Training in progress, epoch 1
Browse files- all_results.json +15 -0
- eval_results.json +10 -0
- pytorch_model.bin +1 -1
- train_results.json +7 -0
- trainer_state.json +1834 -0
- training_args.bin +1 -1
all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 9.97,
|
| 3 |
+
"eval_accuracy": 0.9787045252883763,
|
| 4 |
+
"eval_f1": 0.9828159670126178,
|
| 5 |
+
"eval_loss": 0.08574510365724564,
|
| 6 |
+
"eval_precision": 0.9810208611887528,
|
| 7 |
+
"eval_recall": 0.9847829960486251,
|
| 8 |
+
"eval_runtime": 39.2547,
|
| 9 |
+
"eval_samples_per_second": 28.71,
|
| 10 |
+
"eval_steps_per_second": 3.592,
|
| 11 |
+
"train_loss": 0.16025869953462663,
|
| 12 |
+
"train_runtime": 8523.7997,
|
| 13 |
+
"train_samples_per_second": 10.57,
|
| 14 |
+
"train_steps_per_second": 0.33
|
| 15 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"eval_accuracy": 0.9787045252883763,
|
| 3 |
+
"eval_f1": 0.9828159670126178,
|
| 4 |
+
"eval_loss": 0.08574510365724564,
|
| 5 |
+
"eval_precision": 0.9810208611887528,
|
| 6 |
+
"eval_recall": 0.9847829960486251,
|
| 7 |
+
"eval_runtime": 39.2547,
|
| 8 |
+
"eval_samples_per_second": 28.71,
|
| 9 |
+
"eval_steps_per_second": 3.592
|
| 10 |
+
}
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 350940477
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6f238e5bd1d3666ac49e93bb760594057d588c021bcafa5461f8e017d3ab5174
|
| 3 |
size 350940477
|
train_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 9.97,
|
| 3 |
+
"train_loss": 0.16025869953462663,
|
| 4 |
+
"train_runtime": 8523.7997,
|
| 5 |
+
"train_samples_per_second": 10.57,
|
| 6 |
+
"train_steps_per_second": 0.33
|
| 7 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1834 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9857904085257548,
|
| 3 |
+
"best_model_checkpoint": "10-finetuned-ausSpiders2000/checkpoint-2535",
|
| 4 |
+
"epoch": 9.97338065661047,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 2810,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.04,
|
| 13 |
+
"learning_rate": 1.7793594306049825e-05,
|
| 14 |
+
"loss": 2.4863,
|
| 15 |
+
"step": 10
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.07,
|
| 19 |
+
"learning_rate": 3.558718861209965e-05,
|
| 20 |
+
"loss": 2.1753,
|
| 21 |
+
"step": 20
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.11,
|
| 25 |
+
"learning_rate": 5.338078291814947e-05,
|
| 26 |
+
"loss": 1.6356,
|
| 27 |
+
"step": 30
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.14,
|
| 31 |
+
"learning_rate": 7.11743772241993e-05,
|
| 32 |
+
"loss": 0.9598,
|
| 33 |
+
"step": 40
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.18,
|
| 37 |
+
"learning_rate": 8.896797153024912e-05,
|
| 38 |
+
"loss": 0.5263,
|
| 39 |
+
"step": 50
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.21,
|
| 43 |
+
"learning_rate": 0.00010676156583629893,
|
| 44 |
+
"loss": 0.2692,
|
| 45 |
+
"step": 60
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.25,
|
| 49 |
+
"learning_rate": 0.00012455516014234875,
|
| 50 |
+
"loss": 0.233,
|
| 51 |
+
"step": 70
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.28,
|
| 55 |
+
"learning_rate": 0.0001423487544483986,
|
| 56 |
+
"loss": 0.1485,
|
| 57 |
+
"step": 80
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.32,
|
| 61 |
+
"learning_rate": 0.00016014234875444842,
|
| 62 |
+
"loss": 0.148,
|
| 63 |
+
"step": 90
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.35,
|
| 67 |
+
"learning_rate": 0.00017793594306049823,
|
| 68 |
+
"loss": 0.1282,
|
| 69 |
+
"step": 100
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.39,
|
| 73 |
+
"learning_rate": 0.00019572953736654805,
|
| 74 |
+
"loss": 0.1211,
|
| 75 |
+
"step": 110
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.43,
|
| 79 |
+
"learning_rate": 0.00021352313167259787,
|
| 80 |
+
"loss": 0.1153,
|
| 81 |
+
"step": 120
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.46,
|
| 85 |
+
"learning_rate": 0.0002313167259786477,
|
| 86 |
+
"loss": 0.1512,
|
| 87 |
+
"step": 130
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.5,
|
| 91 |
+
"learning_rate": 0.0002491103202846975,
|
| 92 |
+
"loss": 0.1465,
|
| 93 |
+
"step": 140
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.53,
|
| 97 |
+
"learning_rate": 0.0002669039145907473,
|
| 98 |
+
"loss": 0.1391,
|
| 99 |
+
"step": 150
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.57,
|
| 103 |
+
"learning_rate": 0.0002846975088967972,
|
| 104 |
+
"loss": 0.099,
|
| 105 |
+
"step": 160
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.6,
|
| 109 |
+
"learning_rate": 0.000302491103202847,
|
| 110 |
+
"loss": 0.1788,
|
| 111 |
+
"step": 170
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.64,
|
| 115 |
+
"learning_rate": 0.00032028469750889683,
|
| 116 |
+
"loss": 0.1824,
|
| 117 |
+
"step": 180
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.67,
|
| 121 |
+
"learning_rate": 0.0003380782918149466,
|
| 122 |
+
"loss": 0.1867,
|
| 123 |
+
"step": 190
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.71,
|
| 127 |
+
"learning_rate": 0.00035587188612099647,
|
| 128 |
+
"loss": 0.1699,
|
| 129 |
+
"step": 200
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.75,
|
| 133 |
+
"learning_rate": 0.0003736654804270463,
|
| 134 |
+
"loss": 0.1658,
|
| 135 |
+
"step": 210
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.78,
|
| 139 |
+
"learning_rate": 0.0003914590747330961,
|
| 140 |
+
"loss": 0.1465,
|
| 141 |
+
"step": 220
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.82,
|
| 145 |
+
"learning_rate": 0.00040925266903914595,
|
| 146 |
+
"loss": 0.173,
|
| 147 |
+
"step": 230
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 0.85,
|
| 151 |
+
"learning_rate": 0.00042704626334519574,
|
| 152 |
+
"loss": 0.2829,
|
| 153 |
+
"step": 240
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.89,
|
| 157 |
+
"learning_rate": 0.0004448398576512456,
|
| 158 |
+
"loss": 0.1879,
|
| 159 |
+
"step": 250
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.92,
|
| 163 |
+
"learning_rate": 0.0004626334519572954,
|
| 164 |
+
"loss": 0.2341,
|
| 165 |
+
"step": 260
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.96,
|
| 169 |
+
"learning_rate": 0.0004804270462633452,
|
| 170 |
+
"loss": 0.2539,
|
| 171 |
+
"step": 270
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.99,
|
| 175 |
+
"learning_rate": 0.000498220640569395,
|
| 176 |
+
"loss": 0.1997,
|
| 177 |
+
"step": 280
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 1.0,
|
| 181 |
+
"eval_accuracy": 0.9280639431616341,
|
| 182 |
+
"eval_f1": 0.9044835301236254,
|
| 183 |
+
"eval_loss": 0.2264249473810196,
|
| 184 |
+
"eval_precision": 0.9395261094135103,
|
| 185 |
+
"eval_recall": 0.8868521981209923,
