End of training
Browse files- README.md +3 -3
- all_results.json +13 -0
- eval_results.json +8 -0
- runs/Jan24_17-33-02_ef765fa7224c/events.out.tfevents.1737741311.ef765fa7224c.194.3 +3 -0
- train_results.json +8 -0
- trainer_state.json +2966 -0
README.md
CHANGED
|
@@ -23,7 +23,7 @@ model-index:
|
|
| 23 |
metrics:
|
| 24 |
- name: Accuracy
|
| 25 |
type: accuracy
|
| 26 |
-
value: 0.
|
| 27 |
---
|
| 28 |
|
| 29 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
@@ -33,8 +33,8 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 33 |
|
| 34 |
This model is a fine-tuned version of [facebook/vit-msn-small](https://huggingface.co/facebook/vit-msn-small) on the imagefolder dataset.
|
| 35 |
It achieves the following results on the evaluation set:
|
| 36 |
-
- Loss: 0.
|
| 37 |
-
- Accuracy: 0.
|
| 38 |
|
| 39 |
## Model description
|
| 40 |
|
|
|
|
| 23 |
metrics:
|
| 24 |
- name: Accuracy
|
| 25 |
type: accuracy
|
| 26 |
+
value: 0.7746478873239436
|
| 27 |
---
|
| 28 |
|
| 29 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
|
|
| 33 |
|
| 34 |
This model is a fine-tuned version of [facebook/vit-msn-small](https://huggingface.co/facebook/vit-msn-small) on the imagefolder dataset.
|
| 35 |
It achieves the following results on the evaluation set:
|
| 36 |
+
- Loss: 0.5203
|
| 37 |
+
- Accuracy: 0.7746
|
| 38 |
|
| 39 |
## Model description
|
| 40 |
|
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 285.7142857142857,
|
| 3 |
+
"eval_accuracy": 0.7746478873239436,
|
| 4 |
+
"eval_loss": 0.520267128944397,
|
| 5 |
+
"eval_runtime": 0.6536,
|
| 6 |
+
"eval_samples_per_second": 217.252,
|
| 7 |
+
"eval_steps_per_second": 4.59,
|
| 8 |
+
"total_flos": 2.387684396739963e+18,
|
| 9 |
+
"train_loss": 0.3142274961471558,
|
| 10 |
+
"train_runtime": 1304.9715,
|
| 11 |
+
"train_samples_per_second": 163.605,
|
| 12 |
+
"train_steps_per_second": 0.383
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 285.7142857142857,
|
| 3 |
+
"eval_accuracy": 0.7746478873239436,
|
| 4 |
+
"eval_loss": 0.520267128944397,
|
| 5 |
+
"eval_runtime": 0.6536,
|
| 6 |
+
"eval_samples_per_second": 217.252,
|
| 7 |
+
"eval_steps_per_second": 4.59
|
| 8 |
+
}
|
runs/Jan24_17-33-02_ef765fa7224c/events.out.tfevents.1737741311.ef765fa7224c.194.3
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:604dd8fb6e8fc54c87c603d0ea6014defff7f77476eac5f75a4fb30925f44fdb
|
| 3 |
+
size 411
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 285.7142857142857,
|
| 3 |
+
"total_flos": 2.387684396739963e+18,
|
| 4 |
+
"train_loss": 0.3142274961471558,
|
| 5 |
+
"train_runtime": 1304.9715,
|
| 6 |
+
"train_samples_per_second": 163.605,
|
| 7 |
+
"train_steps_per_second": 0.383
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2966 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.520267128944397,
|
| 3 |
+
"best_model_checkpoint": "vit-msn-small-beta-fia-manually-enhanced_test_2/checkpoint-94",
|
| 4 |
+
"epoch": 285.7142857142857,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 500,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.5714285714285714,
|
| 13 |
+
"eval_accuracy": 0.7464788732394366,
|
| 14 |
+
"eval_loss": 0.6037383675575256,
|
| 15 |
+
"eval_runtime": 0.7043,
|
| 16 |
+
"eval_samples_per_second": 201.628,
|
| 17 |
+
"eval_steps_per_second": 4.26,
|
| 18 |
+
"step": 1
|
| 19 |
+
},
|
| 20 |
+
{
|
| 21 |
+
"epoch": 1.7142857142857144,
|
| 22 |
+
"eval_accuracy": 0.7323943661971831,
|
| 23 |
+
"eval_loss": 0.6071274280548096,
|
| 24 |
+
"eval_runtime": 0.6313,
|
| 25 |
+
"eval_samples_per_second": 224.947,
|
| 26 |
+
"eval_steps_per_second": 4.752,
|
| 27 |
+
"step": 3
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 2.857142857142857,
|
| 31 |
+
"eval_accuracy": 0.7183098591549296,
|
| 32 |
+
"eval_loss": 0.6120076775550842,
|
| 33 |
+
"eval_runtime": 0.6621,
|
| 34 |
+
"eval_samples_per_second": 214.465,
|
| 35 |
+
"eval_steps_per_second": 4.531,
|
| 36 |
+
"step": 5
|
| 37 |
+
},
|
| 38 |
+
{
|
| 39 |
+
"epoch": 4.0,
|
| 40 |
+
"eval_accuracy": 0.7183098591549296,
|
| 41 |
+
"eval_loss": 0.6187500953674316,
|
| 42 |
+
"eval_runtime": 0.617,
|
| 43 |
+
"eval_samples_per_second": 230.15,
|
| 44 |
+
"eval_steps_per_second": 4.862,
|
| 45 |
+
"step": 7
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 4.571428571428571,
|
| 49 |
+
"eval_accuracy": 0.7183098591549296,
|
| 50 |
+
"eval_loss": 0.6205877661705017,
|
| 51 |
+
"eval_runtime": 0.6065,
|
| 52 |
+
"eval_samples_per_second": 234.118,
|
| 53 |
+
"eval_steps_per_second": 4.946,
|
| 54 |
+
"step": 8
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 5.714285714285714,
|
| 58 |
+
"grad_norm": 4.976908206939697,
|
| 59 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 60 |
+
"loss": 0.4866,
|
| 61 |
+
"step": 10
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 5.714285714285714,
|
| 65 |
+
"eval_accuracy": 0.6971830985915493,
|
| 66 |
+
"eval_loss": 0.6271856427192688,
|
| 67 |
+
"eval_runtime": 0.6179,
|
| 68 |
+
"eval_samples_per_second": 229.802,
|
| 69 |
+
"eval_steps_per_second": 4.855,
|
| 70 |
+
"step": 10
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"epoch": 6.857142857142857,
|
| 74 |
+
"eval_accuracy": 0.6901408450704225,
|
| 75 |
+
"eval_loss": 0.6354550123214722,
|
| 76 |
+
"eval_runtime": 0.615,
|
| 77 |
+
"eval_samples_per_second": 230.884,
|
| 78 |
+
"eval_steps_per_second": 4.878,
|
| 79 |
+
"step": 12
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 8.0,
|
| 83 |
+
"eval_accuracy": 0.6901408450704225,
|
| 84 |
+
"eval_loss": 0.6398630142211914,
|
| 85 |
+
"eval_runtime": 0.6614,
|
| 86 |
+
"eval_samples_per_second": 214.71,
|
| 87 |
+
"eval_steps_per_second": 4.536,
|
| 88 |
+
"step": 14
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"epoch": 8.571428571428571,
|
| 92 |
+
"eval_accuracy": 0.6830985915492958,
|
| 93 |
+
"eval_loss": 0.6364104151725769,
|
| 94 |
+
"eval_runtime": 0.6148,
|
| 95 |
+
"eval_samples_per_second": 230.963,
|
| 96 |
+
"eval_steps_per_second": 4.88,
|
| 97 |
+
"step": 15
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"epoch": 9.714285714285714,
|
| 101 |
+
"eval_accuracy": 0.6830985915492958,
|
| 102 |
+
"eval_loss": 0.6294547915458679,
|
| 103 |
+
"eval_runtime": 0.6631,
|
| 104 |
+
"eval_samples_per_second": 214.147,
|
| 105 |
+
"eval_steps_per_second": 4.524,
|
| 106 |
+
"step": 17
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"epoch": 10.857142857142858,
|
| 110 |
+
"eval_accuracy": 0.6901408450704225,
|
| 111 |
+
"eval_loss": 0.6287956833839417,
|
| 112 |
+
"eval_runtime": 0.6702,
|
| 113 |
+
"eval_samples_per_second": 211.869,
|
| 114 |
+
"eval_steps_per_second": 4.476,
|
| 115 |
+
"step": 19
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 11.428571428571429,
|
| 119 |
+
"grad_norm": 4.726110458374023,
|
| 120 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 121 |
+
"loss": 0.4519,
|
| 122 |
+
"step": 20
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 12.0,
|
| 126 |
+
"eval_accuracy": 0.6901408450704225,
|
| 127 |
+
"eval_loss": 0.6185110211372375,
|
| 128 |
+
"eval_runtime": 0.6632,
|
| 129 |
+
"eval_samples_per_second": 214.103,
|
| 130 |
+
"eval_steps_per_second": 4.523,
|
| 131 |
+
"step": 21
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 12.571428571428571,
|
| 135 |
+
"eval_accuracy": 0.6901408450704225,
|
| 136 |
+
"eval_loss": 0.6159186363220215,
|
| 137 |
+
"eval_runtime": 0.6273,
|
| 138 |
+
"eval_samples_per_second": 226.383,
|
| 139 |
+
"eval_steps_per_second": 4.783,
|
| 140 |
+
"step": 22
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 13.714285714285714,
|
| 144 |
+
"eval_accuracy": 0.6971830985915493,
|
| 145 |
+
"eval_loss": 0.6113001704216003,
|
| 146 |
+
"eval_runtime": 0.667,
|
| 147 |
+
"eval_samples_per_second": 212.906,
|
| 148 |
+
"eval_steps_per_second": 4.498,
|
| 149 |
+
"step": 24
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 14.857142857142858,
|
| 153 |
+
"eval_accuracy": 0.6901408450704225,
|
| 154 |
+
"eval_loss": 0.5986557006835938,
|
| 155 |
+
"eval_runtime": 0.6998,
|
| 156 |
+
"eval_samples_per_second": 202.92,
|
| 157 |
+
"eval_steps_per_second": 4.287,
|
| 158 |
+
"step": 26
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 16.0,
|
| 162 |
+
"eval_accuracy": 0.6971830985915493,
|
| 163 |
+
"eval_loss": 0.6017070412635803,
|
| 164 |
+
"eval_runtime": 0.6484,
|
| 165 |
+
"eval_samples_per_second": 218.996,
|
| 166 |
+
"eval_steps_per_second": 4.627,
|
| 167 |
+
"step": 28
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 16.571428571428573,
|
| 171 |
+
"eval_accuracy": 0.6971830985915493,
|
| 172 |
+
"eval_loss": 0.6066802144050598,
|
| 173 |
+
"eval_runtime": 0.6148,
|
| 174 |
+
"eval_samples_per_second": 230.971,
|
| 175 |
+
"eval_steps_per_second": 4.88,
|
| 176 |
+
"step": 29
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 17.142857142857142,
|
| 180 |
+
"grad_norm": 9.665369033813477,
|
| 181 |
+
"learning_rate": 3e-06,
|
| 182 |
+
"loss": 0.437,
|
| 183 |
+
"step": 30
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 17.714285714285715,
|
| 187 |
+
"eval_accuracy": 0.6619718309859155,
|
| 188 |
+
"eval_loss": 0.6062361598014832,
|
| 189 |
+
"eval_runtime": 0.628,
|
| 190 |
+
"eval_samples_per_second": 226.123,
|
| 191 |
+
"eval_steps_per_second": 4.777,
|
| 192 |
+
"step": 31
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 18.857142857142858,
|
| 196 |
+
"eval_accuracy": 0.6901408450704225,
|
| 197 |
+
"eval_loss": 0.5966038107872009,
|
| 198 |
+
"eval_runtime": 0.7056,
|
| 199 |
+
"eval_samples_per_second": 201.243,
|
| 200 |
+
"eval_steps_per_second": 4.252,
|
| 201 |
+
"step": 33
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 20.0,
|
| 205 |
+
"eval_accuracy": 0.7112676056338029,
|
| 206 |
+
"eval_loss": 0.5857767462730408,
|
| 207 |
+
"eval_runtime": 0.6152,
|
| 208 |
+
"eval_samples_per_second": 230.806,
|
| 209 |
+
"eval_steps_per_second": 4.876,
|
| 210 |
+
"step": 35
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"epoch": 20.571428571428573,
|
| 214 |
+
"eval_accuracy": 0.704225352112676,
|
| 215 |
+
"eval_loss": 0.5888840556144714,
|
| 216 |
+
"eval_runtime": 0.6769,
|
| 217 |
+
"eval_samples_per_second": 209.765,
|
| 218 |
+
"eval_steps_per_second": 4.432,
|
| 219 |
+
"step": 36
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 21.714285714285715,
|
| 223 |
+
"eval_accuracy": 0.7183098591549296,
|
| 224 |
+
"eval_loss": 0.5767799615859985,
|
| 225 |
+
"eval_runtime": 0.6222,
|
| 226 |
+
"eval_samples_per_second": 228.219,
|
| 227 |
+
"eval_steps_per_second": 4.822,
|
| 228 |
+
"step": 38
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 22.857142857142858,
|
| 232 |
+
"grad_norm": 7.109732151031494,
|
| 233 |
+
"learning_rate": 4.000000000000001e-06,
|
| 234 |
+
"loss": 0.4353,
|
| 235 |
+
"step": 40
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 22.857142857142858,
|
| 239 |
+
"eval_accuracy": 0.7183098591549296,
|
| 240 |
+
"eval_loss": 0.5751827359199524,
|
| 241 |
+
"eval_runtime": 0.6871,
|
| 242 |
+
"eval_samples_per_second": 206.668,
|
| 243 |
