": 0.2568,
|
| 4464 |
+
"step": 7290
|
| 4465 |
+
},
|
| 4466 |
+
{
|
| 4467 |
+
"epoch": 9.15,
|
| 4468 |
+
"learning_rate": 4.734057365636313e-06,
|
| 4469 |
+
"loss": 0.261,
|
| 4470 |
+
"step": 7300
|
| 4471 |
+
},
|
| 4472 |
+
{
|
| 4473 |
+
"epoch": 9.16,
|
| 4474 |
+
"learning_rate": 4.664438874965191e-06,
|
| 4475 |
+
"loss": 0.2548,
|
| 4476 |
+
"step": 7310
|
| 4477 |
+
},
|
| 4478 |
+
{
|
| 4479 |
+
"epoch": 9.17,
|
| 4480 |
+
"learning_rate": 4.5948203842940685e-06,
|
| 4481 |
+
"loss": 0.2698,
|
| 4482 |
+
"step": 7320
|
| 4483 |
+
},
|
| 4484 |
+
{
|
| 4485 |
+
"epoch": 9.19,
|
| 4486 |
+
"learning_rate": 4.5252018936229464e-06,
|
| 4487 |
+
"loss": 0.2965,
|
| 4488 |
+
"step": 7330
|
| 4489 |
+
},
|
| 4490 |
+
{
|
| 4491 |
+
"epoch": 9.2,
|
| 4492 |
+
"learning_rate": 4.455583402951824e-06,
|
| 4493 |
+
"loss": 0.2601,
|
| 4494 |
+
"step": 7340
|
| 4495 |
+
},
|
| 4496 |
+
{
|
| 4497 |
+
"epoch": 9.21,
|
| 4498 |
+
"learning_rate": 4.3859649122807014e-06,
|
| 4499 |
+
"loss": 0.2512,
|
| 4500 |
+
"step": 7350
|
| 4501 |
+
},
|
| 4502 |
+
{
|
| 4503 |
+
"epoch": 9.22,
|
| 4504 |
+
"learning_rate": 4.316346421609579e-06,
|
| 4505 |
+
"loss": 0.3007,
|
| 4506 |
+
"step": 7360
|
| 4507 |
+
},
|
| 4508 |
+
{
|
| 4509 |
+
"epoch": 9.24,
|
| 4510 |
+
"learning_rate": 4.246727930938457e-06,
|
| 4511 |
+
"loss": 0.2669,
|
| 4512 |
+
"step": 7370
|
| 4513 |
+
},
|
| 4514 |
+
{
|
| 4515 |
+
"epoch": 9.25,
|
| 4516 |
+
"learning_rate": 4.177109440267335e-06,
|
| 4517 |
+
"loss": 0.2709,
|
| 4518 |
+
"step": 7380
|
| 4519 |
+
},
|
| 4520 |
+
{
|
| 4521 |
+
"epoch": 9.26,
|
| 4522 |
+
"learning_rate": 4.107490949596213e-06,
|
| 4523 |
+
"loss": 0.2455,
|
| 4524 |
+
"step": 7390
|
| 4525 |
+
},
|
| 4526 |
+
{
|
| 4527 |
+
"epoch": 9.27,
|
| 4528 |
+
"learning_rate": 4.03787245892509e-06,
|
| 4529 |
+
"loss": 0.2737,
|
| 4530 |
+
"step": 7400
|
| 4531 |
+
},
|
| 4532 |
+
{
|
| 4533 |
+
"epoch": 9.29,
|
| 4534 |
+
"learning_rate": 3.968253968253968e-06,
|
| 4535 |
+
"loss": 0.2845,
|
| 4536 |
+
"step": 7410
|
| 4537 |
+
},
|
| 4538 |
+
{
|
| 4539 |
+
"epoch": 9.3,
|
| 4540 |
+
"learning_rate": 3.898635477582846e-06,
|
| 4541 |
+
"loss": 0.2437,
|
| 4542 |
+
"step": 7420
|
| 4543 |
+
},
|
| 4544 |
+
{
|
| 4545 |
+
"epoch": 9.31,
|
| 4546 |
+
"learning_rate": 3.829016986911724e-06,
|
| 4547 |
+
"loss": 0.2716,
|
| 4548 |
+
"step": 7430
|
| 4549 |
+
},
|
| 4550 |
+
{
|
| 4551 |
+
"epoch": 9.32,
|
| 4552 |
+
"learning_rate": 3.7593984962406014e-06,
|
| 4553 |
+