|
| 186 |
+
"eval_runtime": 40.9312,
|
| 187 |
+
"eval_samples_per_second": 27.51,
|
| 188 |
+
"eval_steps_per_second": 3.445,
|
| 189 |
+
"step": 281
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 1.03,
|
| 193 |
+
"learning_rate": 0.000498220640569395,
|
| 194 |
+
"loss": 0.1887,
|
| 195 |
+
"step": 290
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 1.06,
|
| 199 |
+
"learning_rate": 0.0004962435745353895,
|
| 200 |
+
"loss": 0.227,
|
| 201 |
+
"step": 300
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 1.1,
|
| 205 |
+
"learning_rate": 0.0004942665085013839,
|
| 206 |
+
"loss": 0.1633,
|
| 207 |
+
"step": 310
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 1.14,
|
| 211 |
+
"learning_rate": 0.0004922894424673784,
|
| 212 |
+
"loss": 0.2611,
|
| 213 |
+
"step": 320
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 1.17,
|
| 217 |
+
"learning_rate": 0.0004903123764333729,
|
| 218 |
+
"loss": 0.1793,
|
| 219 |
+
"step": 330
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 1.21,
|
| 223 |
+
"learning_rate": 0.0004883353103993673,
|
| 224 |
+
"loss": 0.1217,
|
| 225 |
+
"step": 340
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 1.24,
|
| 229 |
+
"learning_rate": 0.0004863582443653618,
|
| 230 |
+
"loss": 0.2545,
|
| 231 |
+
"step": 350
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 1.28,
|
| 235 |
+
"learning_rate": 0.00048438117833135623,
|
| 236 |
+
"loss": 0.3059,
|
| 237 |
+
"step": 360
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 1.31,
|
| 241 |
+
"learning_rate": 0.00048240411229735075,
|
| 242 |
+
"loss": 0.2185,
|
| 243 |
+
"step": 370
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 1.35,
|
| 247 |
+
"learning_rate": 0.0004804270462633452,
|
| 248 |
+
"loss": 0.1968,
|
| 249 |
+
"step": 380
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 1.38,
|
| 253 |
+
"learning_rate": 0.0004784499802293397,
|
| 254 |
+
"loss": 0.2301,
|
| 255 |
+
"step": 390
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 1.42,
|
| 259 |
+
"learning_rate": 0.0004764729141953341,
|
| 260 |
+
"loss": 0.3476,
|
| 261 |
+
"step": 400
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 1.46,
|
| 265 |
+
"learning_rate": 0.0004744958481613286,
|
| 266 |
+
"loss": 0.2723,
|
| 267 |
+
"step": 410
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 1.49,
|
| 271 |
+
"learning_rate": 0.0004725187821273231,
|
| 272 |
+
"loss": 0.2253,
|
| 273 |
+
"step": 420
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 1.53,
|
| 277 |
+
"learning_rate": 0.0004705417160933175,
|
| 278 |
+
"loss": 0.1927,
|
| 279 |
+
"step": 430
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 1.56,
|
| 283 |
+
"learning_rate": 0.00046856465005931197,
|
| 284 |
+
"loss": 0.2396,
|
| 285 |
+
"step": 440
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 1.6,
|
| 289 |
+
"learning_rate": 0.00046658758402530644,
|
| 290 |
+
"loss": 0.2278,
|
| 291 |
+
"step": 450
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 1.63,
|
| 295 |
+
"learning_rate": 0.00046461051799130096,
|
| 296 |
+
"loss": 0.275,
|
| 297 |
+
"step": 460
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 1.67,
|
| 301 |
+
"learning_rate": 0.0004626334519572954,
|
| 302 |
+
"loss": 0.1884,
|
| 303 |
+
"step": 470
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 1.7,
|
| 307 |
+
"learning_rate": 0.00046065638592328984,
|
| 308 |
+
"loss": 0.2211,
|
| 309 |
+
"step": 480
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 1.74,
|
| 313 |
+
"learning_rate": 0.0004586793198892843,
|
| 314 |
+
"loss": 0.2202,
|
| 315 |
+
"step": 490
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 1.77,
|
| 319 |
+
"learning_rate": 0.0004567022538552787,
|
| 320 |
+
"loss": 0.2851,
|
| 321 |
+
"step": 500
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 1.81,
|
| 325 |
+
"learning_rate": 0.00045472518782127324,
|
| 326 |
+
"loss": 0.2128,
|
| 327 |
+
"step": 510
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 1.85,
|
| 331 |
+
"learning_rate": 0.0004527481217872677,
|
| 332 |
+
"loss": 0.2074,
|
| 333 |
+
"step": 520
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 1.88,
|
| 337 |
+
"learning_rate": 0.0004507710557532622,
|
| 338 |
+
"loss": 0.2324,
|
| 339 |
+
"step": 530
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 1.92,
|
| 343 |
+
"learning_rate": 0.0004487939897192566,
|
| 344 |
+
"loss": 0.1803,
|
| 345 |
+
"step": 540
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 1.95,
|
| 349 |
+
"learning_rate": 0.0004468169236852511,
|
| 350 |
+
"loss": 0.209,
|
| 351 |
+
"step": 550
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 1.99,
|
| 355 |
+
"learning_rate": 0.0004448398576512456,
|
| 356 |
+
"loss": 0.17,
|
| 357 |
+
"step": 560
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 2.0,
|
| 361 |
+
"eval_accuracy": 0.9564831261101243,
|
| 362 |
+
"eval_f1": 0.8380915867647825,
|
| 363 |
+
"eval_loss": 0.13818518817424774,
|
| 364 |
+
"eval_precision": 0.8539785317750744,
|
| 365 |
+
"eval_recall": 0.8265891631354911,
|
| 366 |
+
"eval_runtime": 39.8834,
|
| 367 |
+
"eval_samples_per_second": 28.232,
|
| 368 |
+
"eval_steps_per_second": 3.535,
|
| 369 |
+
"step": 563
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 2.02,
|
| 373 |
+
"learning_rate": 0.00044286279161724005,
|
| 374 |
+
"loss": 0.2291,
|
| 375 |
+
"step": 570
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 2.06,
|
| 379 |
+
"learning_rate": 0.00044088572558323446,
|
| 380 |
+
"loss": 0.1695,
|
| 381 |
+
"step": 580
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 2.09,
|
| 385 |
+
"learning_rate": 0.00043890865954922893,
|
| 386 |
+
"loss": 0.2086,
|
| 387 |
+
"step": 590
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 2.13,
|
| 391 |
+
"learning_rate": 0.00043693159351522345,
|
| 392 |
+
"loss": 0.1849,
|
| 393 |
+
"step": 600
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 2.17,
|
| 397 |
+
"learning_rate": 0.00043495452748121787,
|
| 398 |
+
"loss": 0.2532,
|
| 399 |
+
"step": 610
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 2.2,
|
| 403 |
+
"learning_rate": 0.00043297746144721233,
|
| 404 |
+
"loss": 0.2137,
|
| 405 |
+
"step": 620
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 2.24,
|
| 409 |
+
"learning_rate": 0.0004310003954132068,
|
| 410 |
+
"loss": 0.1389,
|
| 411 |
+
"step": 630
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 2.27,
|
| 415 |
+
"learning_rate": 0.0004290233293792013,
|
| 416 |
+
"loss": 0.2308,
|
| 417 |
+
"step": 640
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 2.31,
|
| 421 |
+
"learning_rate": 0.00042704626334519574,
|
| 422 |
+
"loss": 0.1713,
|
| 423 |
+
"step": 650
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 2.34,
