End of training
Browse files- README.md +5 -5
- all_results.json +16 -0
- eval_results.json +11 -0
- runs/Mar16_11-27-33_774ccf98b3fe/events.out.tfevents.1742127600.774ccf98b3fe.5066.2 +3 -0
- train_results.json +8 -0
- trainer_state.json +1030 -0
README.md
CHANGED
|
@@ -21,11 +21,11 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 21 |
|
| 22 |
This model is a fine-tuned version of [microsoft/beit-base-patch16-224-pt22k-ft22k](https://huggingface.co/microsoft/beit-base-patch16-224-pt22k-ft22k) on an unknown dataset.
|
| 23 |
It achieves the following results on the evaluation set:
|
| 24 |
-
- Loss: 0.
|
| 25 |
-
- Accuracy: 0.
|
| 26 |
-
- F1: 0.
|
| 27 |
-
- Precision: 0.
|
| 28 |
-
- Recall: 0.
|
| 29 |
|
| 30 |
## Model description
|
| 31 |
|
|
|
|
| 21 |
|
| 22 |
This model is a fine-tuned version of [microsoft/beit-base-patch16-224-pt22k-ft22k](https://huggingface.co/microsoft/beit-base-patch16-224-pt22k-ft22k) on an unknown dataset.
|
| 23 |
It achieves the following results on the evaluation set:
|
| 24 |
+
- Loss: 0.2029
|
| 25 |
+
- Accuracy: 0.9222
|
| 26 |
+
- F1: 0.9214
|
| 27 |
+
- Precision: 0.9234
|
| 28 |
+
- Recall: 0.9222
|
| 29 |
|
| 30 |
## Model description
|
| 31 |
|
all_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 47.9907120743034,
|
| 3 |
+
"eval_accuracy": 0.9222071460877431,
|
| 4 |
+
"eval_f1": 0.9213530294983667,
|
| 5 |
+
"eval_loss": 0.20286186039447784,
|
| 6 |
+
"eval_precision": 0.9233953907433776,
|
| 7 |
+
"eval_recall": 0.9222071460877431,
|
| 8 |
+
"eval_runtime": 9.8001,
|
| 9 |
+
"eval_samples_per_second": 225.611,
|
| 10 |
+
"eval_steps_per_second": 28.265,
|
| 11 |
+
"total_flos": 2.738563601264935e+19,
|
| 12 |
+
"train_loss": 0.3012936460475127,
|
| 13 |
+
"train_runtime": 3086.985,
|
| 14 |
+
"train_samples_per_second": 80.203,
|
| 15 |
+
"train_steps_per_second": 1.244
|
| 16 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 47.9907120743034,
|
| 3 |
+
"eval_accuracy": 0.9222071460877431,
|
| 4 |
+
"eval_f1": 0.9213530294983667,
|
| 5 |
+
"eval_loss": 0.20286186039447784,
|
| 6 |
+
"eval_precision": 0.9233953907433776,
|
| 7 |
+
"eval_recall": 0.9222071460877431,
|
| 8 |
+
"eval_runtime": 9.8001,
|
| 9 |
+
"eval_samples_per_second": 225.611,
|
| 10 |
+
"eval_steps_per_second": 28.265
|
| 11 |
+
}
|
runs/Mar16_11-27-33_774ccf98b3fe/events.out.tfevents.1742127600.774ccf98b3fe.5066.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dc918b6f8d2e9be6798aab0da6703d7db791ebfb85d742fa2e2569de3759b156
|
| 3 |
+
size 560
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 47.9907120743034,
|
| 3 |
+
"total_flos": 2.738563601264935e+19,
|
| 4 |
+
"train_loss": 0.3012936460475127,
|
| 5 |
+
"train_runtime": 3086.985,
|
| 6 |
+
"train_samples_per_second": 80.203,
|
| 7 |
+
"train_steps_per_second": 1.244
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1030 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9213530294983667,
|
| 3 |
+
"best_model_checkpoint": "beit-base-patch16-224-pt22k-ft22k-finetuned-stroke-binary/checkpoint-3200",
|
| 4 |
+
"epoch": 47.9907120743034,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 3840,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.6191950464396285,
|
| 13 |
+
"grad_norm": 5.044180870056152,
|
| 14 |
+
"learning_rate": 2.604166666666667e-06,
|
| 15 |
+
"loss": 0.744,
|
| 16 |
+
"step": 50
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 1.2476780185758514,
|
| 20 |
+
"grad_norm": 1.894531011581421,
|
| 21 |
+
"learning_rate": 5.208333333333334e-06,
|
| 22 |
+
"loss": 0.7256,
|
| 23 |
+
"step": 100
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 1.2476780185758514,
|
| 27 |
+
"eval_accuracy": 0.5685210312075983,
|
| 28 |
+
"eval_f1": 0.48230573725672854,
|
| 29 |
+
"eval_loss": 0.6912839412689209,
|
| 30 |
+
"eval_precision": 0.47305459682975004,
|
| 31 |
+
"eval_recall": 0.5685210312075983,
|
| 32 |
+
"eval_runtime": 9.7307,
|
| 33 |
+
"eval_samples_per_second": 227.22,
|
| 34 |
+
"eval_steps_per_second": 28.467,
|
| 35 |
+
"step": 100
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"epoch": 1.86687306501548,
|
| 39 |
+
"grad_norm": 1.9874528646469116,
|
| 40 |
+
"learning_rate": 7.8125e-06,
|
| 41 |
+
"loss": 0.7033,
|
| 42 |
+
"step": 150
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 2.4953560371517027,
|
| 46 |
+
"grad_norm": 2.1426079273223877,
|
| 47 |
+
"learning_rate": 1.0416666666666668e-05,
|
| 48 |
+
"loss": 0.6695,
|
| 49 |
+
"step": 200
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"epoch": 2.4953560371517027,
|
| 53 |
+
"eval_accuracy": 0.6209859791949344,
|
| 54 |
+
"eval_f1": 0.5163511984739979,
|
| 55 |
+
"eval_loss": 0.64801025390625,
|
| 56 |
+
"eval_precision": 0.5986617059765201,
|
| 57 |
+
"eval_recall": 0.6209859791949344,
|
| 58 |
+
"eval_runtime": 9.7733,
|
| 59 |
+