+
"eval_steps_per_second": 4.366,
|
| 244 |
+
"step": 40
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 24.0,
|
| 248 |
+
"eval_accuracy": 0.7183098591549296,
|
| 249 |
+
"eval_loss": 0.5728872418403625,
|
| 250 |
+
"eval_runtime": 0.61,
|
| 251 |
+
"eval_samples_per_second": 232.791,
|
| 252 |
+
"eval_steps_per_second": 4.918,
|
| 253 |
+
"step": 42
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"epoch": 24.571428571428573,
|
| 257 |
+
"eval_accuracy": 0.6971830985915493,
|
| 258 |
+
"eval_loss": 0.5909363627433777,
|
| 259 |
+
"eval_runtime": 0.6389,
|
| 260 |
+
"eval_samples_per_second": 222.273,
|
| 261 |
+
"eval_steps_per_second": 4.696,
|
| 262 |
+
"step": 43
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 25.714285714285715,
|
| 266 |
+
"eval_accuracy": 0.676056338028169,
|
| 267 |
+
"eval_loss": 0.6038447618484497,
|
| 268 |
+
"eval_runtime": 0.611,
|
| 269 |
+
"eval_samples_per_second": 232.415,
|
| 270 |
+
"eval_steps_per_second": 4.91,
|
| 271 |
+
"step": 45
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 26.857142857142858,
|
| 275 |
+
"eval_accuracy": 0.6901408450704225,
|
| 276 |
+
"eval_loss": 0.5903939008712769,
|
| 277 |
+
"eval_runtime": 0.6408,
|
| 278 |
+
"eval_samples_per_second": 221.615,
|
| 279 |
+
"eval_steps_per_second": 4.682,
|
| 280 |
+
"step": 47
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 28.0,
|
| 284 |
+
"eval_accuracy": 0.6830985915492958,
|
| 285 |
+
"eval_loss": 0.5846758484840393,
|
| 286 |
+
"eval_runtime": 0.6786,
|
| 287 |
+
"eval_samples_per_second": 209.242,
|
| 288 |
+
"eval_steps_per_second": 4.421,
|
| 289 |
+
"step": 49
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 28.571428571428573,
|
| 293 |
+
"grad_norm": 6.319047927856445,
|
| 294 |
+
"learning_rate": 5e-06,
|
| 295 |
+
"loss": 0.4141,
|
| 296 |
+
"step": 50
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 28.571428571428573,
|
| 300 |
+
"eval_accuracy": 0.7112676056338029,
|
| 301 |
+
"eval_loss": 0.5615125894546509,
|
| 302 |
+
"eval_runtime": 0.7037,
|
| 303 |
+
"eval_samples_per_second": 201.787,
|
| 304 |
+
"eval_steps_per_second": 4.263,
|
| 305 |
+
"step": 50
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 29.714285714285715,
|
| 309 |
+
"eval_accuracy": 0.7253521126760564,
|
| 310 |
+
"eval_loss": 0.5543914437294006,
|
| 311 |
+
"eval_runtime": 0.707,
|
| 312 |
+
"eval_samples_per_second": 200.841,
|
| 313 |
+
"eval_steps_per_second": 4.243,
|
| 314 |
+
"step": 52
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 30.857142857142858,
|
| 318 |
+
"eval_accuracy": 0.6690140845070423,
|
| 319 |
+
"eval_loss": 0.5903507471084595,
|
| 320 |
+
"eval_runtime": 0.6246,
|
| 321 |
+
"eval_samples_per_second": 227.36,
|
| 322 |
+
"eval_steps_per_second": 4.803,
|
| 323 |
+
"step": 54
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 32.0,
|
| 327 |
+
"eval_accuracy": 0.6830985915492958,
|
| 328 |
+
"eval_loss": 0.5948190689086914,
|
| 329 |
+
"eval_runtime": 0.67,
|
| 330 |
+
"eval_samples_per_second": 211.956,
|
| 331 |
+
"eval_steps_per_second": 4.478,
|
| 332 |
+
"step": 56
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 32.57142857142857,
|
| 336 |
+
"eval_accuracy": 0.6971830985915493,
|
| 337 |
+
"eval_loss": 0.5799919962882996,
|
| 338 |
+
"eval_runtime": 0.6151,
|
| 339 |
+
"eval_samples_per_second": 230.865,
|
| 340 |
+
"eval_steps_per_second": 4.877,
|
| 341 |
+
"step": 57
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 33.714285714285715,
|
| 345 |
+
"eval_accuracy": 0.6971830985915493,
|
| 346 |
+
"eval_loss": 0.5902391076087952,
|
| 347 |
+
"eval_runtime": 0.6636,
|
| 348 |
+
"eval_samples_per_second": 213.993,
|
| 349 |
+
"eval_steps_per_second": 4.521,
|
| 350 |
+
"step": 59
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 34.285714285714285,
|
| 354 |
+
"grad_norm": 7.730890274047852,
|
| 355 |
+
"learning_rate": 6e-06,
|
| 356 |
+
"loss": 0.4066,
|
| 357 |
+
"step": 60
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 34.857142857142854,
|
| 361 |
+
"eval_accuracy": 0.6690140845070423,
|
| 362 |
+
"eval_loss": 0.5950306057929993,
|
| 363 |
+
"eval_runtime": 0.5961,
|
| 364 |
+
"eval_samples_per_second": 238.215,
|
| 365 |
+
"eval_steps_per_second": 5.033,
|
| 366 |
+
"step": 61
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 36.0,
|
| 370 |
+
"eval_accuracy": 0.7323943661971831,
|
| 371 |
+
"eval_loss": 0.550020694732666,
|
| 372 |
+
"eval_runtime": 0.7196,
|
| 373 |
+
"eval_samples_per_second": 197.33,
|
| 374 |
+
"eval_steps_per_second": 4.169,
|
| 375 |
+
"step": 63
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 36.57142857142857,
|
| 379 |
+
"eval_accuracy": 0.7323943661971831,
|
| 380 |
+
"eval_loss": 0.5470194220542908,
|
| 381 |
+
"eval_runtime": 0.6083,
|
| 382 |
+
"eval_samples_per_second": 233.447,
|
| 383 |
+
"eval_steps_per_second": 4.932,
|
| 384 |
+
"step": 64
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 37.714285714285715,
|
| 388 |
+
"eval_accuracy": 0.6901408450704225,
|
| 389 |
+
"eval_loss": 0.5858845710754395,
|
| 390 |
+
"eval_runtime": 0.6227,
|
| 391 |
+
"eval_samples_per_second": 228.055,
|
| 392 |
+
"eval_steps_per_second": 4.818,
|
| 393 |
+
"step": 66
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 38.857142857142854,
|
| 397 |
+
"eval_accuracy": 0.6830985915492958,
|
| 398 |
+
"eval_loss": 0.5954940319061279,
|
| 399 |
+
"eval_runtime": 0.6205,
|
| 400 |
+
"eval_samples_per_second": 228.835,
|
| 401 |
+
"eval_steps_per_second": 4.835,
|
| 402 |
+
"step": 68
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 40.0,
|
| 406 |
+
"grad_norm": 13.489781379699707,
|
| 407 |
+
"learning_rate": 7e-06,
|
| 408 |
+
"loss": 0.3827,
|
| 409 |
+
"step": 70
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 40.0,
|
| 413 |
+
"eval_accuracy": 0.676056338028169,
|
| 414 |
+
"eval_loss": 0.5966904163360596,
|
| 415 |
+
"eval_runtime": 0.6376,
|
| 416 |
+
"eval_samples_per_second": 222.718,
|
| 417 |
+
"eval_steps_per_second": 4.705,
|
| 418 |
+
"step": 70
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 40.57142857142857,
|
| 422 |
+
"eval_accuracy": 0.6901408450704225,
|
| 423 |
+
"eval_loss": 0.5809391736984253,
|
| 424 |
+
"eval_runtime": 0.6365,
|
| 425 |
+
"eval_samples_per_second": 223.09,
|
| 426 |
+
"eval_steps_per_second": 4.713,
|
| 427 |
+
"step": 71
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 41.714285714285715,
|
| 431 |
+
"eval_accuracy": 0.6971830985915493,
|
| 432 |
+
"eval_loss": 0.5720686316490173,
|
| 433 |
+
"eval_runtime": 0.6634,
|
| 434 |
+
"eval_samples_per_second": 214.062,
|
| 435 |
+
"eval_steps_per_second": 4.522,
|
| 436 |
+
"step": 73
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 42.857142857142854,
|
| 440 |
+
"eval_accuracy": 0.6830985915492958,
|
| 441 |
+
"eval_loss": 0.6018939018249512,
|
| 442 |
+
"eval_runtime": 0.6125,
|
| 443 |
+
"eval_samples_per_second": 231.824,
|
| 444 |
+
"eval_steps_per_second": 4.898,
|
| 445 |
+
"step": 75
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 44.0,
|
| 449 |
+
"eval_accuracy": 0.6901408450704225,
|
| 450 |
+
"eval_loss": 0.6070601940155029,
|
| 451 |
+
"eval_runtime": 0.6927,
|
| 452 |
+
"eval_samples_per_second": 204.986,
|
| 453 |
+
"eval_steps_per_second": 4.331,
|
| 454 |
+
"step": 77
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 44.57142857142857,
|
| 458 |
+
"eval_accuracy": 0.6971830985915493,
|
| 459 |
+
"eval_loss": 0.5961697697639465,
|
| 460 |
+
"eval_runtime": 0.6369,
|
| 461 |
+
"eval_samples_per_second": 222.959,
|
| 462 |
+
"eval_steps_per_second": 4.71,
|
| 463 |
+
"step": 78
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 45.714285714285715,
|
| 467 |
+
"grad_norm": 6.058713912963867,
|
| 468 |
+
"learning_rate": 8.000000000000001e-06,
|
| 469 |
+
"loss": 0.37,
|
| 470 |
+
"step": 80
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 45.714285714285715,
|
| 474 |
+
"eval_accuracy": 0.6830985915492958,
|
| 475 |
+
"eval_loss": 0.6113830804824829,
|
| 476 |
+
"eval_runtime": 0.6224,
|
| 477 |
+
"eval_samples_per_second": 228.164,
|
| 478 |
+
"eval_steps_per_second": 4.82,
|
| 479 |
+
"step": 80
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 46.857142857142854,
|
| 483 |
+
"eval_accuracy": 0.7183098591549296,
|
| 484 |
+
"eval_loss": 0.5594072341918945,
|
| 485 |
+
"eval_runtime": 0.6975,
|
| 486 |
+
"eval_samples_per_second": 203.576,
|
| 487 |
+
"eval_steps_per_second": 4.301,
|
| 488 |
+
"step": 82
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 48.0,
|
| 492 |
+
"eval_accuracy": 0.7323943661971831,
|
| 493 |
+
"eval_loss": 0.5492833852767944,
|
| 494 |
+
"eval_runtime": 0.6605,
|
| 495 |
+
"eval_samples_per_second": 214.993,
|
| 496 |
+
"eval_steps_per_second": 4.542,
|
| 497 |
+
"step": 84
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 48.57142857142857,
|
| 501 |
+
"eval_accuracy": 0.7112676056338029,
|
| 502 |
+
"eval_loss": 0.5743516087532043,
|
| 503 |
+
"eval_runtime": 0.627,
|
| 504 |
+
"eval_samples_per_second": 226.457,
|
| 505 |
+
"eval_steps_per_second": 4.784,
|
| 506 |
+
"step": 85
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 49.714285714285715,
|
| 510 |
+
"eval_accuracy": 0.7183098591549296,
|
| 511 |
+
"eval_loss": 0.5442511439323425,
|
| 512 |
+
"eval_runtime": 0.6291,
|
| 513 |
+
"eval_samples_per_second": 225.736,
|
| 514 |
+
"eval_steps_per_second": 4.769,
|
| 515 |
+
"step": 87
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 50.857142857142854,
|
| 519 |
+
"eval_accuracy": 0.7323943661971831,
|
| 520 |
+
"eval_loss": 0.5469312071800232,
|
| 521 |
+
"eval_runtime": 0.6973,
|
| 522 |
+
"eval_samples_per_second": 203.643,
|
| 523 |
+
"eval_steps_per_second": 4.302,
|
| 524 |
+
"step": 89
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 51.42857142857143,
|
| 528 |
+
"grad_norm": 5.536169052124023,
|
| 529 |
+
"learning_rate": 9e-06,
|
| 530 |
+
"loss": 0.3797,
|
| 531 |
+
"step": 90
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 52.0,
|
| 535 |
+
"eval_accuracy": 0.6830985915492958,
|
| 536 |
+
"eval_loss": 0.6002964377403259,
|
| 537 |
+
"eval_runtime": 0.6896,
|
| 538 |
+
"eval_samples_per_second": 205.902,
|
| 539 |
+
"eval_steps_per_second": 4.35,
|
| 540 |
+
"step": 91
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 52.57142857142857,
|
| 544 |
+
"eval_accuracy": 0.6901408450704225,
|
| 545 |
+
"eval_loss": 0.60481196641922,
|
| 546 |
+
"eval_runtime": 0.6487,
|
| 547 |
+
"eval_samples_per_second": 218.884,
|
| 548 |
+
"eval_steps_per_second": 4.624,
|
| 549 |
+
"step": 92
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 53.714285714285715,
|
| 553 |
+
"eval_accuracy": 0.7746478873239436,
|
| 554 |
+
"eval_loss": 0.520267128944397,
|
| 555 |
+
"eval_runtime": 0.6631,
|
| 556 |
+
"eval_samples_per_second": 214.158,
|
| 557 |
+
"eval_steps_per_second": 4.524,
|
| 558 |
+
"step": 94
|
| 559 |
+
},
|
| 560 |
+
{
|
| 561 |
+
"epoch": 54.857142857142854,
|
| 562 |
+
"eval_accuracy": 0.7535211267605634,
|
| 563 |
+
"eval_loss": 0.5326515436172485,
|
| 564 |
+
"eval_runtime": 0.64,