|
| 427 |
+
"learning_rate": 0.0004250691973111902,
|
| 428 |
+
"loss": 0.1593,
|
| 429 |
+
"step": 660
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 2.38,
|
| 433 |
+
"learning_rate": 0.0004230921312771847,
|
| 434 |
+
"loss": 0.199,
|
| 435 |
+
"step": 670
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 2.41,
|
| 439 |
+
"learning_rate": 0.0004211150652431791,
|
| 440 |
+
"loss": 0.2582,
|
| 441 |
+
"step": 680
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 2.45,
|
| 445 |
+
"learning_rate": 0.0004191379992091736,
|
| 446 |
+
"loss": 0.182,
|
| 447 |
+
"step": 690
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 2.48,
|
| 451 |
+
"learning_rate": 0.0004171609331751681,
|
| 452 |
+
"loss": 0.2524,
|
| 453 |
+
"step": 700
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 2.52,
|
| 457 |
+
"learning_rate": 0.00041518386714116254,
|
| 458 |
+
"loss": 0.1836,
|
| 459 |
+
"step": 710
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 2.56,
|
| 463 |
+
"learning_rate": 0.00041320680110715696,
|
| 464 |
+
"loss": 0.1668,
|
| 465 |
+
"step": 720
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 2.59,
|
| 469 |
+
"learning_rate": 0.0004112297350731515,
|
| 470 |
+
"loss": 0.2326,
|
| 471 |
+
"step": 730
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 2.63,
|
| 475 |
+
"learning_rate": 0.00040925266903914595,
|
| 476 |
+
"loss": 0.1824,
|
| 477 |
+
"step": 740
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 2.66,
|
| 481 |
+
"learning_rate": 0.00040727560300514036,
|
| 482 |
+
"loss": 0.1817,
|
| 483 |
+
"step": 750
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 2.7,
|
| 487 |
+
"learning_rate": 0.00040529853697113483,
|
| 488 |
+
"loss": 0.1547,
|
| 489 |
+
"step": 760
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 2.73,
|
| 493 |
+
"learning_rate": 0.0004033214709371293,
|
| 494 |
+
"loss": 0.1652,
|
| 495 |
+
"step": 770
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 2.77,
|
| 499 |
+
"learning_rate": 0.0004013444049031238,
|
| 500 |
+
"loss": 0.2278,
|
| 501 |
+
"step": 780
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 2.8,
|
| 505 |
+
"learning_rate": 0.00039936733886911823,
|
| 506 |
+
"loss": 0.2369,
|
| 507 |
+
"step": 790
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 2.84,
|
| 511 |
+
"learning_rate": 0.0003973902728351127,
|
| 512 |
+
"loss": 0.1947,
|
| 513 |
+
"step": 800
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 2.87,
|
| 517 |
+
"learning_rate": 0.00039541320680110717,
|
| 518 |
+
"loss": 0.1326,
|
| 519 |
+
"step": 810
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 2.91,
|
| 523 |
+
"learning_rate": 0.0003934361407671016,
|
| 524 |
+
"loss": 0.1644,
|
| 525 |
+
"step": 820
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 2.95,
|
| 529 |
+
"learning_rate": 0.0003914590747330961,
|
| 530 |
+
"loss": 0.1888,
|
| 531 |
+
"step": 830
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 2.98,
|
| 535 |
+
"learning_rate": 0.00038948200869909057,
|
| 536 |
+
"loss": 0.21,
|
| 537 |
+
"step": 840
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 3.0,
|
| 541 |
+
"eval_accuracy": 0.9582593250444049,
|
| 542 |
+
"eval_f1": 0.9348978364444572,
|
| 543 |
+
"eval_loss": 0.14037781953811646,
|
| 544 |
+
"eval_precision": 0.9746987945146139,
|
| 545 |
+
"eval_recall": 0.9063700900026892,
|
| 546 |
+
"eval_runtime": 39.7299,
|
| 547 |
+
"eval_samples_per_second": 28.341,
|
| 548 |
+
"eval_steps_per_second": 3.549,
|
| 549 |
+
"step": 845
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 3.02,
|
| 553 |
+
"learning_rate": 0.00038750494266508504,
|
| 554 |
+
"loss": 0.129,
|
| 555 |
+
"step": 850
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 3.05,
|
| 559 |
+
"learning_rate": 0.00038552787663107945,
|
| 560 |
+
"loss": 0.1675,
|
| 561 |
+
"step": 860
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 3.09,
|
| 565 |
+
"learning_rate": 0.000383550810597074,
|
| 566 |
+
"loss": 0.1136,
|
| 567 |
+
"step": 870
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 3.12,
|
| 571 |
+
"learning_rate": 0.00038157374456306844,
|
| 572 |
+
"loss": 0.1937,
|
| 573 |
+
"step": 880
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 3.16,
|
| 577 |
+
"learning_rate": 0.00037959667852906285,
|
| 578 |
+
"loss": 0.1219,
|
| 579 |
+
"step": 890
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 3.19,
|
| 583 |
+
"learning_rate": 0.0003776196124950573,
|
| 584 |
+
"loss": 0.1394,
|
| 585 |
+
"step": 900
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 3.23,
|
| 589 |
+
"learning_rate": 0.0003756425464610518,
|
| 590 |
+
"loss": 0.1505,
|
| 591 |
+
"step": 910
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 3.27,
|
| 595 |
+
"learning_rate": 0.0003736654804270463,
|
| 596 |
+
"loss": 0.1487,
|
| 597 |
+
"step": 920
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 3.3,
|
| 601 |
+
"learning_rate": 0.0003716884143930407,
|
| 602 |
+
"loss": 0.1228,
|
| 603 |
+
"step": 930
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 3.34,
|
| 607 |
+
"learning_rate": 0.0003697113483590352,
|
| 608 |
+
"loss": 0.185,
|
| 609 |
+
"step": 940
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 3.37,
|
| 613 |
+
"learning_rate": 0.00036773428232502966,
|
| 614 |
+
"loss": 0.1313,
|
| 615 |
+
"step": 950
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 3.41,
|
| 619 |
+
"learning_rate": 0.00036575721629102413,
|
| 620 |
+
"loss": 0.1977,
|
| 621 |
+
"step": 960
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 3.44,
|
| 625 |
+
"learning_rate": 0.0003637801502570186,
|
| 626 |
+
"loss": 0.2188,
|
| 627 |
+
"step": 970
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 3.48,
|
| 631 |
+
"learning_rate": 0.00036180308422301306,
|
| 632 |
+
"loss": 0.1717,
|
| 633 |
+
"step": 980
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 3.51,
|
| 637 |
+
"learning_rate": 0.00035982601818900753,
|
| 638 |
+
"loss": 0.1357,
|
| 639 |
+
"step": 990
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 3.55,
|
| 643 |
+
"learning_rate": 0.00035784895215500194,
|
| 644 |
+
"loss": 0.1386,
|
| 645 |
+
"step": 1000
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 3.58,
|
| 649 |
+
"learning_rate": 0.00035587188612099647,
|
| 650 |
+
"loss": 0.169,
|
| 651 |
+
"step": 1010
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 3.62,
|
| 655 |
+
"learning_rate": 0.00035389482008699093,
|
| 656 |
+
"loss": 0.164,
|
| 657 |
+
"step": 1020
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 3.66,
|
| 661 |
+
"learning_rate": 0.00035191775405298535,
|
| 662 |
+
"loss": 0.1334,
|
| 663 |
+
"step": 1030
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 3.69,
|
| 667 |
+
"learning_rate": 0.0003499406880189798,