"eval_samples_per_second": 226.229,
|
| 60 |
+
"eval_steps_per_second": 28.343,
|
| 61 |
+
"step": 200
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 3.123839009287926,
|
| 65 |
+
"grad_norm": 1.4950852394104004,
|
| 66 |
+
"learning_rate": 1.3020833333333334e-05,
|
| 67 |
+
"loss": 0.6399,
|
| 68 |
+
"step": 250
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 3.7430340557275543,
|
| 72 |
+
"grad_norm": 1.932099461555481,
|
| 73 |
+
"learning_rate": 1.5625e-05,
|
| 74 |
+
"loss": 0.5963,
|
| 75 |
+
"step": 300
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 3.7430340557275543,
|
| 79 |
+
"eval_accuracy": 0.6725463591135233,
|
| 80 |
+
"eval_f1": 0.6118497378760206,
|
| 81 |
+
"eval_loss": 0.5882277488708496,
|
| 82 |
+
"eval_precision": 0.6992514519414246,
|
| 83 |
+
"eval_recall": 0.6725463591135233,
|
| 84 |
+
"eval_runtime": 9.9265,
|
| 85 |
+
"eval_samples_per_second": 222.737,
|
| 86 |
+
"eval_steps_per_second": 27.905,
|
| 87 |
+
"step": 300
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 4.371517027863777,
|
| 91 |
+
"grad_norm": 1.4714239835739136,
|
| 92 |
+
"learning_rate": 1.8229166666666668e-05,
|
| 93 |
+
"loss": 0.5616,
|
| 94 |
+
"step": 350
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 4.9907120743034055,
|
| 98 |
+
"grad_norm": 1.6321817636489868,
|
| 99 |
+
"learning_rate": 1.9998942319271076e-05,
|
| 100 |
+
"loss": 0.518,
|
| 101 |
+
"step": 400
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 4.9907120743034055,
|
| 105 |
+
"eval_accuracy": 0.7480777928539123,
|
| 106 |
+
"eval_f1": 0.7167175429836762,
|
| 107 |
+
"eval_loss": 0.49900302290916443,
|
| 108 |
+
"eval_precision": 0.7891493639439078,
|
| 109 |
+
"eval_recall": 0.7480777928539123,
|
| 110 |
+
"eval_runtime": 9.9652,
|
| 111 |
+
"eval_samples_per_second": 221.873,
|
| 112 |
+
"eval_steps_per_second": 27.797,
|
| 113 |
+
"step": 400
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 5.6191950464396285,
|
| 117 |
+
"grad_norm": 2.7871556282043457,
|
| 118 |
+
"learning_rate": 1.998200798188685e-05,
|
| 119 |
+
"loss": 0.4858,
|
| 120 |
+
"step": 450
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"epoch": 6.247678018575852,
|
| 124 |
+
"grad_norm": 3.0406832695007324,
|
| 125 |
+
"learning_rate": 1.9944456171551024e-05,
|
| 126 |
+
"loss": 0.4325,
|
| 127 |
+
"step": 500
|
| 128 |
+
},
|
| 129 |
+
{
|
| 130 |
+
"epoch": 6.247678018575852,
|
| 131 |
+
"eval_accuracy": 0.8073270013568521,
|
| 132 |
+
"eval_f1": 0.7957289617278764,
|
| 133 |
+
"eval_loss": 0.40903061628341675,
|
| 134 |
+
"eval_precision": 0.8231915996306657,
|
| 135 |
+
"eval_recall": 0.8073270013568521,
|
| 136 |
+
"eval_runtime": 10.0105,
|
| 137 |
+
"eval_samples_per_second": 220.868,
|
| 138 |
+
"eval_steps_per_second": 27.671,
|
| 139 |
+
"step": 500
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 6.86687306501548,
|
| 143 |
+
"grad_norm": 1.8095026016235352,
|
| 144 |
+
"learning_rate": 1.9886364450156485e-05,
|
| 145 |
+
"loss": 0.3996,
|
| 146 |
+
"step": 550
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 7.495356037151703,
|
| 150 |
+
"grad_norm": 2.2104787826538086,
|
| 151 |
+
"learning_rate": 1.9807852804032306e-05,
|
| 152 |
+
"loss": 0.3848,
|
| 153 |
+
"step": 600
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 7.495356037151703,
|
| 157 |
+
"eval_accuracy": 0.8340117593848937,
|
| 158 |
+
"eval_f1": 0.8257264682013935,
|
| 159 |
+
"eval_loss": 0.37033286690711975,
|
| 160 |
+
"eval_precision": 0.8481671958516468,
|
| 161 |
+
"eval_recall": 0.8340117593848937,
|
| 162 |
+
"eval_runtime": 10.0319,
|
| 163 |
+
"eval_samples_per_second": 220.396,
|
| 164 |
+
"eval_steps_per_second": 27.612,
|
| 165 |
+
"step": 600
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 8.123839009287925,
|
| 169 |
+
"grad_norm": 1.679534912109375,
|
| 170 |
+
"learning_rate": 1.970908339611638e-05,
|
| 171 |
+
"loss": 0.3628,
|
| 172 |
+
"step": 650
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 8.743034055727554,
|
| 176 |
+
"grad_norm": 1.1306818723678589,
|
| 177 |
+
"learning_rate": 1.9590260231013774e-05,
|
| 178 |
+
"loss": 0.3532,
|
| 179 |
+
"step": 700
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 8.743034055727554,
|
| 183 |
+
"eval_accuracy": 0.8312980551786522,
|
| 184 |
+
"eval_f1": 0.8200843868501977,
|
| 185 |
+
"eval_loss": 0.39578977227211,
|
| 186 |
+
"eval_precision": 0.8564151669725956,
|
| 187 |
+
"eval_recall": 0.8312980551786522,
|
| 188 |
+
"eval_runtime": 10.0177,
|
| 189 |
+
"eval_samples_per_second": 220.71,
|
| 190 |
+
"eval_steps_per_second": 27.651,
|
| 191 |
+
"step": 700
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 9.371517027863778,
|
| 195 |
+
"grad_norm": 2.967256784439087,
|
| 196 |
+
"learning_rate": 1.945162873363268e-05,
|
| 197 |
+
"loss": 0.3368,
|
| 198 |
+
"step": 750