|
| 565 |
+
"eval_samples_per_second": 221.883,
|
| 566 |
+
"eval_steps_per_second": 4.688,
|
| 567 |
+
"step": 96
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 56.0,
|
| 571 |
+
"eval_accuracy": 0.6338028169014085,
|
| 572 |
+
"eval_loss": 0.6414005756378174,
|
| 573 |
+
"eval_runtime": 0.6942,
|
| 574 |
+
"eval_samples_per_second": 204.548,
|
| 575 |
+
"eval_steps_per_second": 4.321,
|
| 576 |
+
"step": 98
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 56.57142857142857,
|
| 580 |
+
"eval_accuracy": 0.6197183098591549,
|
| 581 |
+
"eval_loss": 0.6561903357505798,
|
| 582 |
+
"eval_runtime": 0.6607,
|
| 583 |
+
"eval_samples_per_second": 214.94,
|
| 584 |
+
"eval_steps_per_second": 4.541,
|
| 585 |
+
"step": 99
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 57.142857142857146,
|
| 589 |
+
"grad_norm": 7.4128217697143555,
|
| 590 |
+
"learning_rate": 1e-05,
|
| 591 |
+
"loss": 0.3715,
|
| 592 |
+
"step": 100
|
| 593 |
+
},
|
| 594 |
+
{
|
| 595 |
+
"epoch": 57.714285714285715,
|
| 596 |
+
"eval_accuracy": 0.7183098591549296,
|
| 597 |
+
"eval_loss": 0.575397253036499,
|
| 598 |
+
"eval_runtime": 0.6153,
|
| 599 |
+
"eval_samples_per_second": 230.765,
|
| 600 |
+
"eval_steps_per_second": 4.875,
|
| 601 |
+
"step": 101
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 58.857142857142854,
|
| 605 |
+
"eval_accuracy": 0.7253521126760564,
|
| 606 |
+
"eval_loss": 0.5671846270561218,
|
| 607 |
+
"eval_runtime": 0.6933,
|
| 608 |
+
"eval_samples_per_second": 204.809,
|
| 609 |
+
"eval_steps_per_second": 4.327,
|
| 610 |
+
"step": 103
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 60.0,
|
| 614 |
+
"eval_accuracy": 0.6901408450704225,
|
| 615 |
+
"eval_loss": 0.6060019731521606,
|
| 616 |
+
"eval_runtime": 0.6652,
|
| 617 |
+
"eval_samples_per_second": 213.467,
|
| 618 |
+
"eval_steps_per_second": 4.51,
|
| 619 |
+
"step": 105
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 60.57142857142857,
|
| 623 |
+
"eval_accuracy": 0.6197183098591549,
|
| 624 |
+
"eval_loss": 0.6536318063735962,
|
| 625 |
+
"eval_runtime": 0.689,
|
| 626 |
+
"eval_samples_per_second": 206.095,
|
| 627 |
+
"eval_steps_per_second": 4.354,
|
| 628 |
+
"step": 106
|
| 629 |
+
},
|
| 630 |
+
{
|
| 631 |
+
"epoch": 61.714285714285715,
|
| 632 |
+
"eval_accuracy": 0.647887323943662,
|
| 633 |
+
"eval_loss": 0.6177354454994202,
|
| 634 |
+
"eval_runtime": 0.6168,
|
| 635 |
+
"eval_samples_per_second": 230.219,
|
| 636 |
+
"eval_steps_per_second": 4.864,
|
| 637 |
+
"step": 108
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"epoch": 62.857142857142854,
|
| 641 |
+
"grad_norm": 6.197967052459717,
|
| 642 |
+
"learning_rate": 9.75e-06,
|
| 643 |
+
"loss": 0.3483,
|
| 644 |
+
"step": 110
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 62.857142857142854,
|
| 648 |
+
"eval_accuracy": 0.7535211267605634,
|
| 649 |
+
"eval_loss": 0.538495659828186,
|
| 650 |
+
"eval_runtime": 0.6031,
|
| 651 |
+
"eval_samples_per_second": 235.448,
|
| 652 |
+
"eval_steps_per_second": 4.974,
|
| 653 |
+
"step": 110
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 64.0,
|
| 657 |
+
"eval_accuracy": 0.7394366197183099,
|
| 658 |
+
"eval_loss": 0.5630239844322205,
|
| 659 |
+
"eval_runtime": 0.6172,
|
| 660 |
+
"eval_samples_per_second": 230.084,
|
| 661 |
+
"eval_steps_per_second": 4.861,
|
| 662 |
+
"step": 112
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 64.57142857142857,
|
| 666 |
+
"eval_accuracy": 0.7253521126760564,
|
| 667 |
+
"eval_loss": 0.5817703604698181,
|
| 668 |
+
"eval_runtime": 0.6286,
|
| 669 |
+
"eval_samples_per_second": 225.91,
|
| 670 |
+
"eval_steps_per_second": 4.773,
|
| 671 |
+
"step": 113
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 65.71428571428571,
|
| 675 |
+
"eval_accuracy": 0.6971830985915493,
|
| 676 |
+
"eval_loss": 0.605471670627594,
|
| 677 |
+
"eval_runtime": 0.6515,
|
| 678 |
+
"eval_samples_per_second": 217.951,
|
| 679 |
+
"eval_steps_per_second": 4.605,
|
| 680 |
+
"step": 115
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 66.85714285714286,
|
| 684 |
+
"eval_accuracy": 0.7323943661971831,
|
| 685 |
+
"eval_loss": 0.5736685991287231,
|
| 686 |
+
"eval_runtime": 0.6154,
|
| 687 |
+
"eval_samples_per_second": 230.742,
|
| 688 |
+
"eval_steps_per_second": 4.875,
|
| 689 |
+
"step": 117
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 68.0,
|
| 693 |
+
"eval_accuracy": 0.7394366197183099,
|
| 694 |
+
"eval_loss": 0.5605682730674744,
|
| 695 |
+
"eval_runtime": 0.658,
|
| 696 |
+
"eval_samples_per_second": 215.803,
|
| 697 |
+
"eval_steps_per_second": 4.559,
|
| 698 |
+
"step": 119
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 68.57142857142857,
|
| 702 |
+
"grad_norm": 9.030172348022461,
|
| 703 |
+
"learning_rate": 9.5e-06,
|
| 704 |
+
"loss": 0.3667,
|
| 705 |
+
"step": 120
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 68.57142857142857,
|
| 709 |
+
"eval_accuracy": 0.7183098591549296,
|
| 710 |
+
"eval_loss": 0.582869827747345,
|
| 711 |
+
"eval_runtime": 0.614,
|
| 712 |
+
"eval_samples_per_second": 231.27,
|
| 713 |
+
"eval_steps_per_second": 4.886,
|
| 714 |
+
"step": 120
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 69.71428571428571,
|
| 718 |
+
"eval_accuracy": 0.7112676056338029,
|
| 719 |
+
"eval_loss": 0.5931121110916138,
|
| 720 |
+
"eval_runtime": 0.6622,
|
| 721 |
+
"eval_samples_per_second": 214.426,
|
| 722 |
+
"eval_steps_per_second": 4.53,
|
| 723 |
+
"step": 122
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 70.85714285714286,
|
| 727 |
+
"eval_accuracy": 0.7605633802816901,
|
| 728 |
+
"eval_loss": 0.5375192761421204,
|
| 729 |
+
"eval_runtime": 0.6532,
|
| 730 |
+
"eval_samples_per_second": 217.407,
|
| 731 |
+
"eval_steps_per_second": 4.593,
|
| 732 |
+
"step": 124
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 72.0,
|
| 736 |
+
"eval_accuracy": 0.7112676056338029,
|
| 737 |
+
"eval_loss": 0.5797293186187744,
|
| 738 |
+
"eval_runtime": 0.6229,
|
| 739 |
+
"eval_samples_per_second": 227.957,
|
| 740 |
+
"eval_steps_per_second": 4.816,
|
| 741 |
+
"step": 126
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 72.57142857142857,
|
| 745 |
+
"eval_accuracy": 0.6690140845070423,
|
| 746 |
+
"eval_loss": 0.6182157397270203,
|
| 747 |
+
"eval_runtime": 0.6904,
|
| 748 |
+
"eval_samples_per_second": 205.68,
|
| 749 |
+
"eval_steps_per_second": 4.345,
|
| 750 |
+
"step": 127
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 73.71428571428571,
|
| 754 |
+
"eval_accuracy": 0.6690140845070423,
|
| 755 |
+
"eval_loss": 0.6496893167495728,
|
| 756 |
+
"eval_runtime": 0.6934,
|
| 757 |
+
"eval_samples_per_second": 204.791,
|
| 758 |
+
"eval_steps_per_second": 4.327,
|
| 759 |
+
"step": 129
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 74.28571428571429,
|
| 763 |
+
"grad_norm": 6.506986618041992,
|
| 764 |
+
"learning_rate": 9.250000000000001e-06,
|
| 765 |
+
"loss": 0.3357,
|
| 766 |
+
"step": 130
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 74.85714285714286,
|
| 770 |
+
"eval_accuracy": 0.6830985915492958,
|
| 771 |
+
"eval_loss": 0.6431972980499268,
|
| 772 |
+
"eval_runtime": 0.6977,
|
| 773 |
+
"eval_samples_per_second": 203.534,
|
| 774 |
+
"eval_steps_per_second": 4.3,
|
| 775 |
+
"step": 131
|
| 776 |
+
},
|
| 777 |
+
{
|
| 778 |
+
"epoch": 76.0,
|
| 779 |
+
"eval_accuracy": 0.6619718309859155,
|
| 780 |
+
"eval_loss": 0.6771786212921143,
|
| 781 |
+
"eval_runtime": 0.679,
|
| 782 |
+
"eval_samples_per_second": 209.125,
|
| 783 |
+
"eval_steps_per_second": 4.418,
|
| 784 |
+
"step": 133
|
| 785 |
+
},
|
| 786 |
+
{
|
| 787 |
+
"epoch": 76.57142857142857,
|
| 788 |
+
"eval_accuracy": 0.647887323943662,
|
| 789 |
+
"eval_loss": 0.6394755840301514,
|
| 790 |
+
"eval_runtime": 0.6125,
|
| 791 |
+
"eval_samples_per_second": 231.828,
|
| 792 |
+
"eval_steps_per_second": 4.898,
|
| 793 |
+
"step": 134
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 77.71428571428571,
|
| 797 |
+
"eval_accuracy": 0.704225352112676,
|
| 798 |
+
"eval_loss": 0.5894708633422852,
|
| 799 |
+
"eval_runtime": 0.628,
|
| 800 |
+
"eval_samples_per_second": 226.107,
|
| 801 |
+
"eval_steps_per_second": 4.777,
|
| 802 |
+
"step": 136
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 78.85714285714286,
|
| 806 |
+
"eval_accuracy": 0.6971830985915493,
|
| 807 |
+
"eval_loss": 0.592099130153656,
|
| 808 |
+
"eval_runtime": 0.6114,
|
| 809 |
+
"eval_samples_per_second": 232.269,
|
| 810 |
+
"eval_steps_per_second": 4.907,
|
| 811 |
+
"step": 138
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 80.0,
|
| 815 |
+
"grad_norm": 6.510707855224609,
|
| 816 |
+
"learning_rate": 9e-06,
|
| 817 |
+
"loss": 0.3415,
|
| 818 |
+
"step": 140
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 80.0,
|
| 822 |
+
"eval_accuracy": 0.7253521126760564,
|
| 823 |
+
"eval_loss": 0.5618441104888916,
|
| 824 |
+
"eval_runtime": 0.6014,
|
| 825 |
+
"eval_samples_per_second": 236.1,
|
| 826 |
+
"eval_steps_per_second": 4.988,
|
| 827 |
+
"step": 140
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 80.57142857142857,
|
| 831 |
+
"eval_accuracy": 0.7183098591549296,
|
| 832 |
+
"eval_loss": 0.569664478302002,
|
| 833 |
+
"eval_runtime": 0.6899,
|
| 834 |
+
"eval_samples_per_second": 205.839,
|
| 835 |
+
"eval_steps_per_second": 4.349,
|
| 836 |
+
"step": 141
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 81.71428571428571,
|
| 840 |
+
"eval_accuracy": 0.6197183098591549,
|
| 841 |
+
"eval_loss": 0.6535471677780151,
|
| 842 |
+
"eval_runtime": 0.6581,
|
| 843 |
+
"eval_samples_per_second": 215.764,
|
| 844 |
+
"eval_steps_per_second": 4.558,
|
| 845 |
+
"step": 143
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 82.85714285714286,
|
| 849 |
+
"eval_accuracy": 0.6338028169014085,
|
| 850 |
+
"eval_loss": 0.6626799702644348,
|
| 851 |
+
"eval_runtime": 0.6631,
|
| 852 |
+
"eval_samples_per_second": 214.131,
|
| 853 |
+
"eval_steps_per_second": 4.524,
|
| 854 |
+
"step": 145
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 84.0,
|
| 858 |
+
"eval_accuracy": 0.676056338028169,
|
| 859 |
+
"eval_loss": 0.6194331645965576,
|
| 860 |
+
"eval_runtime": 0.6094,
|
| 861 |
+
"eval_samples_per_second": 233.008,
|
| 862 |
+
"eval_steps_per_second": 4.923,
|
| 863 |
+
"step": 147
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 84.57142857142857,
|
| 867 |
+
"eval_accuracy": 0.6901408450704225,
|
| 868 |
+
"eval_loss": 0.6301348805427551,
|
| 869 |
+
"eval_runtime": 0.6101,
|
| 870 |
+
"eval_samples_per_second": 232.737,
|
| 871 |
+
"eval_steps_per_second": 4.917,
|
| 872 |
+
"step": 148
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 85.71428571428571,
|
| 876 |
+
"grad_norm": 4.926376819610596,
|
| 877 |
+
"learning_rate": 8.750000000000001e-06,
|
| 878 |
+
"loss": 0.3296,
|
| 879 |
+
"step": 150
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 85.71428571428571,
|
| 883 |
+
"eval_accuracy": 0.6690140845070423,
|
| 884 |
+
"eval_loss": 0.6436378955841064,
|
| 885 |
+
"eval_runtime": 0.6603,
|
| 886 |
+
"eval_samples_per_second": 215.038,