"loss": 0.2764,
|
| 4554 |
+
"step": 7440
|
| 4555 |
+
},
|
| 4556 |
+
{
|
| 4557 |
+
"epoch": 9.34,
|
| 4558 |
+
"learning_rate": 3.6897800055694793e-06,
|
| 4559 |
+
"loss": 0.2819,
|
| 4560 |
+
"step": 7450
|
| 4561 |
+
},
|
| 4562 |
+
{
|
| 4563 |
+
"epoch": 9.35,
|
| 4564 |
+
"learning_rate": 3.620161514898357e-06,
|
| 4565 |
+
"loss": 0.2976,
|
| 4566 |
+
"step": 7460
|
| 4567 |
+
},
|
| 4568 |
+
{
|
| 4569 |
+
"epoch": 9.36,
|
| 4570 |
+
"learning_rate": 3.5505430242272347e-06,
|
| 4571 |
+
"loss": 0.2553,
|
| 4572 |
+
"step": 7470
|
| 4573 |
+
},
|
| 4574 |
+
{
|
| 4575 |
+
"epoch": 9.37,
|
| 4576 |
+
"learning_rate": 3.4809245335561126e-06,
|
| 4577 |
+
"loss": 0.2817,
|
| 4578 |
+
"step": 7480
|
| 4579 |
+
},
|
| 4580 |
+
{
|
| 4581 |
+
"epoch": 9.39,
|
| 4582 |
+
"learning_rate": 3.41130604288499e-06,
|
| 4583 |
+
"loss": 0.2809,
|
| 4584 |
+
"step": 7490
|
| 4585 |
+
},
|
| 4586 |
+
{
|
| 4587 |
+
"epoch": 9.4,
|
| 4588 |
+
"learning_rate": 3.341687552213868e-06,
|
| 4589 |
+
"loss": 0.254,
|
| 4590 |
+
"step": 7500
|
| 4591 |
+
},
|
| 4592 |
+
{
|
| 4593 |
+
"epoch": 9.41,
|
| 4594 |
+
"learning_rate": 3.2720690615427456e-06,
|
| 4595 |
+
"loss": 0.2814,
|
| 4596 |
+
"step": 7510
|
| 4597 |
+
},
|
| 4598 |
+
{
|
| 4599 |
+
"epoch": 9.42,
|
| 4600 |
+
"learning_rate": 3.2024505708716235e-06,
|
| 4601 |
+
"loss": 0.2923,
|
| 4602 |
+
"step": 7520
|
| 4603 |
+
},
|
| 4604 |
+
{
|
| 4605 |
+
"epoch": 9.44,
|
| 4606 |
+
"learning_rate": 3.1328320802005014e-06,
|
| 4607 |
+
"loss": 0.2792,
|
| 4608 |
+
"step": 7530
|
| 4609 |
+
},
|
| 4610 |
+
{
|
| 4611 |
+
"epoch": 9.45,
|
| 4612 |
+
"learning_rate": 3.063213589529379e-06,
|
| 4613 |
+
"loss": 0.2843,
|
| 4614 |
+
"step": 7540
|
| 4615 |
+
},
|
| 4616 |
+
{
|
| 4617 |
+
"epoch": 9.46,
|
| 4618 |
+
"learning_rate": 2.993595098858257e-06,
|
| 4619 |
+
"loss": 0.2724,
|
| 4620 |
+
"step": 7550
|
| 4621 |
+
},
|
| 4622 |
+
{
|
| 4623 |
+
"epoch": 9.47,
|
| 4624 |
+
"learning_rate": 2.9239766081871343e-06,
|
| 4625 |
+
"loss": 0.2918,
|
| 4626 |
+
"step": 7560
|
| 4627 |
+
},
|
| 4628 |
+
{
|
| 4629 |
+
"epoch": 9.49,
|
| 4630 |
+
"learning_rate": 2.854358117516012e-06,
|
| 4631 |
+
"loss": 0.2756,
|
| 4632 |
+
"step": 7570
|
| 4633 |
+
},
|
| 4634 |
+
{
|
| 4635 |
+
"epoch": 9.5,
|
| 4636 |
+
"learning_rate": 2.78473962684489e-06,
|
| 4637 |
+
"loss": 0.303,
|
| 4638 |
+
"step": 7580
|
| 4639 |
+
},
|
| 4640 |
+
{
|
| 4641 |
+
"epoch": 9.51,
|
| 4642 |
+
"learning_rate": 2.7151211361737676e-06,
|
| 4643 |
+