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 9.990712074303406,
|
| 202 |
+
"grad_norm": 1.7120331525802612,
|
| 203 |
+
"learning_rate": 1.9293475242268224e-05,
|
| 204 |
+
"loss": 0.3297,
|
| 205 |
+
"step": 800
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 9.990712074303406,
|
| 209 |
+
"eval_accuracy": 0.8611488014473089,
|
| 210 |
+
"eval_f1": 0.8558119987362278,
|
| 211 |
+
"eval_loss": 0.32570937275886536,
|
| 212 |
+
"eval_precision": 0.871840635108417,
|
| 213 |
+
"eval_recall": 0.8611488014473089,
|
| 214 |
+
"eval_runtime": 10.0508,
|
| 215 |
+
"eval_samples_per_second": 219.982,
|
| 216 |
+
"eval_steps_per_second": 27.56,
|
| 217 |
+
"step": 800
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"epoch": 10.619195046439629,
|
| 221 |
+
"grad_norm": 1.371287226676941,
|
| 222 |
+
"learning_rate": 1.9116126417181188e-05,
|
| 223 |
+
"loss": 0.3143,
|
| 224 |
+
"step": 850
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 11.24767801857585,
|
| 228 |
+
"grad_norm": 1.5851930379867554,
|
| 229 |
+
"learning_rate": 1.8919948565893144e-05,
|
| 230 |
+
"loss": 0.3281,
|
| 231 |
+
"step": 900
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 11.24767801857585,
|
| 235 |
+
"eval_accuracy": 0.866576209859792,
|
| 236 |
+
"eval_f1": 0.8611949189284244,
|
| 237 |
+
"eval_loss": 0.3168693780899048,
|
| 238 |
+
"eval_precision": 0.8790588339902896,
|
| 239 |
+
"eval_recall": 0.866576209859792,
|
| 240 |
+
"eval_runtime": 10.0434,
|
| 241 |
+
"eval_samples_per_second": 220.144,
|
| 242 |
+
"eval_steps_per_second": 27.58,
|
| 243 |
+
"step": 900
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 11.86687306501548,
|
| 247 |
+
"grad_norm": 0.8977594375610352,
|
| 248 |
+
"learning_rate": 1.8705346886591667e-05,
|
| 249 |
+
"loss": 0.3036,
|
| 250 |
+
"step": 950
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 12.495356037151703,
|
| 254 |
+
"grad_norm": 2.8011789321899414,
|
| 255 |
+
"learning_rate": 1.847276463120828e-05,
|
| 256 |
+
"loss": 0.2938,
|
| 257 |
+
"step": 1000
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 12.495356037151703,
|
| 261 |
+
"eval_accuracy": 0.8864767073722297,
|
| 262 |
+
"eval_f1": 0.8841053273932936,
|
| 263 |
+
"eval_loss": 0.28136056661605835,
|
| 264 |
+
"eval_precision": 0.8900209862709351,
|
| 265 |
+
"eval_recall": 0.8864767073722297,
|
| 266 |
+
"eval_runtime": 10.0799,
|
| 267 |
+
"eval_samples_per_second": 219.348,
|
| 268 |
+
"eval_steps_per_second": 27.481,
|
| 269 |
+
"step": 1000
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 13.123839009287925,
|
| 273 |
+
"grad_norm": 2.865165948867798,
|
| 274 |
+
"learning_rate": 1.822268218989775e-05,
|
| 275 |
+
"loss": 0.294,
|
| 276 |
+
"step": 1050
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 13.743034055727554,
|
| 280 |
+
"grad_norm": 1.8461377620697021,
|
| 281 |
+
"learning_rate": 1.79556160988098e-05,
|
| 282 |
+
"loss": 0.2866,
|
| 283 |
+
"step": 1100
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 13.743034055727554,
|
| 287 |
+
"eval_accuracy": 0.8869289914066033,
|
| 288 |
+
"eval_f1": 0.883705253652531,
|
| 289 |
+
"eval_loss": 0.2827624976634979,
|
| 290 |
+
"eval_precision": 0.8942761361900506,
|
| 291 |
+
"eval_recall": 0.8869289914066033,
|
| 292 |
+
"eval_runtime": 10.0748,
|
| 293 |
+
"eval_samples_per_second": 219.459,
|
| 294 |
+
"eval_steps_per_second": 27.494,
|
| 295 |
+
"step": 1100
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 14.371517027863778,
|
| 299 |
+
"grad_norm": 1.1585817337036133,
|
| 300 |
+
"learning_rate": 1.7672117973202526e-05,
|
| 301 |
+
"loss": 0.2667,
|
| 302 |
+
"step": 1150
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 14.990712074303406,
|
| 306 |
+
"grad_norm": 1.050703525543213,
|
| 307 |
+
"learning_rate": 1.737277336810124e-05,
|
| 308 |
+
"loss": 0.2884,
|
| 309 |
+
"step": 1200
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 14.990712074303406,
|
| 313 |
+
"eval_accuracy": 0.8846675712347354,
|
| 314 |
+
"eval_f1": 0.8810137319907275,
|
| 315 |
+
"eval_loss": 0.29289892315864563,
|
| 316 |
+
"eval_precision": 0.8936495972210173,
|
| 317 |
+
"eval_recall": 0.8846675712347354,
|
| 318 |
+
"eval_runtime": 10.0424,
|
| 319 |
+
"eval_samples_per_second": 220.167,
|
| 320 |
+
"eval_steps_per_second": 27.583,
|
| 321 |
+
"step": 1200
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 15.619195046439629,
|
| 325 |
+
"grad_norm": 1.1661638021469116,
|
| 326 |
+
"learning_rate": 1.705820056885596e-05,
|
| 327 |
+
"loss": 0.2722,
|
| 328 |
+
"step": 1250
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 16.24767801857585,
|
| 332 |
+
"grad_norm": 1.4716720581054688,
|
| 333 |
+
"learning_rate": 1.6729049314095578e-05,
|
| 334 |
+
"loss": 0.2808,
|
| 335 |
+
"step": 1300
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 16.24767801857585,
|