|
| 887 |
+
"eval_steps_per_second": 4.543,
|
| 888 |
+
"step": 150
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 86.85714285714286,
|
| 892 |
+
"eval_accuracy": 0.6830985915492958,
|
| 893 |
+
"eval_loss": 0.6348147392272949,
|
| 894 |
+
"eval_runtime": 0.6632,
|
| 895 |
+
"eval_samples_per_second": 214.105,
|
| 896 |
+
"eval_steps_per_second": 4.523,
|
| 897 |
+
"step": 152
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 88.0,
|
| 901 |
+
"eval_accuracy": 0.647887323943662,
|
| 902 |
+
"eval_loss": 0.6703845262527466,
|
| 903 |
+
"eval_runtime": 0.6192,
|
| 904 |
+
"eval_samples_per_second": 229.331,
|
| 905 |
+
"eval_steps_per_second": 4.845,
|
| 906 |
+
"step": 154
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 88.57142857142857,
|
| 910 |
+
"eval_accuracy": 0.6338028169014085,
|
| 911 |
+
"eval_loss": 0.7190238237380981,
|
| 912 |
+
"eval_runtime": 0.6914,
|
| 913 |
+
"eval_samples_per_second": 205.385,
|
| 914 |
+
"eval_steps_per_second": 4.339,
|
| 915 |
+
"step": 155
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 89.71428571428571,
|
| 919 |
+
"eval_accuracy": 0.6338028169014085,
|
| 920 |
+
"eval_loss": 0.7063698768615723,
|
| 921 |
+
"eval_runtime": 0.655,
|
| 922 |
+
"eval_samples_per_second": 216.794,
|
| 923 |
+
"eval_steps_per_second": 4.58,
|
| 924 |
+
"step": 157
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 90.85714285714286,
|
| 928 |
+
"eval_accuracy": 0.6549295774647887,
|
| 929 |
+
"eval_loss": 0.6291490793228149,
|
| 930 |
+
"eval_runtime": 0.6988,
|
| 931 |
+
"eval_samples_per_second": 203.195,
|
| 932 |
+
"eval_steps_per_second": 4.293,
|
| 933 |
+
"step": 159
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 91.42857142857143,
|
| 937 |
+
"grad_norm": 7.158923625946045,
|
| 938 |
+
"learning_rate": 8.5e-06,
|
| 939 |
+
"loss": 0.3296,
|
| 940 |
+
"step": 160
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 92.0,
|
| 944 |
+
"eval_accuracy": 0.6197183098591549,
|
| 945 |
+
"eval_loss": 0.6933465003967285,
|
| 946 |
+
"eval_runtime": 0.7003,
|
| 947 |
+
"eval_samples_per_second": 202.767,
|
| 948 |
+
"eval_steps_per_second": 4.284,
|
| 949 |
+
"step": 161
|
| 950 |
+
},
|
| 951 |
+
{
|
| 952 |
+
"epoch": 92.57142857142857,
|
| 953 |
+
"eval_accuracy": 0.6197183098591549,
|
| 954 |
+
"eval_loss": 0.7115197777748108,
|
| 955 |
+
"eval_runtime": 0.6713,
|
| 956 |
+
"eval_samples_per_second": 211.542,
|
| 957 |
+
"eval_steps_per_second": 4.469,
|
| 958 |
+
"step": 162
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"epoch": 93.71428571428571,
|
| 962 |
+
"eval_accuracy": 0.6690140845070423,
|
| 963 |
+
"eval_loss": 0.6228932738304138,
|
| 964 |
+
"eval_runtime": 0.6174,
|
| 965 |
+
"eval_samples_per_second": 230.002,
|
| 966 |
+
"eval_steps_per_second": 4.859,
|
| 967 |
+
"step": 164
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 94.85714285714286,
|
| 971 |
+
"eval_accuracy": 0.7183098591549296,
|
| 972 |
+
"eval_loss": 0.5726597309112549,
|
| 973 |
+
"eval_runtime": 0.6102,
|
| 974 |
+
"eval_samples_per_second": 232.693,
|
| 975 |
+
"eval_steps_per_second": 4.916,
|
| 976 |
+
"step": 166
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 96.0,
|
| 980 |
+
"eval_accuracy": 0.6901408450704225,
|
| 981 |
+
"eval_loss": 0.5964914560317993,
|
| 982 |
+
"eval_runtime": 0.6146,
|
| 983 |
+
"eval_samples_per_second": 231.039,
|
| 984 |
+
"eval_steps_per_second": 4.881,
|
| 985 |
+
"step": 168
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 96.57142857142857,
|
| 989 |
+
"eval_accuracy": 0.6690140845070423,
|
| 990 |
+
"eval_loss": 0.643250584602356,
|
| 991 |
+
"eval_runtime": 0.6226,
|
| 992 |
+
"eval_samples_per_second": 228.094,
|
| 993 |
+
"eval_steps_per_second": 4.819,
|
| 994 |
+
"step": 169
|
| 995 |
+
},
|
| 996 |
+
{
|
| 997 |
+
"epoch": 97.14285714285714,
|
| 998 |
+
"grad_norm": 10.834047317504883,
|
| 999 |
+
"learning_rate": 8.25e-06,
|
| 1000 |
+
"loss": 0.3174,
|
| 1001 |
+
"step": 170
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 97.71428571428571,
|
| 1005 |
+
"eval_accuracy": 0.6408450704225352,
|
| 1006 |
+
"eval_loss": 0.6633855104446411,
|
| 1007 |
+
"eval_runtime": 0.6496,
|
| 1008 |
+
"eval_samples_per_second": 218.595,
|
| 1009 |
+
"eval_steps_per_second": 4.618,
|
| 1010 |
+
"step": 171
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 98.85714285714286,
|
| 1014 |
+
"eval_accuracy": 0.6549295774647887,
|
| 1015 |
+
"eval_loss": 0.6165616512298584,
|
| 1016 |
+
"eval_runtime": 0.6355,
|
| 1017 |
+
"eval_samples_per_second": 223.431,
|
| 1018 |
+
"eval_steps_per_second": 4.72,
|
| 1019 |
+
"step": 173
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 100.0,
|
| 1023 |
+
"eval_accuracy": 0.6971830985915493,
|
| 1024 |
+
"eval_loss": 0.5896428227424622,
|
| 1025 |
+
"eval_runtime": 0.6288,
|
| 1026 |
+
"eval_samples_per_second": 225.845,
|
| 1027 |
+
"eval_steps_per_second": 4.771,
|
| 1028 |
+
"step": 175
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 100.57142857142857,
|
| 1032 |
+
"eval_accuracy": 0.6549295774647887,
|
| 1033 |
+
"eval_loss": 0.6092301607131958,
|
| 1034 |
+
"eval_runtime": 0.6874,
|
| 1035 |
+
"eval_samples_per_second": 206.582,
|
| 1036 |
+
"eval_steps_per_second": 4.364,
|
| 1037 |
+
"step": 176
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 101.71428571428571,
|
| 1041 |
+
"eval_accuracy": 0.6549295774647887,
|
| 1042 |
+
"eval_loss": 0.6022228002548218,
|
| 1043 |
+
"eval_runtime": 0.6122,
|
| 1044 |
+
"eval_samples_per_second": 231.962,
|
| 1045 |
+
"eval_steps_per_second": 4.901,
|
| 1046 |
+
"step": 178
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 102.85714285714286,
|
| 1050 |
+
"grad_norm": 6.7690935134887695,
|
| 1051 |
+
"learning_rate": 8.000000000000001e-06,
|
| 1052 |
+
"loss": 0.3309,
|
| 1053 |
+
"step": 180
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 102.85714285714286,
|
| 1057 |
+
"eval_accuracy": 0.676056338028169,
|
| 1058 |
+
"eval_loss": 0.5928385853767395,
|
| 1059 |
+
"eval_runtime": 0.629,
|
| 1060 |
+
"eval_samples_per_second": 225.738,
|
| 1061 |
+
"eval_steps_per_second": 4.769,
|
| 1062 |
+
"step": 180
|
| 1063 |
+
},
|
| 1064 |
+
{
|
| 1065 |
+
"epoch": 104.0,
|
| 1066 |
+
"eval_accuracy": 0.6408450704225352,
|
| 1067 |
+
"eval_loss": 0.6327298879623413,
|
| 1068 |
+
"eval_runtime": 0.6082,
|
| 1069 |
+
"eval_samples_per_second": 233.473,
|
| 1070 |
+
"eval_steps_per_second": 4.933,
|
| 1071 |
+
"step": 182
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 104.57142857142857,
|
| 1075 |
+
"eval_accuracy": 0.6338028169014085,
|
| 1076 |
+
"eval_loss": 0.6489691138267517,
|
| 1077 |
+
"eval_runtime": 0.6079,
|
| 1078 |
+
"eval_samples_per_second": 233.573,
|
| 1079 |
+
"eval_steps_per_second": 4.935,
|
| 1080 |
+
"step": 183
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 105.71428571428571,
|
| 1084 |
+
"eval_accuracy": 0.647887323943662,
|
| 1085 |
+
"eval_loss": 0.6155086159706116,
|
| 1086 |
+
"eval_runtime": 0.6162,
|
| 1087 |
+
"eval_samples_per_second": 230.459,
|
| 1088 |
+
"eval_steps_per_second": 4.869,
|
| 1089 |
+
"step": 185
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 106.85714285714286,
|
| 1093 |
+
"eval_accuracy": 0.6619718309859155,
|
| 1094 |
+
"eval_loss": 0.6225135922431946,
|
| 1095 |
+
"eval_runtime": 0.6071,
|
| 1096 |
+
"eval_samples_per_second": 233.885,
|
| 1097 |
+
"eval_steps_per_second": 4.941,
|
| 1098 |
+
"step": 187
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 108.0,
|
| 1102 |
+
"eval_accuracy": 0.6408450704225352,
|
| 1103 |
+
"eval_loss": 0.6732430458068848,
|
| 1104 |
+
"eval_runtime": 0.6148,
|
| 1105 |
+
"eval_samples_per_second": 230.978,
|
| 1106 |
+
"eval_steps_per_second": 4.88,
|
| 1107 |
+
"step": 189
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 108.57142857142857,
|
| 1111 |
+
"grad_norm": 6.764002799987793,
|
| 1112 |
+
"learning_rate": 7.75e-06,
|
| 1113 |
+
"loss": 0.3124,
|
| 1114 |
+
"step": 190
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 108.57142857142857,
|
| 1118 |
+
"eval_accuracy": 0.6408450704225352,
|
| 1119 |
+
"eval_loss": 0.6808416247367859,
|
| 1120 |
+
"eval_runtime": 0.6534,
|
| 1121 |
+
"eval_samples_per_second": 217.318,
|
| 1122 |
+
"eval_steps_per_second": 4.591,
|
| 1123 |
+
"step": 190
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 109.71428571428571,
|
| 1127 |
+
"eval_accuracy": 0.647887323943662,
|
| 1128 |
+
"eval_loss": 0.6585180759429932,
|
| 1129 |
+
"eval_runtime": 0.6127,
|
| 1130 |
+
"eval_samples_per_second": 231.773,
|
| 1131 |
+
"eval_steps_per_second": 4.897,
|
| 1132 |
+
"step": 192
|
| 1133 |
+
},
|
| 1134 |
+
{
|
| 1135 |
+
"epoch": 110.85714285714286,
|
| 1136 |
+
"eval_accuracy": 0.676056338028169,
|
| 1137 |
+
"eval_loss": 0.6121569275856018,
|
| 1138 |
+
"eval_runtime": 0.63,
|
| 1139 |
+
"eval_samples_per_second": 225.399,
|
| 1140 |
+
"eval_steps_per_second": 4.762,
|
| 1141 |
+
"step": 194
|
| 1142 |
+
},
|
| 1143 |
+
{
|
| 1144 |
+
"epoch": 112.0,
|
| 1145 |
+
"eval_accuracy": 0.6549295774647887,
|
| 1146 |
+
"eval_loss": 0.6510112285614014,
|
| 1147 |
+
"eval_runtime": 0.6484,
|
| 1148 |
+
"eval_samples_per_second": 218.987,
|
| 1149 |
+
"eval_steps_per_second": 4.626,
|
| 1150 |
+
"step": 196
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 112.57142857142857,
|
| 1154 |
+
"eval_accuracy": 0.6408450704225352,
|
| 1155 |
+
"eval_loss": 0.7098967432975769,
|
| 1156 |
+
"eval_runtime": 0.6667,
|
| 1157 |
+
"eval_samples_per_second": 212.977,
|
| 1158 |
+
"eval_steps_per_second": 4.5,
|
| 1159 |
+
"step": 197
|
| 1160 |
+
},
|
| 1161 |
+
{
|
| 1162 |
+
"epoch": 113.71428571428571,
|
| 1163 |
+
"eval_accuracy": 0.6338028169014085,
|
| 1164 |
+
"eval_loss": 0.7191683053970337,
|
| 1165 |
+
"eval_runtime": 0.6421,
|
| 1166 |
+
"eval_samples_per_second": 221.158,
|
| 1167 |
+
"eval_steps_per_second": 4.672,
|
| 1168 |
+
"step": 199
|
| 1169 |
+
},
|
| 1170 |
+
{
|
| 1171 |
+
"epoch": 114.28571428571429,
|
| 1172 |
+
"grad_norm": 7.113361358642578,
|
| 1173 |
+
"learning_rate": 7.500000000000001e-06,
|
| 1174 |
+
"loss": 0.3158,
|
| 1175 |
+
"step": 200
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"epoch": 114.85714285714286,
|
| 1179 |
+
"eval_accuracy": 0.6901408450704225,
|
| 1180 |
+
"eval_loss": 0.6186498403549194,
|
| 1181 |
+
"eval_runtime": 0.6046,
|
| 1182 |
+
"eval_samples_per_second": 234.87,
|
| 1183 |
+
"eval_steps_per_second": 4.962,
|
| 1184 |
+
"step": 201
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 116.0,
|
| 1188 |
+
"eval_accuracy": 0.704225352112676,
|
| 1189 |
+
"eval_loss": 0.6070767045021057,
|
| 1190 |
+
"eval_runtime": 0.6492,
|
| 1191 |
+
"eval_samples_per_second": 218.723,
|
| 1192 |
+
"eval_steps_per_second": 4.621,
|
| 1193 |
+
"step": 203
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 116.57142857142857,
|
| 1197 |
+
"eval_accuracy": 0.6830985915492958,
|
| 1198 |
+
"eval_loss": 0.6419255137443542,
|
| 1199 |
+
"eval_runtime": 0.7166,
|
| 1200 |
+
"eval_samples_per_second": 198.169,
|
| 1201 |
+
"eval_steps_per_second": 4.187,