"loss": 0.2556,
|
| 4644 |
+
"step": 7590
|
| 4645 |
+
},
|
| 4646 |
+
{
|
| 4647 |
+
"epoch": 9.52,
|
| 4648 |
+
"learning_rate": 2.6455026455026455e-06,
|
| 4649 |
+
"loss": 0.2687,
|
| 4650 |
+
"step": 7600
|
| 4651 |
+
},
|
| 4652 |
+
{
|
| 4653 |
+
"epoch": 9.54,
|
| 4654 |
+
"learning_rate": 2.575884154831523e-06,
|
| 4655 |
+
"loss": 0.2527,
|
| 4656 |
+
"step": 7610
|
| 4657 |
+
},
|
| 4658 |
+
{
|
| 4659 |
+
"epoch": 9.55,
|
| 4660 |
+
"learning_rate": 2.506265664160401e-06,
|
| 4661 |
+
"loss": 0.2665,
|
| 4662 |
+
"step": 7620
|
| 4663 |
+
},
|
| 4664 |
+
{
|
| 4665 |
+
"epoch": 9.56,
|
| 4666 |
+
"learning_rate": 2.436647173489279e-06,
|
| 4667 |
+
"loss": 0.287,
|
| 4668 |
+
"step": 7630
|
| 4669 |
+
},
|
| 4670 |
+
{
|
| 4671 |
+
"epoch": 9.57,
|
| 4672 |
+
"learning_rate": 2.3670286828181563e-06,
|
| 4673 |
+
"loss": 0.3368,
|
| 4674 |
+
"step": 7640
|
| 4675 |
+
},
|
| 4676 |
+
{
|
| 4677 |
+
"epoch": 9.59,
|
| 4678 |
+
"learning_rate": 2.2974101921470343e-06,
|
| 4679 |
+
"loss": 0.2389,
|
| 4680 |
+
"step": 7650
|
| 4681 |
+
},
|
| 4682 |
+
{
|
| 4683 |
+
"epoch": 9.6,
|
| 4684 |
+
"learning_rate": 2.227791701475912e-06,
|
| 4685 |
+
"loss": 0.2775,
|
| 4686 |
+
"step": 7660
|
| 4687 |
+
},
|
| 4688 |
+
{
|
| 4689 |
+
"epoch": 9.61,
|
| 4690 |
+
"learning_rate": 2.1581732108047897e-06,
|
| 4691 |
+
"loss": 0.2854,
|
| 4692 |
+
"step": 7670
|
| 4693 |
+
},
|
| 4694 |
+
{
|
| 4695 |
+
"epoch": 9.62,
|
| 4696 |
+
"learning_rate": 2.0885547201336676e-06,
|
| 4697 |
+
"loss": 0.2705,
|
| 4698 |
+
"step": 7680
|
| 4699 |
+
},
|
| 4700 |
+
{
|
| 4701 |
+
"epoch": 9.64,
|
| 4702 |
+
"learning_rate": 2.018936229462545e-06,
|
| 4703 |
+
"loss": 0.2666,
|
| 4704 |
+
"step": 7690
|
| 4705 |
+
},
|
| 4706 |
+
{
|
| 4707 |
+
"epoch": 9.65,
|
| 4708 |
+
"learning_rate": 1.949317738791423e-06,
|
| 4709 |
+
"loss": 0.2618,
|
| 4710 |
+
"step": 7700
|
| 4711 |
+
},
|
| 4712 |
+
{
|
| 4713 |
+
"epoch": 9.66,
|
| 4714 |
+
"learning_rate": 1.8796992481203007e-06,
|
| 4715 |
+
"loss": 0.2613,
|
| 4716 |
+
"step": 7710
|
| 4717 |
+
},
|
| 4718 |
+
{
|
| 4719 |
+
"epoch": 9.67,
|
| 4720 |
+
"learning_rate": 1.8100807574491784e-06,
|
| 4721 |
+
"loss": 0.2666,
|
| 4722 |
+
"step": 7720
|
| 4723 |
+
},
|
| 4724 |
+
{
|
| 4725 |
+
"epoch": 9.69,
|
| 4726 |
+
"learning_rate": 1.7404622667780563e-06,
|
| 4727 |
+
"loss": 0.2302,
|
| 4728 |
+
"step": 7730
|
| 4729 |
+
},
|
| 4730 |
+
{
|
| 4731 |
+
"epoch": 9.7,
|
| 4732 |
+
"learning_rate": 1.670843776106934e-06,
|
| 4733 |
+