| 339 |
+
"eval_accuracy": 0.9014020805065581,
|
| 340 |
+
"eval_f1": 0.8998791257539092,
|
| 341 |
+
"eval_loss": 0.24578717350959778,
|
| 342 |
+
"eval_precision": 0.9034472507945439,
|
| 343 |
+
"eval_recall": 0.9014020805065581,
|
| 344 |
+
"eval_runtime": 10.1493,
|
| 345 |
+
"eval_samples_per_second": 217.847,
|
| 346 |
+
"eval_steps_per_second": 27.292,
|
| 347 |
+
"step": 1300
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 16.86687306501548,
|
| 351 |
+
"grad_norm": 1.2002067565917969,
|
| 352 |
+
"learning_rate": 1.6385999453716453e-05,
|
| 353 |
+
"loss": 0.2843,
|
| 354 |
+
"step": 1350
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 17.4953560371517,
|
| 358 |
+
"grad_norm": 2.676912307739258,
|
| 359 |
+
"learning_rate": 1.6029759544677298e-05,
|
| 360 |
+
"loss": 0.258,
|
| 361 |
+
"step": 1400
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 17.4953560371517,
|
| 365 |
+
"eval_accuracy": 0.9090909090909091,
|
| 366 |
+
"eval_f1": 0.908023053812085,
|
| 367 |
+
"eval_loss": 0.235076442360878,
|
| 368 |
+
"eval_precision": 0.9101607687708339,
|
| 369 |
+
"eval_recall": 0.9090909090909091,
|
| 370 |
+
"eval_runtime": 10.0503,
|
| 371 |
+
"eval_samples_per_second": 219.993,
|
| 372 |
+
"eval_steps_per_second": 27.561,
|
| 373 |
+
"step": 1400
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 18.123839009287927,
|
| 377 |
+
"grad_norm": 1.2184885740280151,
|
| 378 |
+
"learning_rate": 1.566106538750063e-05,
|
| 379 |
+
"loss": 0.2639,
|
| 380 |
+
"step": 1450
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 18.743034055727556,
|
| 384 |
+
"grad_norm": 1.8000015020370483,
|
| 385 |
+
"learning_rate": 1.528067850650368e-05,
|
| 386 |
+
"loss": 0.2744,
|
| 387 |
+
"step": 1500
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 18.743034055727556,
|
| 391 |
+
"eval_accuracy": 0.9014020805065581,
|
| 392 |
+
"eval_f1": 0.8993606111929588,
|
| 393 |
+
"eval_loss": 0.25163090229034424,
|
| 394 |
+
"eval_precision": 0.9056565314020832,
|
| 395 |
+
"eval_recall": 0.9014020805065581,
|
| 396 |
+
"eval_runtime": 10.0711,
|
| 397 |
+
"eval_samples_per_second": 219.538,
|
| 398 |
+
"eval_steps_per_second": 27.504,
|
| 399 |
+
"step": 1500
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 19.371517027863778,
|
| 403 |
+
"grad_norm": 0.9816193580627441,
|
| 404 |
+
"learning_rate": 1.4889384576897728e-05,
|
| 405 |
+
"loss": 0.2576,
|
| 406 |
+
"step": 1550
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 19.990712074303406,
|
| 410 |
+
"grad_norm": 1.5961647033691406,
|
| 411 |
+
"learning_rate": 1.4487991802004625e-05,
|
| 412 |
+
"loss": 0.261,
|
| 413 |
+
"step": 1600
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 19.990712074303406,
|
| 417 |
+
"eval_accuracy": 0.9068294889190411,
|
| 418 |
+
"eval_f1": 0.9050356054740915,
|
| 419 |
+
"eval_loss": 0.2453160583972931,
|
| 420 |
+
"eval_precision": 0.910682778753074,
|
| 421 |
+
"eval_recall": 0.9068294889190411,
|
| 422 |
+
"eval_runtime": 10.0772,
|
| 423 |
+
"eval_samples_per_second": 219.406,
|
| 424 |
+
"eval_steps_per_second": 27.488,
|
| 425 |
+
"step": 1600
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 20.61919504643963,
|
| 429 |
+
"grad_norm": 1.3451308012008667,
|
| 430 |
+
"learning_rate": 1.4077329243942368e-05,
|
| 431 |
+
"loss": 0.2573,
|
| 432 |
+
"step": 1650
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 21.24767801857585,
|
| 436 |
+
"grad_norm": 2.416846990585327,
|
| 437 |
+
"learning_rate": 1.3658245111227571e-05,
|
| 438 |
+
"loss": 0.2519,
|
| 439 |
+
"step": 1700
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"epoch": 21.24767801857585,
|
| 443 |
+
"eval_accuracy": 0.8986883763003166,
|
| 444 |
+
"eval_f1": 0.8961275295230519,
|
| 445 |
+
"eval_loss": 0.25643372535705566,
|
| 446 |
+
"eval_precision": 0.9051038365282786,
|
| 447 |
+
"eval_recall": 0.8986883763003166,
|
| 448 |
+
"eval_runtime": 10.0812,
|
| 449 |
+
"eval_samples_per_second": 219.319,
|
| 450 |
+
"eval_steps_per_second": 27.477,
|
| 451 |
+
"step": 1700
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 21.86687306501548,
|
| 455 |
+
"grad_norm": 1.3278274536132812,
|
| 456 |
+
"learning_rate": 1.323160500683173e-05,
|
| 457 |
+
"loss": 0.2546,
|
| 458 |
+
"step": 1750
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 22.4953560371517,
|
| 462 |
+
"grad_norm": 2.4079225063323975,
|
| 463 |
+
"learning_rate": 1.2798290140309924e-05,
|
| 464 |
+
"loss": 0.2595,
|
| 465 |
+
"step": 1800
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 22.4953560371517,
|
| 469 |
+
"eval_accuracy": 0.9095431931252826,
|
| 470 |
+
"eval_f1": 0.9079295985892596,
|
| 471 |
+
"eval_loss": 0.23176445066928864,
|
| 472 |
+
"eval_precision": 0.9128858666371455,
|
| 473 |
+
"eval_recall": 0.9095431931252826,
|
| 474 |
+
"eval_runtime": 10.0549,
|
| 475 |