|
| 1202 |
+
"step": 204
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 117.71428571428571,
|
| 1206 |
+
"eval_accuracy": 0.6549295774647887,
|
| 1207 |
+
"eval_loss": 0.6679487824440002,
|
| 1208 |
+
"eval_runtime": 0.6953,
|
| 1209 |
+
"eval_samples_per_second": 204.235,
|
| 1210 |
+
"eval_steps_per_second": 4.315,
|
| 1211 |
+
"step": 206
|
| 1212 |
+
},
|
| 1213 |
+
{
|
| 1214 |
+
"epoch": 118.85714285714286,
|
| 1215 |
+
"eval_accuracy": 0.6267605633802817,
|
| 1216 |
+
"eval_loss": 0.6824603080749512,
|
| 1217 |
+
"eval_runtime": 0.6086,
|
| 1218 |
+
"eval_samples_per_second": 233.311,
|
| 1219 |
+
"eval_steps_per_second": 4.929,
|
| 1220 |
+
"step": 208
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 120.0,
|
| 1224 |
+
"grad_norm": 7.5840373039245605,
|
| 1225 |
+
"learning_rate": 7.25e-06,
|
| 1226 |
+
"loss": 0.3026,
|
| 1227 |
+
"step": 210
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 120.0,
|
| 1231 |
+
"eval_accuracy": 0.6971830985915493,
|
| 1232 |
+
"eval_loss": 0.6090666055679321,
|
| 1233 |
+
"eval_runtime": 0.6137,
|
| 1234 |
+
"eval_samples_per_second": 231.379,
|
| 1235 |
+
"eval_steps_per_second": 4.888,
|
| 1236 |
+
"step": 210
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"epoch": 120.57142857142857,
|
| 1240 |
+
"eval_accuracy": 0.7394366197183099,
|
| 1241 |
+
"eval_loss": 0.5861057043075562,
|
| 1242 |
+
"eval_runtime": 0.7023,
|
| 1243 |
+
"eval_samples_per_second": 202.195,
|
| 1244 |
+
"eval_steps_per_second": 4.272,
|
| 1245 |
+
"step": 211
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 121.71428571428571,
|
| 1249 |
+
"eval_accuracy": 0.7112676056338029,
|
| 1250 |
+
"eval_loss": 0.6036643385887146,
|
| 1251 |
+
"eval_runtime": 0.6393,
|
| 1252 |
+
"eval_samples_per_second": 222.102,
|
| 1253 |
+
"eval_steps_per_second": 4.692,
|
| 1254 |
+
"step": 213
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 122.85714285714286,
|
| 1258 |
+
"eval_accuracy": 0.676056338028169,
|
| 1259 |
+
"eval_loss": 0.6314629912376404,
|
| 1260 |
+
"eval_runtime": 0.6325,
|
| 1261 |
+
"eval_samples_per_second": 224.491,
|
| 1262 |
+
"eval_steps_per_second": 4.743,
|
| 1263 |
+
"step": 215
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 124.0,
|
| 1267 |
+
"eval_accuracy": 0.6690140845070423,
|
| 1268 |
+
"eval_loss": 0.6328482031822205,
|
| 1269 |
+
"eval_runtime": 0.6623,
|
| 1270 |
+
"eval_samples_per_second": 214.389,
|
| 1271 |
+
"eval_steps_per_second": 4.529,
|
| 1272 |
+
"step": 217
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"epoch": 124.57142857142857,
|
| 1276 |
+
"eval_accuracy": 0.6830985915492958,
|
| 1277 |
+
"eval_loss": 0.6187437772750854,
|
| 1278 |
+
"eval_runtime": 0.608,
|
| 1279 |
+
"eval_samples_per_second": 233.549,
|
| 1280 |
+
"eval_steps_per_second": 4.934,
|
| 1281 |
+
"step": 218
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 125.71428571428571,
|
| 1285 |
+
"grad_norm": 5.965949535369873,
|
| 1286 |
+
"learning_rate": 7e-06,
|
| 1287 |
+
"loss": 0.2968,
|
| 1288 |
+
"step": 220
|
| 1289 |
+
},
|
| 1290 |
+
{
|
| 1291 |
+
"epoch": 125.71428571428571,
|
| 1292 |
+
"eval_accuracy": 0.7394366197183099,
|
| 1293 |
+
"eval_loss": 0.5843108296394348,
|
| 1294 |
+
"eval_runtime": 0.6786,
|
| 1295 |
+
"eval_samples_per_second": 209.252,
|
| 1296 |
+
"eval_steps_per_second": 4.421,
|
| 1297 |
+
"step": 220
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 126.85714285714286,
|
| 1301 |
+
"eval_accuracy": 0.704225352112676,
|
| 1302 |
+
"eval_loss": 0.6126313805580139,
|
| 1303 |
+
"eval_runtime": 0.7169,
|
| 1304 |
+
"eval_samples_per_second": 198.068,
|
| 1305 |
+
"eval_steps_per_second": 4.185,
|
| 1306 |
+
"step": 222
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 128.0,
|
| 1310 |
+
"eval_accuracy": 0.6549295774647887,
|
| 1311 |
+
"eval_loss": 0.6784695386886597,
|
| 1312 |
+
"eval_runtime": 0.6207,
|
| 1313 |
+
"eval_samples_per_second": 228.762,
|
| 1314 |
+
"eval_steps_per_second": 4.833,
|
| 1315 |
+
"step": 224
|
| 1316 |
+
},
|
| 1317 |
+
{
|
| 1318 |
+
"epoch": 128.57142857142858,
|
| 1319 |
+
"eval_accuracy": 0.647887323943662,
|
| 1320 |
+
"eval_loss": 0.6705502867698669,
|
| 1321 |
+
"eval_runtime": 0.6147,
|
| 1322 |
+
"eval_samples_per_second": 231.0,
|
| 1323 |
+
"eval_steps_per_second": 4.88,
|
| 1324 |
+
"step": 225
|
| 1325 |
+
},
|
| 1326 |
+
{
|
| 1327 |
+
"epoch": 129.71428571428572,
|
| 1328 |
+
"eval_accuracy": 0.7112676056338029,
|
| 1329 |
+
"eval_loss": 0.606967031955719,
|
| 1330 |
+
"eval_runtime": 0.6267,
|
| 1331 |
+
"eval_samples_per_second": 226.602,
|
| 1332 |
+
"eval_steps_per_second": 4.787,
|
| 1333 |
+
"step": 227
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 130.85714285714286,
|
| 1337 |
+
"eval_accuracy": 0.7253521126760564,
|
| 1338 |
+
"eval_loss": 0.5984058380126953,
|
| 1339 |
+
"eval_runtime": 0.6228,
|
| 1340 |
+
"eval_samples_per_second": 228.01,
|
| 1341 |
+
"eval_steps_per_second": 4.817,
|
| 1342 |
+
"step": 229
|
| 1343 |
+
},
|
| 1344 |
+
{
|
| 1345 |
+
"epoch": 131.42857142857142,
|
| 1346 |
+
"grad_norm": 12.718730926513672,
|
| 1347 |
+
"learning_rate": 6.750000000000001e-06,
|
| 1348 |
+
"loss": 0.294,
|
| 1349 |
+
"step": 230
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 132.0,
|
| 1353 |
+
"eval_accuracy": 0.6619718309859155,
|
| 1354 |
+
"eval_loss": 0.6533382534980774,
|
| 1355 |
+
"eval_runtime": 0.6815,
|
| 1356 |
+
"eval_samples_per_second": 208.355,
|
| 1357 |
+
"eval_steps_per_second": 4.402,
|
| 1358 |
+
"step": 231
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 132.57142857142858,
|
| 1362 |
+
"eval_accuracy": 0.6408450704225352,
|
| 1363 |
+
"eval_loss": 0.6801695227622986,
|
| 1364 |
+
"eval_runtime": 0.6584,
|
| 1365 |
+
"eval_samples_per_second": 215.686,
|
| 1366 |
+
"eval_steps_per_second": 4.557,
|
| 1367 |
+
"step": 232
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 133.71428571428572,
|
| 1371 |
+
"eval_accuracy": 0.6408450704225352,
|
| 1372 |
+
"eval_loss": 0.6803719401359558,
|
| 1373 |
+
"eval_runtime": 0.649,
|
| 1374 |
+
"eval_samples_per_second": 218.787,
|
| 1375 |
+
"eval_steps_per_second": 4.622,
|
| 1376 |
+
"step": 234
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 134.85714285714286,
|
| 1380 |
+
"eval_accuracy": 0.704225352112676,
|
| 1381 |
+
"eval_loss": 0.6228448152542114,
|
| 1382 |
+
"eval_runtime": 0.6502,
|
| 1383 |
+
"eval_samples_per_second": 218.39,
|
| 1384 |
+
"eval_steps_per_second": 4.614,
|
| 1385 |
+
"step": 236
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 136.0,
|
| 1389 |
+
"eval_accuracy": 0.7676056338028169,
|
| 1390 |
+
"eval_loss": 0.5848736763000488,
|
| 1391 |
+
"eval_runtime": 0.6904,
|
| 1392 |
+
"eval_samples_per_second": 205.693,
|
| 1393 |
+
"eval_steps_per_second": 4.346,
|
| 1394 |
+
"step": 238
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 136.57142857142858,
|
| 1398 |
+
"eval_accuracy": 0.7676056338028169,
|
| 1399 |
+
"eval_loss": 0.5873926281929016,
|
| 1400 |
+
"eval_runtime": 0.703,
|
| 1401 |
+
"eval_samples_per_second": 201.984,
|
| 1402 |
+
"eval_steps_per_second": 4.267,
|
| 1403 |
+
"step": 239
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 137.14285714285714,
|
| 1407 |
+
"grad_norm": 9.88498306274414,
|
| 1408 |
+
"learning_rate": 6.5000000000000004e-06,
|
| 1409 |
+
"loss": 0.3009,
|
| 1410 |
+
"step": 240
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 137.71428571428572,
|
| 1414 |
+
"eval_accuracy": 0.704225352112676,
|
| 1415 |
+
"eval_loss": 0.6229640245437622,
|
| 1416 |
+
"eval_runtime": 0.6216,
|
| 1417 |
+
"eval_samples_per_second": 228.454,
|
| 1418 |
+
"eval_steps_per_second": 4.826,
|
| 1419 |
+
"step": 241
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 138.85714285714286,
|
| 1423 |
+
"eval_accuracy": 0.6549295774647887,
|
| 1424 |
+
"eval_loss": 0.6641034483909607,
|
| 1425 |
+
"eval_runtime": 0.6238,
|
| 1426 |
+
"eval_samples_per_second": 227.625,
|
| 1427 |
+
"eval_steps_per_second": 4.809,
|
| 1428 |
+
"step": 243
|
| 1429 |
+
},
|
| 1430 |
+
{
|
| 1431 |
+
"epoch": 140.0,
|
| 1432 |
+
"eval_accuracy": 0.6971830985915493,
|
| 1433 |
+
"eval_loss": 0.6435027122497559,
|
| 1434 |
+
"eval_runtime": 0.6995,
|
| 1435 |
+
"eval_samples_per_second": 202.995,
|
| 1436 |
+
"eval_steps_per_second": 4.289,
|
| 1437 |
+
"step": 245
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 140.57142857142858,
|
| 1441 |
+
"eval_accuracy": 0.7253521126760564,
|
| 1442 |
+
"eval_loss": 0.6133910417556763,
|
| 1443 |
+
"eval_runtime": 0.6661,
|
| 1444 |
+
"eval_samples_per_second": 213.169,
|
| 1445 |
+
"eval_steps_per_second": 4.504,
|
| 1446 |
+
"step": 246
|
| 1447 |
+
},
|
| 1448 |
+
{
|
| 1449 |
+
"epoch": 141.71428571428572,
|
| 1450 |
+
"eval_accuracy": 0.7394366197183099,
|
| 1451 |
+
"eval_loss": 0.6062853932380676,
|
| 1452 |
+
"eval_runtime": 0.6955,
|
| 1453 |
+
"eval_samples_per_second": 204.179,
|
| 1454 |
+
"eval_steps_per_second": 4.314,
|
| 1455 |
+
"step": 248
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 142.85714285714286,
|
| 1459 |
+
"grad_norm": 4.536113262176514,
|
| 1460 |
+
"learning_rate": 6.25e-06,
|
| 1461 |
+
"loss": 0.2873,
|
| 1462 |
+
"step": 250
|
| 1463 |
+
},
|
| 1464 |
+
{
|
| 1465 |
+
"epoch": 142.85714285714286,
|
| 1466 |
+
"eval_accuracy": 0.6971830985915493,
|
| 1467 |
+
"eval_loss": 0.6346603035926819,
|
| 1468 |
+
"eval_runtime": 0.6744,
|
| 1469 |
+
"eval_samples_per_second": 210.569,
|
| 1470 |
+
"eval_steps_per_second": 4.449,
|
| 1471 |
+
"step": 250
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 144.0,
|
| 1475 |
+
"eval_accuracy": 0.6690140845070423,
|
| 1476 |
+
"eval_loss": 0.6992180347442627,
|
| 1477 |
+
"eval_runtime": 0.6915,
|
| 1478 |
+
"eval_samples_per_second": 205.341,
|
| 1479 |
+
"eval_steps_per_second": 4.338,
|
| 1480 |
+
"step": 252
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 144.57142857142858,
|
| 1484 |
+
"eval_accuracy": 0.6408450704225352,
|
| 1485 |
+
"eval_loss": 0.7137159705162048,
|
| 1486 |
+
"eval_runtime": 0.6826,
|
| 1487 |
+
"eval_samples_per_second": 208.024,
|
| 1488 |
+
"eval_steps_per_second": 4.395,
|
| 1489 |
+
"step": 253
|
| 1490 |
+
},
|
| 1491 |
+
{
|
| 1492 |
+
"epoch": 145.71428571428572,
|
| 1493 |
+
"eval_accuracy": 0.6690140845070423,
|
| 1494 |
+
"eval_loss": 0.6737670302391052,
|
| 1495 |
+
"eval_runtime": 0.6116,
|
| 1496 |
+
"eval_samples_per_second": 232.18,
|
| 1497 |
+
"eval_steps_per_second": 4.905,
|
| 1498 |
+
"step": 255
|
| 1499 |
+
},
|
| 1500 |
+
{
|
| 1501 |
+
"epoch": 146.85714285714286,
|
| 1502 |
+
"eval_accuracy": 0.7112676056338029,
|
| 1503 |
+
"eval_loss": 0.6321189999580383,
|
| 1504 |
+
"eval_runtime": 0.6539,
|
| 1505 |
+
"eval_samples_per_second": 217.171,
|
| 1506 |
+
"eval_steps_per_second": 4.588,
|
| 1507 |
+
"step": 257
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 148.0,
|
| 1511 |
+
"eval_accuracy": 0.7183098591549296,
|
| 1512 |
+
"eval_loss": 0.6134991645812988,
|
| 1513 |
+
"eval_runtime": 0.698,
|
| 1514 |
+
"eval_samples_per_second": 203.44,
|
| 1515 |