|
| 668 |
+
"loss": 0.1634,
|
| 669 |
+
"step": 1040
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 3.73,
|
| 673 |
+
"learning_rate": 0.00034796362198497434,
|
| 674 |
+
"loss": 0.1614,
|
| 675 |
+
"step": 1050
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 3.76,
|
| 679 |
+
"learning_rate": 0.0003459865559509688,
|
| 680 |
+
"loss": 0.1819,
|
| 681 |
+
"step": 1060
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 3.8,
|
| 685 |
+
"learning_rate": 0.0003440094899169632,
|
| 686 |
+
"loss": 0.1796,
|
| 687 |
+
"step": 1070
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 3.83,
|
| 691 |
+
"learning_rate": 0.0003420324238829577,
|
| 692 |
+
"loss": 0.1341,
|
| 693 |
+
"step": 1080
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 3.87,
|
| 697 |
+
"learning_rate": 0.00034005535784895215,
|
| 698 |
+
"loss": 0.127,
|
| 699 |
+
"step": 1090
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 3.9,
|
| 703 |
+
"learning_rate": 0.0003380782918149466,
|
| 704 |
+
"loss": 0.1709,
|
| 705 |
+
"step": 1100
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 3.94,
|
| 709 |
+
"learning_rate": 0.0003361012257809411,
|
| 710 |
+
"loss": 0.2003,
|
| 711 |
+
"step": 1110
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 3.98,
|
| 715 |
+
"learning_rate": 0.00033412415974693556,
|
| 716 |
+
"loss": 0.1976,
|
| 717 |
+
"step": 1120
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 4.0,
|
| 721 |
+
"eval_accuracy": 0.9689165186500888,
|
| 722 |
+
"eval_f1": 0.912801315494264,
|
| 723 |
+
"eval_loss": 0.09873761981725693,
|
| 724 |
+
"eval_precision": 0.9716016779252258,
|
| 725 |
+
"eval_recall": 0.8917025082468766,
|
| 726 |
+
"eval_runtime": 39.8355,
|
| 727 |
+
"eval_samples_per_second": 28.266,
|
| 728 |
+
"eval_steps_per_second": 3.54,
|
| 729 |
+
"step": 1127
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 4.01,
|
| 733 |
+
"learning_rate": 0.00033214709371293,
|
| 734 |
+
"loss": 0.1981,
|
| 735 |
+
"step": 1130
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 4.05,
|
| 739 |
+
"learning_rate": 0.00033017002767892444,
|
| 740 |
+
"loss": 0.1504,
|
| 741 |
+
"step": 1140
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 4.08,
|
| 745 |
+
"learning_rate": 0.00032819296164491896,
|
| 746 |
+
"loss": 0.1655,
|
| 747 |
+
"step": 1150
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 4.12,
|
| 751 |
+
"learning_rate": 0.00032621589561091343,
|
| 752 |
+
"loss": 0.0908,
|
| 753 |
+
"step": 1160
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 4.15,
|
| 757 |
+
"learning_rate": 0.00032423882957690784,
|
| 758 |
+
"loss": 0.1043,
|
| 759 |
+
"step": 1170
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 4.19,
|
| 763 |
+
"learning_rate": 0.0003222617635429023,
|
| 764 |
+
"loss": 0.1908,
|
| 765 |
+
"step": 1180
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 4.22,
|
| 769 |
+
"learning_rate": 0.00032028469750889683,
|
| 770 |
+
"loss": 0.1202,
|
| 771 |
+
"step": 1190
|
| 772 |
+
},
|
| 773 |
+
{
|
| 774 |
+
"epoch": 4.26,
|
| 775 |
+
"learning_rate": 0.0003183076314748913,
|
| 776 |
+
"loss": 0.1597,
|
| 777 |
+
"step": 1200
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 4.29,
|
| 781 |
+
"learning_rate": 0.0003163305654408857,
|
| 782 |
+
"loss": 0.1318,
|
| 783 |
+
"step": 1210
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 4.33,
|
| 787 |
+
"learning_rate": 0.0003143534994068802,
|
| 788 |
+
"loss": 0.1674,
|
| 789 |
+
"step": 1220
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 4.37,
|
| 793 |
+
"learning_rate": 0.00031237643337287465,
|
| 794 |
+
"loss": 0.1218,
|
| 795 |
+
"step": 1230
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 4.4,
|
| 799 |
+
"learning_rate": 0.00031039936733886917,
|
| 800 |
+
"loss": 0.156,
|
| 801 |
+
"step": 1240
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 4.44,
|
| 805 |
+
"learning_rate": 0.0003084223013048636,
|
| 806 |
+
"loss": 0.1314,
|
| 807 |
+
"step": 1250
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 4.47,
|
| 811 |
+
"learning_rate": 0.00030644523527085805,
|
| 812 |
+
"loss": 0.1817,
|
| 813 |
+
"step": 1260
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 4.51,
|
| 817 |
+
"learning_rate": 0.0003044681692368525,
|
| 818 |
+
"loss": 0.187,
|
| 819 |
+
"step": 1270
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 4.54,
|
| 823 |
+
"learning_rate": 0.000302491103202847,
|
| 824 |
+
"loss": 0.1318,
|
| 825 |
+
"step": 1280
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 4.58,
|
| 829 |
+
"learning_rate": 0.00030051403716884145,
|
| 830 |
+
"loss": 0.1806,
|
| 831 |
+
"step": 1290
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 4.61,
|
| 835 |
+
"learning_rate": 0.0002985369711348359,
|
| 836 |
+
"loss": 0.1614,
|
| 837 |
+
"step": 1300
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 4.65,
|
| 841 |
+
"learning_rate": 0.0002965599051008304,
|
| 842 |
+
"loss": 0.1221,
|
| 843 |
+
"step": 1310
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 4.69,
|
| 847 |
+
"learning_rate": 0.0002945828390668248,
|
| 848 |
+
"loss": 0.1391,
|
| 849 |
+
"step": 1320
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 4.72,
|
| 853 |
+
"learning_rate": 0.0002926057730328193,
|
| 854 |
+
"loss": 0.1275,
|
| 855 |
+
"step": 1330
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 4.76,
|
| 859 |
+
"learning_rate": 0.0002906287069988138,
|
| 860 |
+
"loss": 0.1265,
|
| 861 |
+
"step": 1340
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 4.79,
|
| 865 |
+
"learning_rate": 0.0002886516409648082,
|
| 866 |
+
"loss": 0.1637,
|
| 867 |
+
"step": 1350
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 4.83,
|
| 871 |
+
"learning_rate": 0.0002866745749308027,
|
| 872 |
+
"loss": 0.1833,
|
| 873 |
+
"step": 1360
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 4.86,
|
| 877 |
+
"learning_rate": 0.0002846975088967972,
|
| 878 |
+
"loss": 0.1235,
|
| 879 |
+
"step": 1370
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 4.9,
|
| 883 |
+
"learning_rate": 0.00028272044286279166,
|
| 884 |
+
"loss": 0.0997,
|
| 885 |
+
"step": 1380
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 4.93,
|
| 889 |
+
"learning_rate": 0.0002807433768287861,
|
| 890 |
+
"loss": 0.0879,
|
| 891 |
+
"step": 1390
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 4.97,
|
| 895 |
+
"learning_rate": 0.00027876631079478054,
|
| 896 |
+
"loss": 0.178,
|
| 897 |
+
"step": 1400
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 5.0,
|
| 901 |
+
"eval_accuracy": 0.9635879218472468,
|
| 902 |
+
"eval_f1": 0.9471608472683106,
|
| 903 |
+
"eval_loss": 0.12194248288869858,
|
| 904 |
+
"eval_precision": 0.9393499332363083,
|
| 905 |
+