+
"eval_samples_per_second": 219.893,
|
| 476 |
+
"eval_steps_per_second": 27.549,
|
| 477 |
+
"step": 1800
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 23.123839009287927,
|
| 481 |
+
"grad_norm": 1.916286587715149,
|
| 482 |
+
"learning_rate": 1.2359195507694633e-05,
|
| 483 |
+
"loss": 0.2383,
|
| 484 |
+
"step": 1850
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 23.743034055727556,
|
| 488 |
+
"grad_norm": 3.8509888648986816,
|
| 489 |
+
"learning_rate": 1.1915228042914144e-05,
|
| 490 |
+
"loss": 0.2548,
|
| 491 |
+
"step": 1900
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 23.743034055727556,
|
| 495 |
+
"eval_accuracy": 0.9136137494346449,
|
| 496 |
+
"eval_f1": 0.912807074710142,
|
| 497 |
+
"eval_loss": 0.21959343552589417,
|
| 498 |
+
"eval_precision": 0.91415122785338,
|
| 499 |
+
"eval_recall": 0.9136137494346449,
|
| 500 |
+
"eval_runtime": 10.0827,
|
| 501 |
+
"eval_samples_per_second": 219.287,
|
| 502 |
+
"eval_steps_per_second": 27.473,
|
| 503 |
+
"step": 1900
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 24.371517027863778,
|
| 507 |
+
"grad_norm": 1.1695117950439453,
|
| 508 |
+
"learning_rate": 1.1467304744553618e-05,
|
| 509 |
+
"loss": 0.2336,
|
| 510 |
+
"step": 1950
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"epoch": 24.990712074303406,
|
| 514 |
+
"grad_norm": 0.93059903383255,
|
| 515 |
+
"learning_rate": 1.101635078182802e-05,
|
| 516 |
+
"loss": 0.2327,
|
| 517 |
+
"step": 2000
|
| 518 |
+
},
|
| 519 |
+
{
|
| 520 |
+
"epoch": 24.990712074303406,
|
| 521 |
+
"eval_accuracy": 0.9068294889190411,
|
| 522 |
+
"eval_f1": 0.904968418219579,
|
| 523 |
+
"eval_loss": 0.23764155805110931,
|
| 524 |
+
"eval_precision": 0.9110196660671599,
|
| 525 |
+
"eval_recall": 0.9068294889190411,
|
| 526 |
+
"eval_runtime": 10.0796,
|
| 527 |
+
"eval_samples_per_second": 219.354,
|
| 528 |
+
"eval_steps_per_second": 27.481,
|
| 529 |
+
"step": 2000
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 25.61919504643963,
|
| 533 |
+
"grad_norm": 1.0239897966384888,
|
| 534 |
+
"learning_rate": 1.0563297583678877e-05,
|
| 535 |
+
"loss": 0.2351,
|
| 536 |
+
"step": 2050
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 26.24767801857585,
|
| 540 |
+
"grad_norm": 1.6756315231323242,
|
| 541 |
+
"learning_rate": 1.0109080914941825e-05,
|
| 542 |
+
"loss": 0.2563,
|
| 543 |
+
"step": 2100
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 26.24767801857585,
|
| 547 |
+
"eval_accuracy": 0.9027589326096789,
|
| 548 |
+
"eval_f1": 0.9005087315122816,
|
| 549 |
+
"eval_loss": 0.2420978993177414,
|
| 550 |
+
"eval_precision": 0.9082977649283199,
|
| 551 |
+
"eval_recall": 0.9027589326096789,
|
| 552 |
+
"eval_runtime": 10.0674,
|
| 553 |
+
"eval_samples_per_second": 219.621,
|
| 554 |
+
"eval_steps_per_second": 27.515,
|
| 555 |
+
"step": 2100
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 26.86687306501548,
|
| 559 |
+
"grad_norm": 2.134220838546753,
|
| 560 |
+
"learning_rate": 9.65463894355851e-06,
|
| 561 |
+
"loss": 0.2373,
|
| 562 |
+
"step": 2150
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 27.4953560371517,
|
| 566 |
+
"grad_norm": 0.8266007304191589,
|
| 567 |
+
"learning_rate": 9.200910302824964e-06,
|
| 568 |
+
"loss": 0.2348,
|
| 569 |
+
"step": 2200
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 27.4953560371517,
|
| 573 |
+
"eval_accuracy": 0.9109000452284034,
|
| 574 |
+
"eval_f1": 0.9095386566495041,
|
| 575 |
+
"eval_loss": 0.22126996517181396,
|
| 576 |
+
"eval_precision": 0.9132386550038264,
|
| 577 |
+
"eval_recall": 0.9109000452284034,
|
| 578 |
+
"eval_runtime": 10.0881,
|
| 579 |
+
"eval_samples_per_second": 219.169,
|
| 580 |
+
"eval_steps_per_second": 27.458,
|
| 581 |
+
"step": 2200
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 28.123839009287927,
|
| 585 |
+
"grad_norm": 1.1402108669281006,
|
| 586 |
+
"learning_rate": 8.74883215267881e-06,
|
| 587 |
+
"loss": 0.2366,
|
| 588 |
+
"step": 2250
|
| 589 |
+
},
|
| 590 |
+
{
|
| 591 |
+
"epoch": 28.743034055727556,
|
| 592 |
+
"grad_norm": 1.5788310766220093,
|
| 593 |
+
"learning_rate": 8.299338244029646e-06,
|
| 594 |
+
"loss": 0.2427,
|
| 595 |
+
"step": 2300
|
| 596 |
+
},
|
| 597 |
+
{
|
| 598 |
+
"epoch": 28.743034055727556,
|
| 599 |
+
"eval_accuracy": 0.9077340569877883,
|
| 600 |
+
"eval_f1": 0.9059575898869645,
|
| 601 |
+
"eval_loss": 0.23078913986682892,
|
| 602 |
+
"eval_precision": 0.9116313077973748,
|
| 603 |
+
"eval_recall": 0.9077340569877883,
|
| 604 |
+
"eval_runtime": 10.0773,
|
| 605 |
+
"eval_samples_per_second": 219.404,
|
| 606 |
+
"eval_steps_per_second": 27.488,
|
| 607 |
+
"step": 2300
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 29.371517027863778,
|
| 611 |
+
"grad_norm": 1.0100857019424438,
|
| 612 |
+
"learning_rate": 7.853356990130625e-06,
|
| 613 |
+