+
"eval_steps_per_second": 4.298,
|
| 1516 |
+
"step": 259
|
| 1517 |
+
},
|
| 1518 |
+
{
|
| 1519 |
+
"epoch": 148.57142857142858,
|
| 1520 |
+
"grad_norm": 7.451202869415283,
|
| 1521 |
+
"learning_rate": 6e-06,
|
| 1522 |
+
"loss": 0.2821,
|
| 1523 |
+
"step": 260
|
| 1524 |
+
},
|
| 1525 |
+
{
|
| 1526 |
+
"epoch": 148.57142857142858,
|
| 1527 |
+
"eval_accuracy": 0.7112676056338029,
|
| 1528 |
+
"eval_loss": 0.6195182204246521,
|
| 1529 |
+
"eval_runtime": 0.6206,
|
| 1530 |
+
"eval_samples_per_second": 228.8,
|
| 1531 |
+
"eval_steps_per_second": 4.834,
|
| 1532 |
+
"step": 260
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 149.71428571428572,
|
| 1536 |
+
"eval_accuracy": 0.676056338028169,
|
| 1537 |
+
"eval_loss": 0.6543940305709839,
|
| 1538 |
+
"eval_runtime": 0.6997,
|
| 1539 |
+
"eval_samples_per_second": 202.943,
|
| 1540 |
+
"eval_steps_per_second": 4.288,
|
| 1541 |
+
"step": 262
|
| 1542 |
+
},
|
| 1543 |
+
{
|
| 1544 |
+
"epoch": 150.85714285714286,
|
| 1545 |
+
"eval_accuracy": 0.6830985915492958,
|
| 1546 |
+
"eval_loss": 0.6463538408279419,
|
| 1547 |
+
"eval_runtime": 0.6738,
|
| 1548 |
+
"eval_samples_per_second": 210.752,
|
| 1549 |
+
"eval_steps_per_second": 4.453,
|
| 1550 |
+
"step": 264
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 152.0,
|
| 1554 |
+
"eval_accuracy": 0.7323943661971831,
|
| 1555 |
+
"eval_loss": 0.6086514592170715,
|
| 1556 |
+
"eval_runtime": 0.6612,
|
| 1557 |
+
"eval_samples_per_second": 214.749,
|
| 1558 |
+
"eval_steps_per_second": 4.537,
|
| 1559 |
+
"step": 266
|
| 1560 |
+
},
|
| 1561 |
+
{
|
| 1562 |
+
"epoch": 152.57142857142858,
|
| 1563 |
+
"eval_accuracy": 0.7394366197183099,
|
| 1564 |
+
"eval_loss": 0.5999845266342163,
|
| 1565 |
+
"eval_runtime": 0.6497,
|
| 1566 |
+
"eval_samples_per_second": 218.565,
|
| 1567 |
+
"eval_steps_per_second": 4.618,
|
| 1568 |
+
"step": 267
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 153.71428571428572,
|
| 1572 |
+
"eval_accuracy": 0.7112676056338029,
|
| 1573 |
+
"eval_loss": 0.6170024871826172,
|
| 1574 |
+
"eval_runtime": 0.6064,
|
| 1575 |
+
"eval_samples_per_second": 234.184,
|
| 1576 |
+
"eval_steps_per_second": 4.948,
|
| 1577 |
+
"step": 269
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 154.28571428571428,
|
| 1581 |
+
"grad_norm": 15.44714069366455,
|
| 1582 |
+
"learning_rate": 5.75e-06,
|
| 1583 |
+
"loss": 0.3017,
|
| 1584 |
+
"step": 270
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 154.85714285714286,
|
| 1588 |
+
"eval_accuracy": 0.6830985915492958,
|
| 1589 |
+
"eval_loss": 0.6674468517303467,
|
| 1590 |
+
"eval_runtime": 0.6065,
|
| 1591 |
+
"eval_samples_per_second": 234.118,
|
| 1592 |
+
"eval_steps_per_second": 4.946,
|
| 1593 |
+
"step": 271
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 156.0,
|
| 1597 |
+
"eval_accuracy": 0.6338028169014085,
|
| 1598 |
+
"eval_loss": 0.7137435078620911,
|
| 1599 |
+
"eval_runtime": 0.6178,
|
| 1600 |
+
"eval_samples_per_second": 229.849,
|
| 1601 |
+
"eval_steps_per_second": 4.856,
|
| 1602 |
+
"step": 273
|
| 1603 |
+
},
|
| 1604 |
+
{
|
| 1605 |
+
"epoch": 156.57142857142858,
|
| 1606 |
+
"eval_accuracy": 0.647887323943662,
|
| 1607 |
+
"eval_loss": 0.7013713717460632,
|
| 1608 |
+
"eval_runtime": 0.6271,
|
| 1609 |
+
"eval_samples_per_second": 226.442,
|
| 1610 |
+
"eval_steps_per_second": 4.784,
|
| 1611 |
+
"step": 274
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 157.71428571428572,
|
| 1615 |
+
"eval_accuracy": 0.7253521126760564,
|
| 1616 |
+
"eval_loss": 0.609072744846344,
|
| 1617 |
+
"eval_runtime": 0.6285,
|
| 1618 |
+
"eval_samples_per_second": 225.936,
|
| 1619 |
+
"eval_steps_per_second": 4.773,
|
| 1620 |
+
"step": 276
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 158.85714285714286,
|
| 1624 |
+
"eval_accuracy": 0.7676056338028169,
|
| 1625 |
+
"eval_loss": 0.5625793933868408,
|
| 1626 |
+
"eval_runtime": 0.6308,
|
| 1627 |
+
"eval_samples_per_second": 225.116,
|
| 1628 |
+
"eval_steps_per_second": 4.756,
|
| 1629 |
+
"step": 278
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 160.0,
|
| 1633 |
+
"grad_norm": 7.80256462097168,
|
| 1634 |
+
"learning_rate": 5.500000000000001e-06,
|
| 1635 |
+
"loss": 0.2857,
|
| 1636 |
+
"step": 280
|
| 1637 |
+
},
|
| 1638 |
+
{
|
| 1639 |
+
"epoch": 160.0,
|
| 1640 |
+
"eval_accuracy": 0.7605633802816901,
|
| 1641 |
+
"eval_loss": 0.5685088634490967,
|
| 1642 |
+
"eval_runtime": 0.6091,
|
| 1643 |
+
"eval_samples_per_second": 233.142,
|
| 1644 |
+
"eval_steps_per_second": 4.926,
|
| 1645 |
+
"step": 280
|
| 1646 |
+
},
|
| 1647 |
+
{
|
| 1648 |
+
"epoch": 160.57142857142858,
|
| 1649 |
+
"eval_accuracy": 0.7112676056338029,
|
| 1650 |
+
"eval_loss": 0.5941457748413086,
|
| 1651 |
+
"eval_runtime": 0.6891,
|
| 1652 |
+
"eval_samples_per_second": 206.055,
|
| 1653 |
+
"eval_steps_per_second": 4.353,
|
| 1654 |
+
"step": 281
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"epoch": 161.71428571428572,
|
| 1658 |
+
"eval_accuracy": 0.7112676056338029,
|
| 1659 |
+
"eval_loss": 0.6219268441200256,
|
| 1660 |
+
"eval_runtime": 0.7133,
|
| 1661 |
+
"eval_samples_per_second": 199.064,
|
| 1662 |
+
"eval_steps_per_second": 4.206,
|
| 1663 |
+
"step": 283
|
| 1664 |
+
},
|
| 1665 |
+
{
|
| 1666 |
+
"epoch": 162.85714285714286,
|
| 1667 |
+
"eval_accuracy": 0.7112676056338029,
|
| 1668 |
+
"eval_loss": 0.6282822489738464,
|
| 1669 |
+
"eval_runtime": 0.6296,
|
| 1670 |
+
"eval_samples_per_second": 225.525,
|
| 1671 |
+
"eval_steps_per_second": 4.765,
|
| 1672 |
+
"step": 285
|
| 1673 |
+
},
|
| 1674 |
+
{
|
| 1675 |
+
"epoch": 164.0,
|
| 1676 |
+
"eval_accuracy": 0.704225352112676,
|
| 1677 |
+
"eval_loss": 0.631372332572937,
|
| 1678 |
+
"eval_runtime": 0.6725,
|
| 1679 |
+
"eval_samples_per_second": 211.158,
|
| 1680 |
+
"eval_steps_per_second": 4.461,
|
| 1681 |
+
"step": 287
|
| 1682 |
+
},
|
| 1683 |
+
{
|
| 1684 |
+
"epoch": 164.57142857142858,
|
| 1685 |
+
"eval_accuracy": 0.6971830985915493,
|
| 1686 |
+
"eval_loss": 0.6368916630744934,
|
| 1687 |
+
"eval_runtime": 0.6185,
|
| 1688 |
+
"eval_samples_per_second": 229.576,
|
| 1689 |
+
"eval_steps_per_second": 4.85,
|
| 1690 |
+
"step": 288
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 165.71428571428572,
|
| 1694 |
+
"grad_norm": 7.8034491539001465,
|
| 1695 |
+
"learning_rate": 5.2500000000000006e-06,
|
| 1696 |
+
"loss": 0.2819,
|
| 1697 |
+
"step": 290
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 165.71428571428572,
|
| 1701 |
+
"eval_accuracy": 0.6971830985915493,
|
| 1702 |
+
"eval_loss": 0.6445870995521545,
|
| 1703 |
+
"eval_runtime": 0.6129,
|
| 1704 |
+
"eval_samples_per_second": 231.7,
|
| 1705 |
+
"eval_steps_per_second": 4.895,
|
| 1706 |
+
"step": 290
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 166.85714285714286,
|
| 1710 |
+
"eval_accuracy": 0.6901408450704225,
|
| 1711 |
+
"eval_loss": 0.6541280150413513,
|
| 1712 |
+
"eval_runtime": 0.6753,
|
| 1713 |
+
"eval_samples_per_second": 210.283,
|
| 1714 |
+
"eval_steps_per_second": 4.443,
|
| 1715 |
+
"step": 292
|
| 1716 |
+
},
|
| 1717 |
+
{
|
| 1718 |
+
"epoch": 168.0,
|
| 1719 |
+
"eval_accuracy": 0.7183098591549296,
|
| 1720 |
+
"eval_loss": 0.6285813450813293,
|
| 1721 |
+
"eval_runtime": 0.6411,
|
| 1722 |
+
"eval_samples_per_second": 221.49,
|
| 1723 |
+
"eval_steps_per_second": 4.679,
|
| 1724 |
+
"step": 294
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 168.57142857142858,
|
| 1728 |
+
"eval_accuracy": 0.7183098591549296,
|
| 1729 |
+
"eval_loss": 0.6063798069953918,
|
| 1730 |
+
"eval_runtime": 0.6266,
|
| 1731 |
+
"eval_samples_per_second": 226.62,
|
| 1732 |
+
"eval_steps_per_second": 4.788,
|
| 1733 |
+
"step": 295
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 169.71428571428572,
|
| 1737 |
+
"eval_accuracy": 0.7253521126760564,
|
| 1738 |
+
"eval_loss": 0.5995392203330994,
|
| 1739 |
+
"eval_runtime": 0.6217,
|
| 1740 |
+
"eval_samples_per_second": 228.41,
|
| 1741 |
+
"eval_steps_per_second": 4.826,
|
| 1742 |
+
"step": 297
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 170.85714285714286,
|
| 1746 |
+
"eval_accuracy": 0.7253521126760564,
|
| 1747 |
+
"eval_loss": 0.6430822610855103,
|
| 1748 |
+
"eval_runtime": 0.6407,
|
| 1749 |
+
"eval_samples_per_second": 221.618,
|
| 1750 |
+
"eval_steps_per_second": 4.682,
|
| 1751 |
+
"step": 299
|
| 1752 |
+
},
|
| 1753 |
+
{
|
| 1754 |
+
"epoch": 171.42857142857142,
|
| 1755 |
+
"grad_norm": 6.425961494445801,
|
| 1756 |
+
"learning_rate": 5e-06,
|
| 1757 |
+
"loss": 0.2744,
|
| 1758 |
+
"step": 300
|
| 1759 |
+
},
|
| 1760 |
+
{
|
| 1761 |
+
"epoch": 172.0,
|
| 1762 |
+
"eval_accuracy": 0.6901408450704225,
|
| 1763 |
+
"eval_loss": 0.6797394156455994,
|
| 1764 |
+
"eval_runtime": 0.6591,
|
| 1765 |
+
"eval_samples_per_second": 215.44,
|
| 1766 |
+
"eval_steps_per_second": 4.552,
|
| 1767 |
+
"step": 301
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 172.57142857142858,
|
| 1771 |
+
"eval_accuracy": 0.6971830985915493,
|
| 1772 |
+
"eval_loss": 0.6715541481971741,
|
| 1773 |
+
"eval_runtime": 0.6541,
|
| 1774 |
+
"eval_samples_per_second": 217.092,
|
| 1775 |
+
"eval_steps_per_second": 4.586,
|
| 1776 |
+
"step": 302
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 173.71428571428572,
|
| 1780 |
+
"eval_accuracy": 0.7253521126760564,
|
| 1781 |
+
"eval_loss": 0.6510439515113831,
|
| 1782 |
+
"eval_runtime": 0.7018,
|
| 1783 |
+
"eval_samples_per_second": 202.337,
|
| 1784 |
+
"eval_steps_per_second": 4.275,
|
| 1785 |
+
"step": 304
|
| 1786 |
+
},
|
| 1787 |
+
{
|
| 1788 |
+
"epoch": 174.85714285714286,
|
| 1789 |
+
"eval_accuracy": 0.7464788732394366,
|
| 1790 |
+
"eval_loss": 0.6362167000770569,
|
| 1791 |
+
"eval_runtime": 0.6994,
|
| 1792 |
+
"eval_samples_per_second": 203.037,
|
| 1793 |
+
"eval_steps_per_second": 4.29,
|
| 1794 |
+
"step": 306
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 176.0,
|
| 1798 |
+
"eval_accuracy": 0.7605633802816901,
|
| 1799 |
+
"eval_loss": 0.615811824798584,
|
| 1800 |
+
"eval_runtime": 0.6046,
|
| 1801 |
+
"eval_samples_per_second": 234.884,
|
| 1802 |
+
"eval_steps_per_second": 4.962,
|
| 1803 |
+
"step": 308
|
| 1804 |
+
},
|
| 1805 |
+
{
|
| 1806 |
+
"epoch": 176.57142857142858,
|
| 1807 |
+
"eval_accuracy": 0.7676056338028169,
|
| 1808 |
+
"eval_loss": 0.6099287867546082,
|
| 1809 |
+
"eval_runtime": 0.667,
|
| 1810 |
+
"eval_samples_per_second": 212.882,
|
| 1811 |
+
"eval_steps_per_second": 4.498,
|
| 1812 |
+
"step": 309
|
| 1813 |
+
},
|
| 1814 |
+
{
|
| 1815 |
+
"epoch": 177.14285714285714,
|
| 1816 |
+
"grad_norm": 7.763293266296387,
|
| 1817 |
+
"learning_rate": 4.75e-06,
|
| 1818 |
+
"loss": 0.2867,
|
| 1819 |
+
"step": 310
|
| 1820 |
+
},
|
| 1821 |
+
{
|
| 1822 |
+
"epoch": 177.71428571428572,
|
| 1823 |
+
"eval_accuracy": 0.7535211267605634,
|
| 1824 |
+
"eval_loss": 0.6111687421798706,
|
| 1825 |
+
"eval_runtime": 0.6977,
|
| 1826 |
+
"eval_samples_per_second": 203.536,
|
| 1827 |
+
"eval_steps_per_second": 4.3,
|
| 1828 |
+
"step": 311
|
| 1829 |
+
},
|
| 1830 |
+
{
|