"eval_recall": 0.9600466296790979,
|
| 906 |
+
"eval_runtime": 39.8939,
|
| 907 |
+
"eval_samples_per_second": 28.225,
|
| 908 |
+
"eval_steps_per_second": 3.534,
|
| 909 |
+
"step": 1408
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 5.0,
|
| 913 |
+
"learning_rate": 0.000276789244760775,
|
| 914 |
+
"loss": 0.0922,
|
| 915 |
+
"step": 1410
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 5.04,
|
| 919 |
+
"learning_rate": 0.0002748121787267695,
|
| 920 |
+
"loss": 0.0684,
|
| 921 |
+
"step": 1420
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 5.08,
|
| 925 |
+
"learning_rate": 0.00027283511269276395,
|
| 926 |
+
"loss": 0.1164,
|
| 927 |
+
"step": 1430
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 5.11,
|
| 931 |
+
"learning_rate": 0.0002708580466587584,
|
| 932 |
+
"loss": 0.0982,
|
| 933 |
+
"step": 1440
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 5.15,
|
| 937 |
+
"learning_rate": 0.0002688809806247529,
|
| 938 |
+
"loss": 0.1041,
|
| 939 |
+
"step": 1450
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 5.18,
|
| 943 |
+
"learning_rate": 0.0002669039145907473,
|
| 944 |
+
"loss": 0.1143,
|
| 945 |
+
"step": 1460
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 5.22,
|
| 949 |
+
"learning_rate": 0.0002649268485567418,
|
| 950 |
+
"loss": 0.1927,
|
| 951 |
+
"step": 1470
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 5.25,
|
| 955 |
+
"learning_rate": 0.0002629497825227363,
|
| 956 |
+
"loss": 0.1207,
|
| 957 |
+
"step": 1480
|
| 958 |
+
},
|
| 959 |
+
{
|
| 960 |
+
"epoch": 5.29,
|
| 961 |
+
"learning_rate": 0.0002609727164887307,
|
| 962 |
+
"loss": 0.1076,
|
| 963 |
+
"step": 1490
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 5.32,
|
| 967 |
+
"learning_rate": 0.00025899565045472517,
|
| 968 |
+
"loss": 0.1068,
|
| 969 |
+
"step": 1500
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 5.36,
|
| 973 |
+
"learning_rate": 0.0002570185844207197,
|
| 974 |
+
"loss": 0.1254,
|
| 975 |
+
"step": 1510
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 5.39,
|
| 979 |
+
"learning_rate": 0.00025504151838671416,
|
| 980 |
+
"loss": 0.0679,
|
| 981 |
+
"step": 1520
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 5.43,
|
| 985 |
+
"learning_rate": 0.00025306445235270857,
|
| 986 |
+
"loss": 0.12,
|
| 987 |
+
"step": 1530
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 5.47,
|
| 991 |
+
"learning_rate": 0.00025108738631870304,
|
| 992 |
+
"loss": 0.1157,
|
| 993 |
+
"step": 1540
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 5.5,
|
| 997 |
+
"learning_rate": 0.0002491103202846975,
|
| 998 |
+
"loss": 0.1034,
|
| 999 |
+
"step": 1550
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 5.54,
|
| 1003 |
+
"learning_rate": 0.00024713325425069197,
|
| 1004 |
+
"loss": 0.1317,
|
| 1005 |
+
"step": 1560
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 5.57,
|
| 1009 |
+
"learning_rate": 0.00024515618821668644,
|
| 1010 |
+
"loss": 0.0942,
|
| 1011 |
+
"step": 1570
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 5.61,
|
| 1015 |
+
"learning_rate": 0.0002431791221826809,
|
| 1016 |
+
"loss": 0.1381,
|
| 1017 |
+
"step": 1580
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 5.64,
|
| 1021 |
+
"learning_rate": 0.00024120205614867538,
|
| 1022 |
+
"loss": 0.1109,
|
| 1023 |
+
"step": 1590
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 5.68,
|
| 1027 |
+
"learning_rate": 0.00023922499011466984,
|
| 1028 |
+
"loss": 0.1132,
|
| 1029 |
+
"step": 1600
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 5.71,
|
| 1033 |
+
"learning_rate": 0.0002372479240806643,
|
| 1034 |
+
"loss": 0.0736,
|
| 1035 |
+
"step": 1610
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 5.75,
|
| 1039 |
+
"learning_rate": 0.00023527085804665875,
|
| 1040 |
+
"loss": 0.1418,
|
| 1041 |
+
"step": 1620
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 5.79,
|
| 1045 |
+
"learning_rate": 0.00023329379201265322,
|
| 1046 |
+
"loss": 0.1318,
|
| 1047 |
+
"step": 1630
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 5.82,
|
| 1051 |
+
"learning_rate": 0.0002313167259786477,
|
| 1052 |
+
"loss": 0.1313,
|
| 1053 |
+
"step": 1640
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 5.86,
|
| 1057 |
+
"learning_rate": 0.00022933965994464215,
|
| 1058 |
+
"loss": 0.1749,
|
| 1059 |
+
"step": 1650
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 5.89,
|
| 1063 |
+
"learning_rate": 0.00022736259391063662,
|
| 1064 |
+
"loss": 0.1526,
|
| 1065 |
+
"step": 1660
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 5.93,
|
| 1069 |
+
"learning_rate": 0.0002253855278766311,
|
| 1070 |
+
"loss": 0.0911,
|
| 1071 |
+
"step": 1670
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 5.96,
|
| 1075 |
+
"learning_rate": 0.00022340846184262556,
|
| 1076 |
+
"loss": 0.1377,
|
| 1077 |
+
"step": 1680
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 6.0,
|
| 1081 |
+
"learning_rate": 0.00022143139580862003,
|
| 1082 |
+
"loss": 0.0659,
|
| 1083 |
+
"step": 1690
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 6.0,
|
| 1087 |
+
"eval_accuracy": 0.9813499111900533,
|
| 1088 |
+
"eval_f1": 0.9807178013915328,
|
| 1089 |
+
"eval_loss": 0.08037910610437393,
|
| 1090 |
+
"eval_precision": 0.9815439633215112,
|
| 1091 |
+
"eval_recall": 0.9801148444477591,
|
| 1092 |
+
"eval_runtime": 39.9217,
|
| 1093 |
+
"eval_samples_per_second": 28.205,
|
| 1094 |
+
"eval_steps_per_second": 3.532,
|
| 1095 |
+
"step": 1690
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 6.03,
|
| 1099 |
+
"learning_rate": 0.00021945432977461447,
|
| 1100 |
+
"loss": 0.1303,
|
| 1101 |
+
"step": 1700
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 6.07,
|
| 1105 |
+
"learning_rate": 0.00021747726374060893,
|
| 1106 |
+
"loss": 0.0513,
|
| 1107 |
+
"step": 1710
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 6.1,
|
| 1111 |
+
"learning_rate": 0.0002155001977066034,
|
| 1112 |
+
"loss": 0.0856,
|
| 1113 |
+
"step": 1720
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 6.14,
|
| 1117 |
+
"learning_rate": 0.00021352313167259787,
|
| 1118 |
+
"loss": 0.1485,
|
| 1119 |
+
"step": 1730
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 6.18,
|
| 1123 |
+
"learning_rate": 0.00021154606563859234,
|
| 1124 |
+
"loss": 0.118,
|
| 1125 |
+
"step": 1740
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 6.21,
|
| 1129 |
+
"learning_rate": 0.0002095689996045868,
|
| 1130 |
+
"loss": 0.0684,
|
| 1131 |
+
"step": 1750
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 6.25,
|
| 1135 |
+
"learning_rate": 0.00020759193357058127,
|
| 1136 |
+
"loss": 0.1394,
|
| 1137 |
+
"step": 1760
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 6.28,
|
| 1141 |
+