"loss": 0.2462,
|
| 614 |
+
"step": 2350
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 29.990712074303406,
|
| 618 |
+
"grad_norm": 1.6099952459335327,
|
| 619 |
+
"learning_rate": 7.411809548974792e-06,
|
| 620 |
+
"loss": 0.2166,
|
| 621 |
+
"step": 2400
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 29.990712074303406,
|
| 625 |
+
"eval_accuracy": 0.9140660334690185,
|
| 626 |
+
"eval_f1": 0.9127673040043155,
|
| 627 |
+
"eval_loss": 0.21517515182495117,
|
| 628 |
+
"eval_precision": 0.9164570638030407,
|
| 629 |
+
"eval_recall": 0.9140660334690185,
|
| 630 |
+
"eval_runtime": 10.0698,
|
| 631 |
+
"eval_samples_per_second": 219.568,
|
| 632 |
+
"eval_steps_per_second": 27.508,
|
| 633 |
+
"step": 2400
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 30.61919504643963,
|
| 637 |
+
"grad_norm": 2.7743020057678223,
|
| 638 |
+
"learning_rate": 6.975607920676901e-06,
|
| 639 |
+
"loss": 0.2315,
|
| 640 |
+
"step": 2450
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 31.24767801857585,
|
| 644 |
+
"grad_norm": 2.1687729358673096,
|
| 645 |
+
"learning_rate": 6.545653063770458e-06,
|
| 646 |
+
"loss": 0.2345,
|
| 647 |
+
"step": 2500
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 31.24767801857585,
|
| 651 |
+
"eval_accuracy": 0.9068294889190411,
|
| 652 |
+
"eval_f1": 0.9049003940630713,
|
| 653 |
+
"eval_loss": 0.22831733524799347,
|
| 654 |
+
"eval_precision": 0.9113708908998591,
|
| 655 |
+
"eval_recall": 0.9068294889190411,
|
| 656 |
+
"eval_runtime": 10.1159,
|
| 657 |
+
"eval_samples_per_second": 218.567,
|
| 658 |
+
"eval_steps_per_second": 27.383,
|
| 659 |
+
"step": 2500
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 31.86687306501548,
|
| 663 |
+
"grad_norm": 2.7222249507904053,
|
| 664 |
+
"learning_rate": 6.122833034310794e-06,
|
| 665 |
+
"loss": 0.2283,
|
| 666 |
+
"step": 2550
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 32.4953560371517,
|
| 670 |
+
"grad_norm": 1.5686038732528687,
|
| 671 |
+
"learning_rate": 5.708021151627712e-06,
|
| 672 |
+
"loss": 0.2355,
|
| 673 |
+
"step": 2600
|
| 674 |
+
},
|
| 675 |
+
{
|
| 676 |
+
"epoch": 32.4953560371517,
|
| 677 |
+
"eval_accuracy": 0.9118046132971506,
|
| 678 |
+
"eval_f1": 0.9103077963732922,
|
| 679 |
+
"eval_loss": 0.21731863915920258,
|
| 680 |
+
"eval_precision": 0.9148692757405781,
|
| 681 |
+
"eval_recall": 0.9118046132971506,
|
| 682 |
+
"eval_runtime": 10.0717,
|
| 683 |
+
"eval_samples_per_second": 219.526,
|
| 684 |
+
"eval_steps_per_second": 27.503,
|
| 685 |
+
"step": 2600
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 33.12383900928793,
|
| 689 |
+
"grad_norm": 2.4550020694732666,
|
| 690 |
+
"learning_rate": 5.302074194516291e-06,
|
| 691 |
+
"loss": 0.2207,
|
| 692 |
+
"step": 2650
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 33.743034055727556,
|
| 696 |
+
"grad_norm": 2.3107786178588867,
|
| 697 |
+
"learning_rate": 4.9058306315915826e-06,
|
| 698 |
+
"loss": 0.2291,
|
| 699 |
+
"step": 2700
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 33.743034055727556,
|
| 703 |
+
"eval_accuracy": 0.9127091813658977,
|
| 704 |
+
"eval_f1": 0.9112873759668951,
|
| 705 |
+
"eval_loss": 0.214884415268898,
|
| 706 |
+
"eval_precision": 0.9155228611707341,
|
| 707 |
+
"eval_recall": 0.9127091813658977,
|
| 708 |
+
"eval_runtime": 10.1219,
|
| 709 |
+
"eval_samples_per_second": 218.437,
|
| 710 |
+
"eval_steps_per_second": 27.366,
|
| 711 |
+
"step": 2700
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 34.371517027863774,
|
| 715 |
+
"grad_norm": 1.5490918159484863,
|
| 716 |
+
"learning_rate": 4.52010888946231e-06,
|
| 717 |
+
"loss": 0.2344,
|
| 718 |
+
"step": 2750
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 34.9907120743034,
|
| 722 |
+
"grad_norm": 1.1831104755401611,
|
| 723 |
+
"learning_rate": 4.1457056623005954e-06,
|
| 724 |
+
"loss": 0.2319,
|
| 725 |
+
"step": 2800
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 34.9907120743034,
|
| 729 |
+
"eval_accuracy": 0.9140660334690185,
|
| 730 |
+
"eval_f1": 0.912709884455563,
|
| 731 |
+
"eval_loss": 0.2123376727104187,
|
| 732 |
+
"eval_precision": 0.9167245062463679,
|
| 733 |
+
"eval_recall": 0.9140660334690185,
|
| 734 |
+
"eval_runtime": 10.0267,
|
| 735 |
+
"eval_samples_per_second": 220.512,
|
| 736 |
+
"eval_steps_per_second": 27.626,
|
| 737 |
+
"step": 2800
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 35.61919504643963,
|
| 741 |
+
"grad_norm": 1.047865629196167,
|
| 742 |
+
"learning_rate": 3.7833942662992286e-06,
|
| 743 |
+
"loss": 0.2253,
|
| 744 |
+
"step": 2850
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 36.247678018575854,
|
| 748 |
+
"grad_norm": 1.880428433418274,
|
| 749 |
+
"learning_rate": 3.4339230424153225e-06,
|
| 750 |
+
"loss": 0.222,
|
| 751 |
+
"step": 2900