| 1831 |
+
"epoch": 178.85714285714286,
|
| 1832 |
+
"eval_accuracy": 0.7464788732394366,
|
| 1833 |
+
"eval_loss": 0.6035374402999878,
|
| 1834 |
+
"eval_runtime": 0.6171,
|
| 1835 |
+
"eval_samples_per_second": 230.092,
|
| 1836 |
+
"eval_steps_per_second": 4.861,
|
| 1837 |
+
"step": 313
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 180.0,
|
| 1841 |
+
"eval_accuracy": 0.7676056338028169,
|
| 1842 |
+
"eval_loss": 0.5815895795822144,
|
| 1843 |
+
"eval_runtime": 0.6334,
|
| 1844 |
+
"eval_samples_per_second": 224.175,
|
| 1845 |
+
"eval_steps_per_second": 4.736,
|
| 1846 |
+
"step": 315
|
| 1847 |
+
},
|
| 1848 |
+
{
|
| 1849 |
+
"epoch": 180.57142857142858,
|
| 1850 |
+
"eval_accuracy": 0.7676056338028169,
|
| 1851 |
+
"eval_loss": 0.5818370580673218,
|
| 1852 |
+
"eval_runtime": 0.6341,
|
| 1853 |
+
"eval_samples_per_second": 223.945,
|
| 1854 |
+
"eval_steps_per_second": 4.731,
|
| 1855 |
+
"step": 316
|
| 1856 |
+
},
|
| 1857 |
+
{
|
| 1858 |
+
"epoch": 181.71428571428572,
|
| 1859 |
+
"eval_accuracy": 0.7676056338028169,
|
| 1860 |
+
"eval_loss": 0.6077942252159119,
|
| 1861 |
+
"eval_runtime": 0.7182,
|
| 1862 |
+
"eval_samples_per_second": 197.71,
|
| 1863 |
+
"eval_steps_per_second": 4.177,
|
| 1864 |
+
"step": 318
|
| 1865 |
+
},
|
| 1866 |
+
{
|
| 1867 |
+
"epoch": 182.85714285714286,
|
| 1868 |
+
"grad_norm": 6.135729789733887,
|
| 1869 |
+
"learning_rate": 4.5e-06,
|
| 1870 |
+
"loss": 0.2883,
|
| 1871 |
+
"step": 320
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 182.85714285714286,
|
| 1875 |
+
"eval_accuracy": 0.7535211267605634,
|
| 1876 |
+
"eval_loss": 0.6083420515060425,
|
| 1877 |
+
"eval_runtime": 0.6578,
|
| 1878 |
+
"eval_samples_per_second": 215.872,
|
| 1879 |
+
"eval_steps_per_second": 4.561,
|
| 1880 |
+
"step": 320
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 184.0,
|
| 1884 |
+
"eval_accuracy": 0.7464788732394366,
|
| 1885 |
+
"eval_loss": 0.5928220152854919,
|
| 1886 |
+
"eval_runtime": 0.6159,
|
| 1887 |
+
"eval_samples_per_second": 230.544,
|
| 1888 |
+
"eval_steps_per_second": 4.871,
|
| 1889 |
+
"step": 322
|
| 1890 |
+
},
|
| 1891 |
+
{
|
| 1892 |
+
"epoch": 184.57142857142858,
|
| 1893 |
+
"eval_accuracy": 0.7535211267605634,
|
| 1894 |
+
"eval_loss": 0.5862187743186951,
|
| 1895 |
+
"eval_runtime": 0.6212,
|
| 1896 |
+
"eval_samples_per_second": 228.598,
|
| 1897 |
+
"eval_steps_per_second": 4.83,
|
| 1898 |
+
"step": 323
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 185.71428571428572,
|
| 1902 |
+
"eval_accuracy": 0.7676056338028169,
|
| 1903 |
+
"eval_loss": 0.562484622001648,
|
| 1904 |
+
"eval_runtime": 0.6399,
|
| 1905 |
+
"eval_samples_per_second": 221.922,
|
| 1906 |
+
"eval_steps_per_second": 4.688,
|
| 1907 |
+
"step": 325
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"epoch": 186.85714285714286,
|
| 1911 |
+
"eval_accuracy": 0.7816901408450704,
|
| 1912 |
+
"eval_loss": 0.5579776763916016,
|
| 1913 |
+
"eval_runtime": 0.6043,
|
| 1914 |
+
"eval_samples_per_second": 234.992,
|
| 1915 |
+
"eval_steps_per_second": 4.965,
|
| 1916 |
+
"step": 327
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 188.0,
|
| 1920 |
+
"eval_accuracy": 0.7535211267605634,
|
| 1921 |
+
"eval_loss": 0.5944541692733765,
|
| 1922 |
+
"eval_runtime": 0.652,
|
| 1923 |
+
"eval_samples_per_second": 217.789,
|
| 1924 |
+
"eval_steps_per_second": 4.601,
|
| 1925 |
+
"step": 329
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 188.57142857142858,
|
| 1929 |
+
"grad_norm": 7.581370830535889,
|
| 1930 |
+
"learning_rate": 4.25e-06,
|
| 1931 |
+
"loss": 0.2852,
|
| 1932 |
+
"step": 330
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 188.57142857142858,
|
| 1936 |
+
"eval_accuracy": 0.6971830985915493,
|
| 1937 |
+
"eval_loss": 0.6320931911468506,
|
| 1938 |
+
"eval_runtime": 0.6488,
|
| 1939 |
+
"eval_samples_per_second": 218.867,
|
| 1940 |
+
"eval_steps_per_second": 4.624,
|
| 1941 |
+
"step": 330
|
| 1942 |
+
},
|
| 1943 |
+
{
|
| 1944 |
+
"epoch": 189.71428571428572,
|
| 1945 |
+
"eval_accuracy": 0.6619718309859155,
|
| 1946 |
+
"eval_loss": 0.6650228500366211,
|
| 1947 |
+
"eval_runtime": 0.6204,
|
| 1948 |
+
"eval_samples_per_second": 228.895,
|
| 1949 |
+
"eval_steps_per_second": 4.836,
|
| 1950 |
+
"step": 332
|
| 1951 |
+
},
|
| 1952 |
+
{
|
| 1953 |
+
"epoch": 190.85714285714286,
|
| 1954 |
+
"eval_accuracy": 0.6690140845070423,
|
| 1955 |
+
"eval_loss": 0.6611641645431519,
|
| 1956 |
+
"eval_runtime": 0.6316,
|
| 1957 |
+
"eval_samples_per_second": 224.843,
|
| 1958 |
+
"eval_steps_per_second": 4.75,
|
| 1959 |
+
"step": 334
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 192.0,
|
| 1963 |
+
"eval_accuracy": 0.676056338028169,
|
| 1964 |
+
"eval_loss": 0.6455404758453369,
|
| 1965 |
+
"eval_runtime": 0.6769,
|
| 1966 |
+
"eval_samples_per_second": 209.794,
|
| 1967 |
+
"eval_steps_per_second": 4.432,
|
| 1968 |
+
"step": 336
|
| 1969 |
+
},
|
| 1970 |
+
{
|
| 1971 |
+
"epoch": 192.57142857142858,
|
| 1972 |
+
"eval_accuracy": 0.7112676056338029,
|
| 1973 |
+
"eval_loss": 0.6290314197540283,
|
| 1974 |
+
"eval_runtime": 0.6314,
|
| 1975 |
+
"eval_samples_per_second": 224.887,
|
| 1976 |
+
"eval_steps_per_second": 4.751,
|
| 1977 |
+
"step": 337
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 193.71428571428572,
|
| 1981 |
+
"eval_accuracy": 0.7394366197183099,
|
| 1982 |
+
"eval_loss": 0.6035693287849426,
|
| 1983 |
+
"eval_runtime": 0.6972,
|
| 1984 |
+
"eval_samples_per_second": 203.665,
|
| 1985 |
+
"eval_steps_per_second": 4.303,
|
| 1986 |
+
"step": 339
|
| 1987 |
+
},
|
| 1988 |
+
{
|
| 1989 |
+
"epoch": 194.28571428571428,
|
| 1990 |
+
"grad_norm": 8.776147842407227,
|
| 1991 |
+
"learning_rate": 4.000000000000001e-06,
|
| 1992 |
+
"loss": 0.2941,
|
| 1993 |
+
"step": 340
|
| 1994 |
+
},
|
| 1995 |
+
{
|
| 1996 |
+
"epoch": 194.85714285714286,
|
| 1997 |
+
"eval_accuracy": 0.7535211267605634,
|
| 1998 |
+
"eval_loss": 0.5879209637641907,
|
| 1999 |
+
"eval_runtime": 0.6487,
|
| 2000 |
+
"eval_samples_per_second": 218.901,
|
| 2001 |
+
"eval_steps_per_second": 4.625,
|
| 2002 |
+
"step": 341
|
| 2003 |
+
},
|
| 2004 |
+
{
|
| 2005 |
+
"epoch": 196.0,
|
| 2006 |
+
"eval_accuracy": 0.7253521126760564,
|
| 2007 |
+
"eval_loss": 0.6135103702545166,
|
| 2008 |
+
"eval_runtime": 0.6192,
|
| 2009 |
+
"eval_samples_per_second": 229.335,
|
| 2010 |
+
"eval_steps_per_second": 4.845,
|
| 2011 |
+
"step": 343
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 196.57142857142858,
|
| 2015 |
+
"eval_accuracy": 0.7112676056338029,
|
| 2016 |
+
"eval_loss": 0.6294575929641724,
|
| 2017 |
+
"eval_runtime": 0.6245,
|
| 2018 |
+
"eval_samples_per_second": 227.392,
|
| 2019 |
+
"eval_steps_per_second": 4.804,
|
| 2020 |
+
"step": 344
|
| 2021 |
+
},
|
| 2022 |
+
{
|
| 2023 |
+
"epoch": 197.71428571428572,
|
| 2024 |
+
"eval_accuracy": 0.6830985915492958,
|
| 2025 |
+
"eval_loss": 0.6444960832595825,
|
| 2026 |
+
"eval_runtime": 0.6631,
|
| 2027 |
+
"eval_samples_per_second": 214.15,
|
| 2028 |
+
"eval_steps_per_second": 4.524,
|
| 2029 |
+
"step": 346
|
| 2030 |
+
},
|
| 2031 |
+
{
|
| 2032 |
+
"epoch": 198.85714285714286,
|
| 2033 |
+
"eval_accuracy": 0.6690140845070423,
|
| 2034 |
+
"eval_loss": 0.6591252088546753,
|
| 2035 |
+
"eval_runtime": 0.646,
|
| 2036 |
+
"eval_samples_per_second": 219.801,
|
| 2037 |
+
"eval_steps_per_second": 4.644,
|
| 2038 |
+
"step": 348
|
| 2039 |
+
},
|
| 2040 |
+
{
|
| 2041 |
+
"epoch": 200.0,
|
| 2042 |
+
"grad_norm": 9.722872734069824,
|
| 2043 |
+
"learning_rate": 3.7500000000000005e-06,
|
| 2044 |
+
"loss": 0.2692,
|
| 2045 |
+
"step": 350
|
| 2046 |
+
},
|
| 2047 |
+
{
|
| 2048 |
+
"epoch": 200.0,
|
| 2049 |
+
"eval_accuracy": 0.6830985915492958,
|
| 2050 |
+
"eval_loss": 0.6556887626647949,
|
| 2051 |
+
"eval_runtime": 0.6171,
|
| 2052 |
+
"eval_samples_per_second": 230.104,
|
| 2053 |
+
"eval_steps_per_second": 4.861,
|
| 2054 |
+
"step": 350
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 200.57142857142858,
|
| 2058 |
+
"eval_accuracy": 0.7112676056338029,
|
| 2059 |
+
"eval_loss": 0.648453414440155,
|
| 2060 |
+
"eval_runtime": 0.6913,
|
| 2061 |
+
"eval_samples_per_second": 205.406,
|
| 2062 |
+
"eval_steps_per_second": 4.34,
|
| 2063 |
+
"step": 351
|
| 2064 |
+
},
|
| 2065 |
+
{
|
| 2066 |
+
"epoch": 201.71428571428572,
|
| 2067 |
+
"eval_accuracy": 0.7183098591549296,
|
| 2068 |
+
"eval_loss": 0.6519960761070251,
|
| 2069 |
+
"eval_runtime": 0.6264,
|
| 2070 |
+
"eval_samples_per_second": 226.708,
|
| 2071 |
+
"eval_steps_per_second": 4.79,
|
| 2072 |
+
"step": 353
|
| 2073 |
+
},
|
| 2074 |
+
{
|
| 2075 |
+
"epoch": 202.85714285714286,
|
| 2076 |
+
"eval_accuracy": 0.7112676056338029,
|
| 2077 |
+
"eval_loss": 0.6673240065574646,
|
| 2078 |
+
"eval_runtime": 0.6108,
|
| 2079 |
+
"eval_samples_per_second": 232.475,
|
| 2080 |
+
"eval_steps_per_second": 4.911,
|
| 2081 |
+
"step": 355
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 204.0,
|
| 2085 |
+
"eval_accuracy": 0.7183098591549296,
|
| 2086 |
+
"eval_loss": 0.6813691258430481,
|
| 2087 |
+
"eval_runtime": 0.6655,
|
| 2088 |
+
"eval_samples_per_second": 213.38,
|
| 2089 |
+
"eval_steps_per_second": 4.508,
|
| 2090 |
+
"step": 357
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 204.57142857142858,
|
| 2094 |
+
"eval_accuracy": 0.7112676056338029,
|
| 2095 |
+
"eval_loss": 0.6693511605262756,
|
| 2096 |
+
"eval_runtime": 0.6097,
|
| 2097 |
+
"eval_samples_per_second": 232.913,
|
| 2098 |
+
"eval_steps_per_second": 4.921,
|
| 2099 |
+
"step": 358
|
| 2100 |
+
},
|
| 2101 |
+
{
|
| 2102 |
+
"epoch": 205.71428571428572,
|
| 2103 |
+
"grad_norm": 9.632829666137695,
|
| 2104 |
+
"learning_rate": 3.5e-06,
|
| 2105 |
+
"loss": 0.2666,
|
| 2106 |
+
"step": 360
|
| 2107 |
+
},
|
| 2108 |
+
{
|
| 2109 |
+
"epoch": 205.71428571428572,
|
| 2110 |
+
"eval_accuracy": 0.7253521126760564,
|
| 2111 |
+
"eval_loss": 0.6349532604217529,
|
| 2112 |
+
"eval_runtime": 0.6697,
|
| 2113 |
+
"eval_samples_per_second": 212.051,
|
| 2114 |
+
"eval_steps_per_second": 4.48,
|
| 2115 |
+
"step": 360
|
| 2116 |
+
},
|
| 2117 |
+
{
|
| 2118 |
+
"epoch": 206.85714285714286,
|
| 2119 |
+
"eval_accuracy": 0.7464788732394366,
|
| 2120 |
+
"eval_loss": 0.6090918779373169,
|
| 2121 |
+
"eval_runtime": 0.6143,
|
| 2122 |
+
"eval_samples_per_second": 231.175,
|
| 2123 |
+
"eval_steps_per_second": 4.884,
|
| 2124 |
+
"step": 362
|
| 2125 |
+
},
|
| 2126 |
+
{
|
| 2127 |
+
"epoch": 208.0,
|
| 2128 |
+
"eval_accuracy": 0.7394366197183099,
|
| 2129 |
+
"eval_loss": 0.6221663951873779,
|
| 2130 |
+
"eval_runtime": 0.629,
|
| 2131 |
+
"eval_samples_per_second": 225.743,
|
| 2132 |
+
"eval_steps_per_second": 4.769,
|
| 2133 |
+
"step": 364
|
| 2134 |
+
},
|
| 2135 |
+
{
|
| 2136 |
+
"epoch": 208.57142857142858,
|
| 2137 |
+
"eval_accuracy": 0.7394366197183099,
|
| 2138 |
+
"eval_loss": 0.6363082528114319,
|
| 2139 |
+
"eval_runtime": 0.6924,
|
| 2140 |
+
"eval_samples_per_second": 205.083,
|
| 2141 |
+
"eval_steps_per_second": 4.333,
|
| 2142 |
+
"step": 365
|
| 2143 |
+
},
|
| 2144 |