"learning_rate": 0.00020561486753657574,
|
| 1142 |
+
"loss": 0.1556,
|
| 1143 |
+
"step": 1770
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 6.32,
|
| 1147 |
+
"learning_rate": 0.00020363780150257018,
|
| 1148 |
+
"loss": 0.1246,
|
| 1149 |
+
"step": 1780
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 6.35,
|
| 1153 |
+
"learning_rate": 0.00020166073546856465,
|
| 1154 |
+
"loss": 0.1257,
|
| 1155 |
+
"step": 1790
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 6.39,
|
| 1159 |
+
"learning_rate": 0.00019968366943455912,
|
| 1160 |
+
"loss": 0.17,
|
| 1161 |
+
"step": 1800
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 6.42,
|
| 1165 |
+
"learning_rate": 0.00019770660340055358,
|
| 1166 |
+
"loss": 0.0932,
|
| 1167 |
+
"step": 1810
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 6.46,
|
| 1171 |
+
"learning_rate": 0.00019572953736654805,
|
| 1172 |
+
"loss": 0.1288,
|
| 1173 |
+
"step": 1820
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 6.5,
|
| 1177 |
+
"learning_rate": 0.00019375247133254252,
|
| 1178 |
+
"loss": 0.0677,
|
| 1179 |
+
"step": 1830
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 6.53,
|
| 1183 |
+
"learning_rate": 0.000191775405298537,
|
| 1184 |
+
"loss": 0.1177,
|
| 1185 |
+
"step": 1840
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 6.57,
|
| 1189 |
+
"learning_rate": 0.00018979833926453143,
|
| 1190 |
+
"loss": 0.1093,
|
| 1191 |
+
"step": 1850
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 6.6,
|
| 1195 |
+
"learning_rate": 0.0001878212732305259,
|
| 1196 |
+
"loss": 0.1438,
|
| 1197 |
+
"step": 1860
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 6.64,
|
| 1201 |
+
"learning_rate": 0.00018584420719652036,
|
| 1202 |
+
"loss": 0.0854,
|
| 1203 |
+
"step": 1870
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 6.67,
|
| 1207 |
+
"learning_rate": 0.00018386714116251483,
|
| 1208 |
+
"loss": 0.0865,
|
| 1209 |
+
"step": 1880
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 6.71,
|
| 1213 |
+
"learning_rate": 0.0001818900751285093,
|
| 1214 |
+
"loss": 0.1254,
|
| 1215 |
+
"step": 1890
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 6.74,
|
| 1219 |
+
"learning_rate": 0.00017991300909450377,
|
| 1220 |
+
"loss": 0.0999,
|
| 1221 |
+
"step": 1900
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 6.78,
|
| 1225 |
+
"learning_rate": 0.00017793594306049823,
|
| 1226 |
+
"loss": 0.1072,
|
| 1227 |
+
"step": 1910
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 6.81,
|
| 1231 |
+
"learning_rate": 0.00017595887702649267,
|
| 1232 |
+
"loss": 0.1191,
|
| 1233 |
+
"step": 1920
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 6.85,
|
| 1237 |
+
"learning_rate": 0.00017398181099248717,
|
| 1238 |
+
"loss": 0.1147,
|
| 1239 |
+
"step": 1930
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 6.89,
|
| 1243 |
+
"learning_rate": 0.0001720047449584816,
|
| 1244 |
+
"loss": 0.1028,
|
| 1245 |
+
"step": 1940
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 6.92,
|
| 1249 |
+
"learning_rate": 0.00017002767892447608,
|
| 1250 |
+
"loss": 0.0695,
|
| 1251 |
+
"step": 1950
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 6.96,
|
| 1255 |
+
"learning_rate": 0.00016805061289047054,
|
| 1256 |
+
"loss": 0.0913,
|
| 1257 |
+
"step": 1960
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 6.99,
|
| 1261 |
+
"learning_rate": 0.000166073546856465,
|
| 1262 |
+
"loss": 0.0917,
|
| 1263 |
+
"step": 1970
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 7.0,
|
| 1267 |
+
"eval_accuracy": 0.9733570159857904,
|
| 1268 |
+
"eval_f1": 0.9715717407405214,
|
| 1269 |
+
"eval_loss": 0.10624586045742035,
|
| 1270 |
+
"eval_precision": 0.9764811708120374,
|
| 1271 |
+
"eval_recall": 0.9675866233506976,
|
| 1272 |
+
"eval_runtime": 39.8459,
|
| 1273 |
+
"eval_samples_per_second": 28.259,
|
| 1274 |
+
"eval_steps_per_second": 3.539,
|
| 1275 |
+
"step": 1972
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 7.03,
|
| 1279 |
+
"learning_rate": 0.00016409648082245948,
|
| 1280 |
+
"loss": 0.1021,
|
| 1281 |
+
"step": 1980
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 7.06,
|
| 1285 |
+
"learning_rate": 0.00016211941478845392,
|
| 1286 |
+
"loss": 0.0706,
|
| 1287 |
+
"step": 1990
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 7.1,
|
| 1291 |
+
"learning_rate": 0.00016014234875444842,
|
| 1292 |
+
"loss": 0.0677,
|
| 1293 |
+
"step": 2000
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 7.13,
|
| 1297 |
+
"learning_rate": 0.00015816528272044286,
|
| 1298 |
+
"loss": 0.0857,
|
| 1299 |
+
"step": 2010
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 7.17,
|
| 1303 |
+
"learning_rate": 0.00015618821668643732,
|
| 1304 |
+
"loss": 0.0865,
|
| 1305 |
+
"step": 2020
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 7.2,
|
| 1309 |
+
"learning_rate": 0.0001542111506524318,
|
| 1310 |
+
"loss": 0.0641,
|
| 1311 |
+
"step": 2030
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 7.24,
|
| 1315 |
+
"learning_rate": 0.00015223408461842626,
|
| 1316 |
+
"loss": 0.0446,
|
| 1317 |
+
"step": 2040
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 7.28,
|
| 1321 |
+
"learning_rate": 0.00015025701858442073,
|
| 1322 |
+
"loss": 0.053,
|
| 1323 |
+
"step": 2050
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 7.31,
|
| 1327 |
+
"learning_rate": 0.0001482799525504152,
|
| 1328 |
+
"loss": 0.087,
|
| 1329 |
+
"step": 2060
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 7.35,
|
| 1333 |
+
"learning_rate": 0.00014630288651640966,
|
| 1334 |
+
"loss": 0.0873,
|
| 1335 |
+
"step": 2070
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"epoch": 7.38,
|
| 1339 |
+
"learning_rate": 0.0001443258204824041,
|
| 1340 |
+
"loss": 0.099,
|
| 1341 |
+
"step": 2080
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 7.42,
|
| 1345 |
+
"learning_rate": 0.0001423487544483986,
|
| 1346 |
+
"loss": 0.1161,
|
| 1347 |
+
"step": 2090
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 7.45,
|
| 1351 |
+
"learning_rate": 0.00014037168841439304,
|
| 1352 |
+
"loss": 0.0951,
|
| 1353 |
+
"step": 2100
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 7.49,
|
| 1357 |
+
"learning_rate": 0.0001383946223803875,
|
| 1358 |
+
"loss": 0.0605,
|
| 1359 |
+
"step": 2110
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 7.52,
|
| 1363 |
+
"learning_rate": 0.00013641755634638197,
|
| 1364 |
+
"loss": 0.0857,
|
| 1365 |
+
"step": 2120
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 7.56,
|
| 1369 |
+
"learning_rate": 0.00013444049031237644,
|
| 1370 |
+
"loss": 0.1027,
|
| 1371 |
+
"step": 2130
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 7.6,
|
| 1375 |
+
"learning_rate": 0.0001324634242783709,
|
| 1376 |