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 36.247678018575854,
|
| 755 |
+
"eval_accuracy": 0.9181365897783809,
|
| 756 |
+
"eval_f1": 0.9171239767848,
|
| 757 |
+
"eval_loss": 0.2052914798259735,
|
| 758 |
+
"eval_precision": 0.9196609395815928,
|
| 759 |
+
"eval_recall": 0.9181365897783809,
|
| 760 |
+
"eval_runtime": 10.0315,
|
| 761 |
+
"eval_samples_per_second": 220.405,
|
| 762 |
+
"eval_steps_per_second": 27.613,
|
| 763 |
+
"step": 2900
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 36.86687306501548,
|
| 767 |
+
"grad_norm": 1.5235450267791748,
|
| 768 |
+
"learning_rate": 3.098013810699404e-06,
|
| 769 |
+
"loss": 0.2256,
|
| 770 |
+
"step": 2950
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 37.4953560371517,
|
| 774 |
+
"grad_norm": 1.710253357887268,
|
| 775 |
+
"learning_rate": 2.776360379402445e-06,
|
| 776 |
+
"loss": 0.2235,
|
| 777 |
+
"step": 3000
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 37.4953560371517,
|
| 781 |
+
"eval_accuracy": 0.9140660334690185,
|
| 782 |
+
"eval_f1": 0.9127386875836824,
|
| 783 |
+
"eval_loss": 0.21209371089935303,
|
| 784 |
+
"eval_precision": 0.9165890729592928,
|
| 785 |
+
"eval_recall": 0.9140660334690185,
|
| 786 |
+
"eval_runtime": 10.0492,
|
| 787 |
+
"eval_samples_per_second": 220.018,
|
| 788 |
+
"eval_steps_per_second": 27.564,
|
| 789 |
+
"step": 3000
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 38.12383900928793,
|
| 793 |
+
"grad_norm": 1.8632524013519287,
|
| 794 |
+
"learning_rate": 2.469627111940258e-06,
|
| 795 |
+
"loss": 0.2207,
|
| 796 |
+
"step": 3050
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 38.743034055727556,
|
| 800 |
+
"grad_norm": 1.2123332023620605,
|
| 801 |
+
"learning_rate": 2.178447554675136e-06,
|
| 802 |
+
"loss": 0.2221,
|
| 803 |
+
"step": 3100
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 38.743034055727556,
|
| 807 |
+
"eval_accuracy": 0.9194934418815016,
|
| 808 |
+
"eval_f1": 0.9188000621156364,
|
| 809 |
+
"eval_loss": 0.20125794410705566,
|
| 810 |
+
"eval_precision": 0.9199821991038324,
|
| 811 |
+
"eval_recall": 0.9194934418815016,
|
| 812 |
+
"eval_runtime": 10.0477,
|
| 813 |
+
"eval_samples_per_second": 220.05,
|
| 814 |
+
"eval_steps_per_second": 27.568,
|
| 815 |
+
"step": 3100
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 39.371517027863774,
|
| 819 |
+
"grad_norm": 1.6572494506835938,
|
| 820 |
+
"learning_rate": 1.903423128348959e-06,
|
| 821 |
+
"loss": 0.2411,
|
| 822 |
+
"step": 3150
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 39.9907120743034,
|
| 826 |
+
"grad_norm": 1.047083854675293,
|
| 827 |
+
"learning_rate": 1.6451218858706374e-06,
|
| 828 |
+
"loss": 0.2262,
|
| 829 |
+
"step": 3200
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 39.9907120743034,
|
| 833 |
+
"eval_accuracy": 0.9222071460877431,
|
| 834 |
+
"eval_f1": 0.9213530294983667,
|
| 835 |
+
"eval_loss": 0.20286186039447784,
|
| 836 |
+
"eval_precision": 0.9233953907433776,
|
| 837 |
+
"eval_recall": 0.9222071460877431,
|
| 838 |
+
"eval_runtime": 10.0854,
|
| 839 |
+
"eval_samples_per_second": 219.227,
|
| 840 |
+
"eval_steps_per_second": 27.465,
|
| 841 |
+
"step": 3200
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 40.61919504643963,
|
| 845 |
+
"grad_norm": 1.171499490737915,
|
| 846 |
+
"learning_rate": 1.4040773390235463e-06,
|
| 847 |
+
"loss": 0.2213,
|
| 848 |
+
"step": 3250
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 41.247678018575854,
|
| 852 |
+
"grad_norm": 2.6431760787963867,
|
| 853 |
+
"learning_rate": 1.1807873565164507e-06,
|
| 854 |
+
"loss": 0.2171,
|
| 855 |
+
"step": 3300
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 41.247678018575854,
|
| 859 |
+
"eval_accuracy": 0.9181365897783809,
|
| 860 |
+
"eval_f1": 0.9169933978820265,
|
| 861 |
+
"eval_loss": 0.2075406163930893,
|
| 862 |
+
"eval_precision": 0.9202292831462632,
|
| 863 |
+
"eval_recall": 0.9181365897783809,
|
| 864 |
+
"eval_runtime": 10.0343,
|
| 865 |
+
"eval_samples_per_second": 220.344,
|
| 866 |
+
"eval_steps_per_second": 27.605,
|
| 867 |
+
"step": 3300
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 41.86687306501548,
|
| 871 |
+
"grad_norm": 1.0989552736282349,
|
| 872 |
+
"learning_rate": 9.757131356538408e-07,
|
| 873 |
+
"loss": 0.2162,
|
| 874 |
+
"step": 3350
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 42.4953560371517,
|
| 878 |
+
"grad_norm": 1.126570463180542,
|
| 879 |
+
"learning_rate": 7.892782497497642e-07,
|
| 880 |
+
"loss": 0.2268,
|
| 881 |
+
"step": 3400
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 42.4953560371517,
|
| 885 |
+
"eval_accuracy": 0.919041157847128,
|
| 886 |
+
"eval_f1": 0.9179885980860357,
|
| 887 |
+
"eval_loss": 0.2045469731092453,
|
| 888 |
+
"eval_precision": 0.9208108708737649,
|
| 889 |
+
"eval_recall": 0.919041157847128,
|
| 890 |
+