"loss": 0.2484,
|
| 4734 |
+
"step": 7740
|
| 4735 |
+
},
|
| 4736 |
+
{
|
| 4737 |
+
"epoch": 9.71,
|
| 4738 |
+
"learning_rate": 1.6012252854358117e-06,
|
| 4739 |
+
"loss": 0.2419,
|
| 4740 |
+
"step": 7750
|
| 4741 |
+
},
|
| 4742 |
+
{
|
| 4743 |
+
"epoch": 9.72,
|
| 4744 |
+
"learning_rate": 1.5316067947646894e-06,
|
| 4745 |
+
"loss": 0.2479,
|
| 4746 |
+
"step": 7760
|
| 4747 |
+
},
|
| 4748 |
+
{
|
| 4749 |
+
"epoch": 9.74,
|
| 4750 |
+
"learning_rate": 1.4619883040935671e-06,
|
| 4751 |
+
"loss": 0.2976,
|
| 4752 |
+
"step": 7770
|
| 4753 |
+
},
|
| 4754 |
+
{
|
| 4755 |
+
"epoch": 9.75,
|
| 4756 |
+
"learning_rate": 1.392369813422445e-06,
|
| 4757 |
+
"loss": 0.2675,
|
| 4758 |
+
"step": 7780
|
| 4759 |
+
},
|
| 4760 |
+
{
|
| 4761 |
+
"epoch": 9.76,
|
| 4762 |
+
"learning_rate": 1.3227513227513228e-06,
|
| 4763 |
+
"loss": 0.2993,
|
| 4764 |
+
"step": 7790
|
| 4765 |
+
},
|
| 4766 |
+
{
|
| 4767 |
+
"epoch": 9.77,
|
| 4768 |
+
"learning_rate": 1.2531328320802005e-06,
|
| 4769 |
+
"loss": 0.2943,
|
| 4770 |
+
"step": 7800
|
| 4771 |
+
},
|
| 4772 |
+
{
|
| 4773 |
+
"epoch": 9.79,
|
| 4774 |
+
"learning_rate": 1.1835143414090782e-06,
|
| 4775 |
+
"loss": 0.3028,
|
| 4776 |
+
"step": 7810
|
| 4777 |
+
},
|
| 4778 |
+
{
|
| 4779 |
+
"epoch": 9.8,
|
| 4780 |
+
"learning_rate": 1.113895850737956e-06,
|
| 4781 |
+
"loss": 0.2926,
|
| 4782 |
+
"step": 7820
|
| 4783 |
+
},
|
| 4784 |
+
{
|
| 4785 |
+
"epoch": 9.81,
|
| 4786 |
+
"learning_rate": 1.0442773600668338e-06,
|
| 4787 |
+
"loss": 0.2558,
|
| 4788 |
+
"step": 7830
|
| 4789 |
+
},
|
| 4790 |
+
{
|
| 4791 |
+
"epoch": 9.82,
|
| 4792 |
+
"learning_rate": 9.746588693957115e-07,
|
| 4793 |
+
"loss": 0.2556,
|
| 4794 |
+
"step": 7840
|
| 4795 |
+
},
|
| 4796 |
+
{
|
| 4797 |
+
"epoch": 9.84,
|
| 4798 |
+
"learning_rate": 9.050403787245892e-07,
|
| 4799 |
+
"loss": 0.2531,
|
| 4800 |
+
"step": 7850
|
| 4801 |
+
},
|
| 4802 |
+
{
|
| 4803 |
+
"epoch": 9.85,
|
| 4804 |
+
"learning_rate": 8.35421888053467e-07,
|
| 4805 |
+
"loss": 0.2996,
|
| 4806 |
+
"step": 7860
|
| 4807 |
+
},
|
| 4808 |
+
{
|
| 4809 |
+
"epoch": 9.86,
|
| 4810 |
+
"learning_rate": 7.658033973823447e-07,
|
| 4811 |
+
"loss": 0.2825,
|
| 4812 |
+
"step": 7870
|
| 4813 |
+
},
|
| 4814 |
+
{
|
| 4815 |
+
"epoch": 9.87,
|
| 4816 |
+
"learning_rate": 6.961849067112225e-07,
|
| 4817 |
+
"loss": 0.2828,
|
| 4818 |
+
"step": 7880
|
| 4819 |
+
},
|
| 4820 |
+
{
|
| 4821 |
+
"epoch": 9.89,
|
| 4822 |
+
"learning_rate": 6.265664160401002e-07,
|
| 4823 |