+
"loss": 0.1222,
|
| 1377 |
+
"step": 2140
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 7.63,
|
| 1381 |
+
"learning_rate": 0.00013048635824436535,
|
| 1382 |
+
"loss": 0.1055,
|
| 1383 |
+
"step": 2150
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 7.67,
|
| 1387 |
+
"learning_rate": 0.00012850929221035984,
|
| 1388 |
+
"loss": 0.1326,
|
| 1389 |
+
"step": 2160
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 7.7,
|
| 1393 |
+
"learning_rate": 0.00012653222617635428,
|
| 1394 |
+
"loss": 0.0953,
|
| 1395 |
+
"step": 2170
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 7.74,
|
| 1399 |
+
"learning_rate": 0.00012455516014234875,
|
| 1400 |
+
"loss": 0.0672,
|
| 1401 |
+
"step": 2180
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 7.77,
|
| 1405 |
+
"learning_rate": 0.00012257809410834322,
|
| 1406 |
+
"loss": 0.0827,
|
| 1407 |
+
"step": 2190
|
| 1408 |
+
},
|
| 1409 |
+
{
|
| 1410 |
+
"epoch": 7.81,
|
| 1411 |
+
"learning_rate": 0.00012060102807433769,
|
| 1412 |
+
"loss": 0.0651,
|
| 1413 |
+
"step": 2200
|
| 1414 |
+
},
|
| 1415 |
+
{
|
| 1416 |
+
"epoch": 7.84,
|
| 1417 |
+
"learning_rate": 0.00011862396204033216,
|
| 1418 |
+
"loss": 0.0662,
|
| 1419 |
+
"step": 2210
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 7.88,
|
| 1423 |
+
"learning_rate": 0.00011664689600632661,
|
| 1424 |
+
"loss": 0.1431,
|
| 1425 |
+
"step": 2220
|
| 1426 |
+
},
|
| 1427 |
+
{
|
| 1428 |
+
"epoch": 7.91,
|
| 1429 |
+
"learning_rate": 0.00011466982997232108,
|
| 1430 |
+
"loss": 0.0315,
|
| 1431 |
+
"step": 2230
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 7.95,
|
| 1435 |
+
"learning_rate": 0.00011269276393831554,
|
| 1436 |
+
"loss": 0.0996,
|
| 1437 |
+
"step": 2240
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 7.99,
|
| 1441 |
+
"learning_rate": 0.00011071569790431001,
|
| 1442 |
+
"loss": 0.108,
|
| 1443 |
+
"step": 2250
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 8.0,
|
| 1447 |
+
"eval_accuracy": 0.9849023090586145,
|
| 1448 |
+
"eval_f1": 0.9828334388632164,
|
| 1449 |
+
"eval_loss": 0.05676717311143875,
|
| 1450 |
+
"eval_precision": 0.9868491250375566,
|
| 1451 |
+
"eval_recall": 0.9794172540958566,
|
| 1452 |
+
"eval_runtime": 39.7132,
|
| 1453 |
+
"eval_samples_per_second": 28.353,
|
| 1454 |
+
"eval_steps_per_second": 3.55,
|
| 1455 |
+
"step": 2254
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 8.02,
|
| 1459 |
+
"learning_rate": 0.00010873863187030447,
|
| 1460 |
+
"loss": 0.0307,
|
| 1461 |
+
"step": 2260
|
| 1462 |
+
},
|
| 1463 |
+
{
|
| 1464 |
+
"epoch": 8.06,
|
| 1465 |
+
"learning_rate": 0.00010676156583629893,
|
| 1466 |
+
"loss": 0.0847,
|
| 1467 |
+
"step": 2270
|
| 1468 |
+
},
|
| 1469 |
+
{
|
| 1470 |
+
"epoch": 8.09,
|
| 1471 |
+
"learning_rate": 0.0001047844998022934,
|
| 1472 |
+
"loss": 0.1177,
|
| 1473 |
+
"step": 2280
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 8.13,
|
| 1477 |
+
"learning_rate": 0.00010280743376828787,
|
| 1478 |
+
"loss": 0.0689,
|
| 1479 |
+
"step": 2290
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 8.16,
|
| 1483 |
+
"learning_rate": 0.00010083036773428232,
|
| 1484 |
+
"loss": 0.0873,
|
| 1485 |
+
"step": 2300
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 8.2,
|
| 1489 |
+
"learning_rate": 9.885330170027679e-05,
|
| 1490 |
+
"loss": 0.1209,
|
| 1491 |
+
"step": 2310
|
| 1492 |
+
},
|
| 1493 |
+
{
|
| 1494 |
+
"epoch": 8.23,
|
| 1495 |
+
"learning_rate": 9.687623566627126e-05,
|
| 1496 |
+
"loss": 0.0488,
|
| 1497 |
+
"step": 2320
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 8.27,
|
| 1501 |
+
"learning_rate": 9.489916963226571e-05,
|
| 1502 |
+
"loss": 0.0558,
|
| 1503 |
+
"step": 2330
|
| 1504 |
+
},
|
| 1505 |
+
{
|
| 1506 |
+
"epoch": 8.31,
|
| 1507 |
+
"learning_rate": 9.292210359826018e-05,
|
| 1508 |
+
"loss": 0.067,
|
| 1509 |
+
"step": 2340
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 8.34,
|
| 1513 |
+
"learning_rate": 9.094503756425465e-05,
|
| 1514 |
+
"loss": 0.13,
|
| 1515 |
+
"step": 2350
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 8.38,
|
| 1519 |
+
"learning_rate": 8.896797153024912e-05,
|
| 1520 |
+
"loss": 0.0743,
|
| 1521 |
+
"step": 2360
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 8.41,
|
| 1525 |
+
"learning_rate": 8.699090549624358e-05,
|
| 1526 |
+
"loss": 0.0797,
|
| 1527 |
+
"step": 2370
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 8.45,
|
| 1531 |
+
"learning_rate": 8.501383946223804e-05,
|
| 1532 |
+
"loss": 0.0913,
|
| 1533 |
+
"step": 2380
|
| 1534 |
+
},
|
| 1535 |
+
{
|
| 1536 |
+
"epoch": 8.48,
|
| 1537 |
+
"learning_rate": 8.30367734282325e-05,
|
| 1538 |
+
"loss": 0.0772,
|
| 1539 |
+
"step": 2390
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 8.52,
|
| 1543 |
+
"learning_rate": 8.105970739422696e-05,
|
| 1544 |
+
"loss": 0.072,
|
| 1545 |
+
"step": 2400
|
| 1546 |
+
},
|
| 1547 |
+
{
|
| 1548 |
+
"epoch": 8.55,
|
| 1549 |
+
"learning_rate": 7.908264136022143e-05,
|
| 1550 |
+
"loss": 0.0566,
|
| 1551 |
+
"step": 2410
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 8.59,
|
| 1555 |
+
"learning_rate": 7.71055753262159e-05,
|
| 1556 |
+
"loss": 0.1198,
|
| 1557 |
+
"step": 2420
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 8.62,
|
| 1561 |
+
"learning_rate": 7.512850929221036e-05,
|
| 1562 |
+
"loss": 0.0647,
|
| 1563 |
+
"step": 2430
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 8.66,
|
| 1567 |
+
"learning_rate": 7.315144325820483e-05,
|
| 1568 |
+
"loss": 0.1079,
|
| 1569 |
+
"step": 2440
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 8.7,
|
| 1573 |
+
"learning_rate": 7.11743772241993e-05,
|
| 1574 |
+
"loss": 0.0792,
|
| 1575 |
+
"step": 2450
|
| 1576 |
+
},
|
| 1577 |
+
{
|
| 1578 |
+
"epoch": 8.73,
|
| 1579 |
+
"learning_rate": 6.919731119019375e-05,
|
| 1580 |
+
"loss": 0.0712,
|
| 1581 |
+
"step": 2460
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 8.77,
|
| 1585 |
+
"learning_rate": 6.722024515618822e-05,
|
| 1586 |
+
"loss": 0.0707,
|
| 1587 |
+
"step": 2470
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"epoch": 8.8,
|
| 1591 |
+
"learning_rate": 6.524317912218267e-05,
|
| 1592 |
+
"loss": 0.0571,
|
| 1593 |
+
"step": 2480
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 8.84,
|
| 1597 |
+
"learning_rate": 6.326611308817714e-05,
|
| 1598 |
+
"loss": 0.0802,
|
| 1599 |
+
"step": 2490
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 8.87,
|
| 1603 |
+
"learning_rate": 6.128904705417161e-05,
|
| 1604 |
+
"loss": 0.0533,
|
| 1605 |
+
"step": 2500
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 8.91,
|
| 1609 |
+
"learning_rate": 5.931198102016608e-05,
|
| 1610 |
+
"loss": 0.078,
|
| 1611 |
+
"step": 2510