"eval_runtime": 10.0534,
|
| 891 |
+
"eval_samples_per_second": 219.925,
|
| 892 |
+
"eval_steps_per_second": 27.553,
|
| 893 |
+
"step": 3400
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 43.12383900928793,
|
| 897 |
+
"grad_norm": 1.0992859601974487,
|
| 898 |
+
"learning_rate": 6.218677732526035e-07,
|
| 899 |
+
"loss": 0.2284,
|
| 900 |
+
"step": 3450
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 43.743034055727556,
|
| 904 |
+
"grad_norm": 0.8807310461997986,
|
| 905 |
+
"learning_rate": 4.7382748638786336e-07,
|
| 906 |
+
"loss": 0.2222,
|
| 907 |
+
"step": 3500
|
| 908 |
+
},
|
| 909 |
+
{
|
| 910 |
+
"epoch": 43.743034055727556,
|
| 911 |
+
"eval_accuracy": 0.9203980099502488,
|
| 912 |
+
"eval_f1": 0.9193757241191305,
|
| 913 |
+
"eval_loss": 0.20502084493637085,
|
| 914 |
+
"eval_precision": 0.9221532972644962,
|
| 915 |
+
"eval_recall": 0.9203980099502488,
|
| 916 |
+
"eval_runtime": 10.0674,
|
| 917 |
+
"eval_samples_per_second": 219.619,
|
| 918 |
+
"eval_steps_per_second": 27.514,
|
| 919 |
+
"step": 3500
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 44.371517027863774,
|
| 923 |
+
"grad_norm": 1.0601508617401123,
|
| 924 |
+
"learning_rate": 3.454631609617487e-07,
|
| 925 |
+
"loss": 0.2232,
|
| 926 |
+
"step": 3550
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 44.9907120743034,
|
| 930 |
+
"grad_norm": 1.5610824823379517,
|
| 931 |
+
"learning_rate": 2.370399288006664e-07,
|
| 932 |
+
"loss": 0.2169,
|
| 933 |
+
"step": 3600
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 44.9907120743034,
|
| 937 |
+
"eval_accuracy": 0.9176843057440073,
|
| 938 |
+
"eval_f1": 0.9165481260183203,
|
| 939 |
+
"eval_loss": 0.20695888996124268,
|
| 940 |
+
"eval_precision": 0.9197013630605291,
|
| 941 |
+
"eval_recall": 0.9176843057440073,
|
| 942 |
+
"eval_runtime": 10.0683,
|
| 943 |
+
"eval_samples_per_second": 219.6,
|
| 944 |
+
"eval_steps_per_second": 27.512,
|
| 945 |
+
"step": 3600
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 45.61919504643963,
|
| 949 |
+
"grad_norm": 1.097316861152649,
|
| 950 |
+
"learning_rate": 1.4878173413111485e-07,
|
| 951 |
+
"loss": 0.2269,
|
| 952 |
+
"step": 3650
|
| 953 |
+
},
|
| 954 |
+
{
|
| 955 |
+
"epoch": 46.247678018575854,
|
| 956 |
+
"grad_norm": 1.628519892692566,
|
| 957 |
+
"learning_rate": 8.087087103106461e-08,
|
| 958 |
+
"loss": 0.2245,
|
| 959 |
+
"step": 3700
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 46.247678018575854,
|
| 963 |
+
"eval_accuracy": 0.9181365897783809,
|
| 964 |
+
"eval_f1": 0.9170198637903132,
|
| 965 |
+
"eval_loss": 0.20643840730190277,
|
| 966 |
+
"eval_precision": 0.9201088948883855,
|
| 967 |
+
"eval_recall": 0.9181365897783809,
|
| 968 |
+
"eval_runtime": 10.0506,
|
| 969 |
+
"eval_samples_per_second": 219.986,
|
| 970 |
+
"eval_steps_per_second": 27.56,
|
| 971 |
+
"step": 3700
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 46.86687306501548,
|
| 975 |
+
"grad_norm": 0.9928280115127563,
|
| 976 |
+
"learning_rate": 3.3447606908196815e-08,
|
| 977 |
+
"loss": 0.212,
|
| 978 |
+
"step": 3750
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 47.4953560371517,
|
| 982 |
+
"grad_norm": 1.0919773578643799,
|
| 983 |
+
"learning_rate": 6.609892782699634e-09,
|
| 984 |
+
"loss": 0.2148,
|
| 985 |
+
"step": 3800
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 47.4953560371517,
|
| 989 |
+
"eval_accuracy": 0.9181365897783809,
|
| 990 |
+
"eval_f1": 0.9170198637903132,
|
| 991 |
+
"eval_loss": 0.20658649504184723,
|
| 992 |
+
"eval_precision": 0.9201088948883855,
|
| 993 |
+
"eval_recall": 0.9181365897783809,
|
| 994 |
+
"eval_runtime": 10.0329,
|
| 995 |
+
"eval_samples_per_second": 220.375,
|
| 996 |
+
"eval_steps_per_second": 27.609,
|
| 997 |
+
"step": 3800
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 47.9907120743034,
|
| 1001 |
+
"step": 3840,
|
| 1002 |
+
"total_flos": 2.738563601264935e+19,
|
| 1003 |
+
"train_loss": 0.3012936460475127,
|
| 1004 |
+
"train_runtime": 3086.985,
|
| 1005 |
+
"train_samples_per_second": 80.203,
|
| 1006 |
+
"train_steps_per_second": 1.244
|
| 1007 |
+
}
|
| 1008 |
+
],
|
| 1009 |
+
"logging_steps": 50,
|
| 1010 |
+
"max_steps": 3840,
|
| 1011 |
+
"num_input_tokens_seen": 0,
|
| 1012 |
+
"num_train_epochs": 48,
|
| 1013 |
+
"save_steps": 100,
|
| 1014 |
+
"stateful_callbacks": {
|
| 1015 |
+
"TrainerControl": {
|
| 1016 |
+
"args": {
|
| 1017 |
+
"should_epoch_stop": false,
|
| 1018 |
+
"should_evaluate": false,
|
| 1019 |
+
"should_log": false,
|
| 1020 |
+
"should_save": true,
|
| 1021 |
+
"should_training_stop": true
|
| 1022 |
+
},
|
| 1023 |
+
"attributes": {}
|
| 1024 |
+
}
|
| 1025 |
+
},
|
| 1026 |
+
"total_flos": 2.738563601264935e+19,
|
| 1027 |
+
"train_batch_size": 16,
|
| 1028 |
+
"trial_name": null,
|
| 1029 |
+
"trial_params": null
|
| 1030 |
+
}
|