+
"loss": 0.2608,
|
| 4824 |
+
"step": 7890
|
| 4825 |
+
},
|
| 4826 |
+
{
|
| 4827 |
+
"epoch": 9.9,
|
| 4828 |
+
"learning_rate": 5.56947925368978e-07,
|
| 4829 |
+
"loss": 0.2855,
|
| 4830 |
+
"step": 7900
|
| 4831 |
+
},
|
| 4832 |
+
{
|
| 4833 |
+
"epoch": 9.91,
|
| 4834 |
+
"learning_rate": 4.873294346978557e-07,
|
| 4835 |
+
"loss": 0.255,
|
| 4836 |
+
"step": 7910
|
| 4837 |
+
},
|
| 4838 |
+
{
|
| 4839 |
+
"epoch": 9.92,
|
| 4840 |
+
"learning_rate": 4.177109440267335e-07,
|
| 4841 |
+
"loss": 0.2583,
|
| 4842 |
+
"step": 7920
|
| 4843 |
+
},
|
| 4844 |
+
{
|
| 4845 |
+
"epoch": 9.94,
|
| 4846 |
+
"learning_rate": 3.4809245335561126e-07,
|
| 4847 |
+
"loss": 0.2945,
|
| 4848 |
+
"step": 7930
|
| 4849 |
+
},
|
| 4850 |
+
{
|
| 4851 |
+
"epoch": 9.95,
|
| 4852 |
+
"learning_rate": 2.78473962684489e-07,
|
| 4853 |
+
"loss": 0.2671,
|
| 4854 |
+
"step": 7940
|
| 4855 |
+
},
|
| 4856 |
+
{
|
| 4857 |
+
"epoch": 9.96,
|
| 4858 |
+
"learning_rate": 2.0885547201336675e-07,
|
| 4859 |
+
"loss": 0.2847,
|
| 4860 |
+
"step": 7950
|
| 4861 |
+
},
|
| 4862 |
+
{
|
| 4863 |
+
"epoch": 9.97,
|
| 4864 |
+
"learning_rate": 1.392369813422445e-07,
|
| 4865 |
+
"loss": 0.2756,
|
| 4866 |
+
"step": 7960
|
| 4867 |
+
},
|
| 4868 |
+
{
|
| 4869 |
+
"epoch": 9.99,
|
| 4870 |
+
"learning_rate": 6.961849067112226e-08,
|
| 4871 |
+
"loss": 0.2689,
|
| 4872 |
+
"step": 7970
|
| 4873 |
+
},
|
| 4874 |
+
{
|
| 4875 |
+
"epoch": 10.0,
|
| 4876 |
+
"learning_rate": 0.0,
|
| 4877 |
+
"loss": 0.2692,
|
| 4878 |
+
"step": 7980
|
| 4879 |
+
},
|
| 4880 |
+
{
|
| 4881 |
+
"epoch": 10.0,
|
| 4882 |
+
"eval_accuracy": 0.8571428571428571,
|
| 4883 |
+
"eval_loss": 0.4352613687515259,
|
| 4884 |
+
"eval_runtime": 236.1141,
|
| 4885 |
+
"eval_samples_per_second": 108.151,
|
| 4886 |
+
"eval_steps_per_second": 3.38,
|
| 4887 |
+
"step": 7980
|
| 4888 |
+
},
|
| 4889 |
+
{
|
| 4890 |
+
"epoch": 10.0,
|
| 4891 |
+
"step": 7980,
|
| 4892 |
+
"total_flos": 7.915696500716863e+19,
|
| 4893 |
+
"train_loss": 0.14746467811720712,
|
| 4894 |
+
"train_runtime": 9966.943,
|
| 4895 |
+
"train_samples_per_second": 102.483,
|
| 4896 |
+
"train_steps_per_second": 0.801
|
| 4897 |
}
|
| 4898 |
],
|
| 4899 |
"logging_steps": 10,
|
| 4900 |
+
"max_steps": 7980,
|
| 4901 |
"num_input_tokens_seen": 0,
|
| 4902 |
+
"num_train_epochs": 10,
|
| 4903 |
"save_steps": 500,
|
| 4904 |
+
"total_flos": 7.915696500716863e+19,
|
| 4905 |
"train_batch_size": 32,
|
| 4906 |
"trial_name": null,
|
| 4907 |
"trial_params": null
|