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 8.94,
|
| 1615 |
+
"learning_rate": 5.733491498616054e-05,
|
| 1616 |
+
"loss": 0.0795,
|
| 1617 |
+
"step": 2520
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 8.98,
|
| 1621 |
+
"learning_rate": 5.5357848952155006e-05,
|
| 1622 |
+
"loss": 0.1151,
|
| 1623 |
+
"step": 2530
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 9.0,
|
| 1627 |
+
"eval_accuracy": 0.9857904085257548,
|
| 1628 |
+
"eval_f1": 0.9868558448721283,
|
| 1629 |
+
"eval_loss": 0.050765056163072586,
|
| 1630 |
+
"eval_precision": 0.987576986984624,
|
| 1631 |
+
"eval_recall": 0.9863465915836483,
|
| 1632 |
+
"eval_runtime": 39.9806,
|
| 1633 |
+
"eval_samples_per_second": 28.164,
|
| 1634 |
+
"eval_steps_per_second": 3.527,
|
| 1635 |
+
"step": 2535
|
| 1636 |
+
},
|
| 1637 |
+
{
|
| 1638 |
+
"epoch": 9.02,
|
| 1639 |
+
"learning_rate": 5.338078291814947e-05,
|
| 1640 |
+
"loss": 0.0986,
|
| 1641 |
+
"step": 2540
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 9.05,
|
| 1645 |
+
"learning_rate": 5.1403716884143935e-05,
|
| 1646 |
+
"loss": 0.0261,
|
| 1647 |
+
"step": 2550
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 9.09,
|
| 1651 |
+
"learning_rate": 4.9426650850138396e-05,
|
| 1652 |
+
"loss": 0.0691,
|
| 1653 |
+
"step": 2560
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 9.12,
|
| 1657 |
+
"learning_rate": 4.744958481613286e-05,
|
| 1658 |
+
"loss": 0.0361,
|
| 1659 |
+
"step": 2570
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 9.16,
|
| 1663 |
+
"learning_rate": 4.5472518782127324e-05,
|
| 1664 |
+
"loss": 0.0829,
|
| 1665 |
+
"step": 2580
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 9.19,
|
| 1669 |
+
"learning_rate": 4.349545274812179e-05,
|
| 1670 |
+
"loss": 0.0701,
|
| 1671 |
+
"step": 2590
|
| 1672 |
+
},
|
| 1673 |
+
{
|
| 1674 |
+
"epoch": 9.23,
|
| 1675 |
+
"learning_rate": 4.151838671411625e-05,
|
| 1676 |
+
"loss": 0.0608,
|
| 1677 |
+
"step": 2600
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 9.26,
|
| 1681 |
+
"learning_rate": 3.9541320680110714e-05,
|
| 1682 |
+
"loss": 0.056,
|
| 1683 |
+
"step": 2610
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 9.3,
|
| 1687 |
+
"learning_rate": 3.756425464610518e-05,
|
| 1688 |
+
"loss": 0.0595,
|
| 1689 |
+
"step": 2620
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 9.33,
|
| 1693 |
+
"learning_rate": 3.558718861209965e-05,
|
| 1694 |
+
"loss": 0.0891,
|
| 1695 |
+
"step": 2630
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 9.37,
|
| 1699 |
+
"learning_rate": 3.361012257809411e-05,
|
| 1700 |
+
"loss": 0.0716,
|
| 1701 |
+
"step": 2640
|
| 1702 |
+
},
|
| 1703 |
+
{
|
| 1704 |
+
"epoch": 9.41,
|
| 1705 |
+
"learning_rate": 3.163305654408857e-05,
|
| 1706 |
+
"loss": 0.0674,
|
| 1707 |
+
"step": 2650
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 9.44,
|
| 1711 |
+
"learning_rate": 2.965599051008304e-05,
|
| 1712 |
+
"loss": 0.0568,
|
| 1713 |
+
"step": 2660
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 9.48,
|
| 1717 |
+
"learning_rate": 2.7678924476077503e-05,
|
| 1718 |
+
"loss": 0.0531,
|
| 1719 |
+
"step": 2670
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 9.51,
|
| 1723 |
+
"learning_rate": 2.5701858442071967e-05,
|
| 1724 |
+
"loss": 0.0749,
|
| 1725 |
+
"step": 2680
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 9.55,
|
| 1729 |
+
"learning_rate": 2.372479240806643e-05,
|
| 1730 |
+
"loss": 0.09,
|
| 1731 |
+
"step": 2690
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 9.58,
|
| 1735 |
+
"learning_rate": 2.1747726374060896e-05,
|
| 1736 |
+
"loss": 0.0794,
|
| 1737 |
+
"step": 2700
|
| 1738 |
+
},
|
| 1739 |
+
{
|
| 1740 |
+
"epoch": 9.62,
|
| 1741 |
+
"learning_rate": 1.9770660340055357e-05,
|
| 1742 |
+
"loss": 0.0452,
|
| 1743 |
+
"step": 2710
|
| 1744 |
+
},
|
| 1745 |
+
{
|
| 1746 |
+
"epoch": 9.65,
|
| 1747 |
+
"learning_rate": 1.7793594306049825e-05,
|
| 1748 |
+
"loss": 0.0822,
|
| 1749 |
+
"step": 2720
|
| 1750 |
+
},
|
| 1751 |
+
{
|
| 1752 |
+
"epoch": 9.69,
|
| 1753 |
+
"learning_rate": 1.5816528272044286e-05,
|
| 1754 |
+
"loss": 0.0461,
|
| 1755 |
+
"step": 2730
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 9.72,
|
| 1759 |
+
"learning_rate": 1.3839462238038752e-05,
|
| 1760 |
+
"loss": 0.043,
|
| 1761 |
+
"step": 2740
|
| 1762 |
+
},
|
| 1763 |
+
{
|
| 1764 |
+
"epoch": 9.76,
|
| 1765 |
+
"learning_rate": 1.1862396204033214e-05,
|
| 1766 |
+
"loss": 0.1089,
|
| 1767 |
+
"step": 2750
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 9.8,
|
| 1771 |
+
"learning_rate": 9.885330170027678e-06,
|
| 1772 |
+
"loss": 0.0963,
|
| 1773 |
+
"step": 2760
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 9.83,
|
| 1777 |
+
"learning_rate": 7.908264136022143e-06,
|
| 1778 |
+
"loss": 0.0683,
|
| 1779 |
+
"step": 2770
|
| 1780 |
+
},
|
| 1781 |
+
{
|
| 1782 |
+
"epoch": 9.87,
|
| 1783 |
+
"learning_rate": 5.931198102016607e-06,
|
| 1784 |
+
"loss": 0.047,
|
| 1785 |
+
"step": 2780
|
| 1786 |
+
},
|
| 1787 |
+
{
|
| 1788 |
+
"epoch": 9.9,
|
| 1789 |
+
"learning_rate": 3.954132068011071e-06,
|
| 1790 |
+
"loss": 0.0792,
|
| 1791 |
+
"step": 2790
|
| 1792 |
+
},
|
| 1793 |
+
{
|
| 1794 |
+
"epoch": 9.94,
|
| 1795 |
+
"learning_rate": 1.9770660340055357e-06,
|
| 1796 |
+
"loss": 0.0441,
|
| 1797 |
+
"step": 2800
|
| 1798 |
+
},
|
| 1799 |
+
{
|
| 1800 |
+
"epoch": 9.97,
|
| 1801 |
+
"learning_rate": 0.0,
|
| 1802 |
+
"loss": 0.049,
|
| 1803 |
+
"step": 2810
|
| 1804 |
+
},
|
| 1805 |
+
{
|
| 1806 |
+
"epoch": 9.97,
|
| 1807 |
+
"eval_accuracy": 0.9857904085257548,
|
| 1808 |
+
"eval_f1": 0.9877505171364073,
|
| 1809 |
+
"eval_loss": 0.050837442278862,
|
| 1810 |
+
"eval_precision": 0.9891672844560683,
|
| 1811 |
+
"eval_recall": 0.9865008922572412,
|
| 1812 |
+
"eval_runtime": 39.8176,
|
| 1813 |
+
"eval_samples_per_second": 28.279,
|
| 1814 |
+
"eval_steps_per_second": 3.541,
|
| 1815 |
+
"step": 2810
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 9.97,
|
| 1819 |
+
"step": 2810,
|
| 1820 |
+
"total_flos": 2.0918954997191803e+19,
|
| 1821 |
+
"train_loss": 0.16025869953462663,
|
| 1822 |
+
"train_runtime": 8523.7997,
|
| 1823 |
+
"train_samples_per_second": 10.57,
|
| 1824 |
+
"train_steps_per_second": 0.33
|
| 1825 |
+
}
|
| 1826 |
+
],
|
| 1827 |
+
"logging_steps": 10,
|
| 1828 |
+
"max_steps": 2810,
|
| 1829 |
+
"num_train_epochs": 10,
|
| 1830 |
+
"save_steps": 500,
|
| 1831 |
+
"total_flos": 2.0918954997191803e+19,
|
| 1832 |
+
"trial_name": null,
|
| 1833 |
+
"trial_params": null
|
| 1834 |
+
}
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4091
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0e7ebea4550f56e28f137084d224481b5a37d5ed28f8c1d765b5589a7c364f76
|
| 3 |
size 4091
|