+
{
|
| 2145 |
+
"epoch": 209.71428571428572,
|
| 2146 |
+
"eval_accuracy": 0.7394366197183099,
|
| 2147 |
+
"eval_loss": 0.639771044254303,
|
| 2148 |
+
"eval_runtime": 0.6787,
|
| 2149 |
+
"eval_samples_per_second": 209.212,
|
| 2150 |
+
"eval_steps_per_second": 4.42,
|
| 2151 |
+
"step": 367
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 210.85714285714286,
|
| 2155 |
+
"eval_accuracy": 0.7253521126760564,
|
| 2156 |
+
"eval_loss": 0.6554509401321411,
|
| 2157 |
+
"eval_runtime": 0.6896,
|
| 2158 |
+
"eval_samples_per_second": 205.904,
|
| 2159 |
+
"eval_steps_per_second": 4.35,
|
| 2160 |
+
"step": 369
|
| 2161 |
+
},
|
| 2162 |
+
{
|
| 2163 |
+
"epoch": 211.42857142857142,
|
| 2164 |
+
"grad_norm": 10.489874839782715,
|
| 2165 |
+
"learning_rate": 3.2500000000000002e-06,
|
| 2166 |
+
"loss": 0.2745,
|
| 2167 |
+
"step": 370
|
| 2168 |
+
},
|
| 2169 |
+
{
|
| 2170 |
+
"epoch": 212.0,
|
| 2171 |
+
"eval_accuracy": 0.7253521126760564,
|
| 2172 |
+
"eval_loss": 0.6555443406105042,
|
| 2173 |
+
"eval_runtime": 0.6446,
|
| 2174 |
+
"eval_samples_per_second": 220.289,
|
| 2175 |
+
"eval_steps_per_second": 4.654,
|
| 2176 |
+
"step": 371
|
| 2177 |
+
},
|
| 2178 |
+
{
|
| 2179 |
+
"epoch": 212.57142857142858,
|
| 2180 |
+
"eval_accuracy": 0.7394366197183099,
|
| 2181 |
+
"eval_loss": 0.6466799974441528,
|
| 2182 |
+
"eval_runtime": 0.6722,
|
| 2183 |
+
"eval_samples_per_second": 211.262,
|
| 2184 |
+
"eval_steps_per_second": 4.463,
|
| 2185 |
+
"step": 372
|
| 2186 |
+
},
|
| 2187 |
+
{
|
| 2188 |
+
"epoch": 213.71428571428572,
|
| 2189 |
+
"eval_accuracy": 0.7605633802816901,
|
| 2190 |
+
"eval_loss": 0.6215640902519226,
|
| 2191 |
+
"eval_runtime": 0.6123,
|
| 2192 |
+
"eval_samples_per_second": 231.925,
|
| 2193 |
+
"eval_steps_per_second": 4.9,
|
| 2194 |
+
"step": 374
|
| 2195 |
+
},
|
| 2196 |
+
{
|
| 2197 |
+
"epoch": 214.85714285714286,
|
| 2198 |
+
"eval_accuracy": 0.7676056338028169,
|
| 2199 |
+
"eval_loss": 0.6065743565559387,
|
| 2200 |
+
"eval_runtime": 0.678,
|
| 2201 |
+
"eval_samples_per_second": 209.443,
|
| 2202 |
+
"eval_steps_per_second": 4.425,
|
| 2203 |
+
"step": 376
|
| 2204 |
+
},
|
| 2205 |
+
{
|
| 2206 |
+
"epoch": 216.0,
|
| 2207 |
+
"eval_accuracy": 0.7605633802816901,
|
| 2208 |
+
"eval_loss": 0.6082801222801208,
|
| 2209 |
+
"eval_runtime": 0.6148,
|
| 2210 |
+
"eval_samples_per_second": 230.957,
|
| 2211 |
+
"eval_steps_per_second": 4.879,
|
| 2212 |
+
"step": 378
|
| 2213 |
+
},
|
| 2214 |
+
{
|
| 2215 |
+
"epoch": 216.57142857142858,
|
| 2216 |
+
"eval_accuracy": 0.7535211267605634,
|
| 2217 |
+
"eval_loss": 0.6151567697525024,
|
| 2218 |
+
"eval_runtime": 0.6887,
|
| 2219 |
+
"eval_samples_per_second": 206.18,
|
| 2220 |
+
"eval_steps_per_second": 4.356,
|
| 2221 |
+
"step": 379
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 217.14285714285714,
|
| 2225 |
+
"grad_norm": 8.25847339630127,
|
| 2226 |
+
"learning_rate": 3e-06,
|
| 2227 |
+
"loss": 0.2578,
|
| 2228 |
+
"step": 380
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 217.71428571428572,
|
| 2232 |
+
"eval_accuracy": 0.7535211267605634,
|
| 2233 |
+
"eval_loss": 0.6162380576133728,
|
| 2234 |
+
"eval_runtime": 0.6165,
|
| 2235 |
+
"eval_samples_per_second": 230.314,
|
| 2236 |
+
"eval_steps_per_second": 4.866,
|
| 2237 |
+
"step": 381
|
| 2238 |
+
},
|
| 2239 |
+
{
|
| 2240 |
+
"epoch": 218.85714285714286,
|
| 2241 |
+
"eval_accuracy": 0.7535211267605634,
|
| 2242 |
+
"eval_loss": 0.6096570491790771,
|
| 2243 |
+
"eval_runtime": 0.6922,
|
| 2244 |
+
"eval_samples_per_second": 205.145,
|
| 2245 |
+
"eval_steps_per_second": 4.334,
|
| 2246 |
+
"step": 383
|
| 2247 |
+
},
|
| 2248 |
+
{
|
| 2249 |
+
"epoch": 220.0,
|
| 2250 |
+
"eval_accuracy": 0.7464788732394366,
|
| 2251 |
+
"eval_loss": 0.6003305315971375,
|
| 2252 |
+
"eval_runtime": 0.7013,
|
| 2253 |
+
"eval_samples_per_second": 202.483,
|
| 2254 |
+
"eval_steps_per_second": 4.278,
|
| 2255 |
+
"step": 385
|
| 2256 |
+
},
|
| 2257 |
+
{
|
| 2258 |
+
"epoch": 220.57142857142858,
|
| 2259 |
+
"eval_accuracy": 0.7535211267605634,
|
| 2260 |
+
"eval_loss": 0.6063842177391052,
|
| 2261 |
+
"eval_runtime": 0.6714,
|
| 2262 |
+
"eval_samples_per_second": 211.501,
|
| 2263 |
+
"eval_steps_per_second": 4.468,
|
| 2264 |
+
"step": 386
|
| 2265 |
+
},
|
| 2266 |
+
{
|
| 2267 |
+
"epoch": 221.71428571428572,
|
| 2268 |
+
"eval_accuracy": 0.7535211267605634,
|
| 2269 |
+
"eval_loss": 0.6182224154472351,
|
| 2270 |
+
"eval_runtime": 0.6756,
|
| 2271 |
+
"eval_samples_per_second": 210.176,
|
| 2272 |
+
"eval_steps_per_second": 4.44,
|
| 2273 |
+
"step": 388
|
| 2274 |
+
},
|
| 2275 |
+
{
|
| 2276 |
+
"epoch": 222.85714285714286,
|
| 2277 |
+
"grad_norm": 7.586206436157227,
|
| 2278 |
+
"learning_rate": 2.7500000000000004e-06,
|
| 2279 |
+
"loss": 0.2637,
|
| 2280 |
+
"step": 390
|
| 2281 |
+
},
|
| 2282 |
+
{
|
| 2283 |
+
"epoch": 222.85714285714286,
|
| 2284 |
+
"eval_accuracy": 0.7464788732394366,
|
| 2285 |
+
"eval_loss": 0.6464552283287048,
|
| 2286 |
+
"eval_runtime": 0.6274,
|
| 2287 |
+
"eval_samples_per_second": 226.325,
|
| 2288 |
+
"eval_steps_per_second": 4.782,
|
| 2289 |
+
"step": 390
|
| 2290 |
+
},
|
| 2291 |
+
{
|
| 2292 |
+
"epoch": 224.0,
|
| 2293 |
+
"eval_accuracy": 0.7535211267605634,
|
| 2294 |
+
"eval_loss": 0.6461374163627625,
|
| 2295 |
+
"eval_runtime": 0.6544,
|
| 2296 |
+
"eval_samples_per_second": 216.993,
|
| 2297 |
+
"eval_steps_per_second": 4.584,
|
| 2298 |
+
"step": 392
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 224.57142857142858,
|
| 2302 |
+
"eval_accuracy": 0.7535211267605634,
|
| 2303 |
+
"eval_loss": 0.6352179050445557,
|
| 2304 |
+
"eval_runtime": 0.6812,
|
| 2305 |
+
"eval_samples_per_second": 208.441,
|
| 2306 |
+
"eval_steps_per_second": 4.404,
|
| 2307 |
+
"step": 393
|
| 2308 |
+
},
|
| 2309 |
+
{
|
| 2310 |
+
"epoch": 225.71428571428572,
|
| 2311 |
+
"eval_accuracy": 0.7605633802816901,
|
| 2312 |
+
"eval_loss": 0.6018115282058716,
|
| 2313 |
+
"eval_runtime": 0.7046,
|
| 2314 |
+
"eval_samples_per_second": 201.545,
|
| 2315 |
+
"eval_steps_per_second": 4.258,
|
| 2316 |
+
"step": 395
|
| 2317 |
+
},
|
| 2318 |
+
{
|
| 2319 |
+
"epoch": 226.85714285714286,
|
| 2320 |
+
"eval_accuracy": 0.7746478873239436,
|
| 2321 |
+
"eval_loss": 0.5855334401130676,
|
| 2322 |
+
"eval_runtime": 0.6735,
|
| 2323 |
+
"eval_samples_per_second": 210.832,
|
| 2324 |
+
"eval_steps_per_second": 4.454,
|
| 2325 |
+
"step": 397
|
| 2326 |
+
},
|
| 2327 |
+
{
|
| 2328 |
+
"epoch": 228.0,
|
| 2329 |
+
"eval_accuracy": 0.7605633802816901,
|
| 2330 |
+
"eval_loss": 0.5916351675987244,
|
| 2331 |
+
"eval_runtime": 0.6135,
|
| 2332 |
+
"eval_samples_per_second": 231.446,
|
| 2333 |
+
"eval_steps_per_second": 4.89,
|
| 2334 |
+
"step": 399
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"epoch": 228.57142857142858,
|
| 2338 |
+
"grad_norm": 6.688592910766602,
|
| 2339 |
+
"learning_rate": 2.5e-06,
|
| 2340 |
+
"loss": 0.2696,
|
| 2341 |
+
"step": 400
|
| 2342 |
+
},
|
| 2343 |
+
{
|
| 2344 |
+
"epoch": 228.57142857142858,
|
| 2345 |
+
"eval_accuracy": 0.7605633802816901,
|
| 2346 |
+
"eval_loss": 0.6031278371810913,
|
| 2347 |
+
"eval_runtime": 0.7042,
|
| 2348 |
+
"eval_samples_per_second": 201.644,
|
| 2349 |
+
"eval_steps_per_second": 4.26,
|
| 2350 |
+
"step": 400
|
| 2351 |
+
},
|
| 2352 |
+
{
|
| 2353 |
+
"epoch": 229.71428571428572,
|
| 2354 |
+
"eval_accuracy": 0.7605633802816901,
|
| 2355 |
+
"eval_loss": 0.6307948231697083,
|
| 2356 |
+
"eval_runtime": 0.6195,
|
| 2357 |
+
"eval_samples_per_second": 229.23,
|
| 2358 |
+
"eval_steps_per_second": 4.843,
|
| 2359 |
+
"step": 402
|
| 2360 |
+
},
|
| 2361 |
+
{
|
| 2362 |
+
"epoch": 230.85714285714286,
|
| 2363 |
+
"eval_accuracy": 0.7464788732394366,
|
| 2364 |
+
"eval_loss": 0.6435055732727051,
|
| 2365 |
+
"eval_runtime": 0.6377,
|
| 2366 |
+
"eval_samples_per_second": 222.671,
|
| 2367 |
+
"eval_steps_per_second": 4.704,
|
| 2368 |
+
"step": 404
|
| 2369 |
+
},
|
| 2370 |
+
{
|
| 2371 |
+
"epoch": 232.0,
|
| 2372 |
+
"eval_accuracy": 0.7464788732394366,
|
| 2373 |
+
"eval_loss": 0.6324659585952759,
|
| 2374 |
+
"eval_runtime": 0.6593,
|
| 2375 |
+
"eval_samples_per_second": 215.368,
|
| 2376 |
+
"eval_steps_per_second": 4.55,
|
| 2377 |
+
"step": 406
|
| 2378 |
+
},
|
| 2379 |
+
{
|
| 2380 |
+
"epoch": 232.57142857142858,
|
| 2381 |
+
"eval_accuracy": 0.7535211267605634,
|
| 2382 |
+
"eval_loss": 0.6212195158004761,
|
| 2383 |
+
"eval_runtime": 0.6515,
|
| 2384 |
+
"eval_samples_per_second": 217.97,
|
| 2385 |
+
"eval_steps_per_second": 4.605,
|
| 2386 |
+
"step": 407
|
| 2387 |
+
},
|
| 2388 |
+
{
|
| 2389 |
+
"epoch": 233.71428571428572,
|
| 2390 |
+
"eval_accuracy": 0.7535211267605634,
|
| 2391 |
+
"eval_loss": 0.5985831618309021,
|
| 2392 |
+
"eval_runtime": 0.6199,
|
| 2393 |
+
"eval_samples_per_second": 229.057,
|
| 2394 |
+
"eval_steps_per_second": 4.839,
|
| 2395 |
+
"step": 409
|
| 2396 |
+
},
|
| 2397 |
+
{
|
| 2398 |
+
"epoch": 234.28571428571428,
|
| 2399 |
+
"grad_norm": 10.518420219421387,
|
| 2400 |
+
"learning_rate": 2.25e-06,
|
| 2401 |
+
"loss": 0.2697,
|
| 2402 |
+
"step": 410
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 234.85714285714286,
|
| 2406 |
+
"eval_accuracy": 0.7464788732394366,
|
| 2407 |
+
"eval_loss": 0.5964060425758362,
|
| 2408 |
+
"eval_runtime": 0.6379,
|
| 2409 |
+
"eval_samples_per_second": 222.588,
|
| 2410 |
+
"eval_steps_per_second": 4.703,
|
| 2411 |
+
"step": 411
|
| 2412 |
+
},
|
| 2413 |
+
{
|
| 2414 |
+
"epoch": 236.0,
|
| 2415 |
+
"eval_accuracy": 0.7464788732394366,
|
| 2416 |
+
"eval_loss": 0.5949541330337524,
|
| 2417 |
+
"eval_runtime": 0.6955,
|
| 2418 |
+
"eval_samples_per_second": 204.178,
|
| 2419 |
+
"eval_steps_per_second": 4.314,
|
| 2420 |
+
"step": 413
|
| 2421 |
+
},
|
| 2422 |
+
{
|
| 2423 |
+
"epoch": 236.57142857142858,
|
| 2424 |
+
"eval_accuracy": 0.7464788732394366,
|
| 2425 |
+
"eval_loss": 0.5986204743385315,
|
| 2426 |
+
"eval_runtime": 0.6833,
|
| 2427 |
+
"eval_samples_per_second": 207.804,
|
| 2428 |
+
"eval_steps_per_second": 4.39,
|
| 2429 |
+
"step": 414
|
| 2430 |
+
},
|
| 2431 |
+
{
|
| 2432 |
+
"epoch": 237.71428571428572,
|
| 2433 |
+
"eval_accuracy": 0.7535211267605634,
|
| 2434 |
+
"eval_loss": 0.6066287755966187,
|
| 2435 |
+
"eval_runtime": 0.6316,
|
| 2436 |
+
"eval_samples_per_second": 224.842,
|
| 2437 |
+
"eval_steps_per_second": 4.75,
|
| 2438 |
+
"step": 416
|
| 2439 |
+
},
|
| 2440 |
+
{
|
| 2441 |
+
"epoch": 238.85714285714286,
|
| 2442 |
+
"eval_accuracy": 0.7535211267605634,
|
| 2443 |
+
"eval_loss": 0.6035082936286926,
|
| 2444 |
+
"eval_runtime": 0.6521,
|
| 2445 |
+
"eval_samples_per_second": 217.753,
|
| 2446 |
+
"eval_steps_per_second": 4.6,
|
| 2447 |
+
"step": 418
|
| 2448 |
+
},
|
| 2449 |
+
{
|
| 2450 |
+
"epoch": 240.0,
|
| 2451 |
+
"grad_norm": 8.466328620910645,
|
| 2452 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 2453 |
+
"loss": 0.2659,
|
| 2454 |
+
"step": 420
|
| 2455 |
+
},
|
| 2456 |
+
{
|
| 2457 |
+
"epoch": 240.0,
|
| 2458 |
+
"eval_accuracy": 0.7535211267605634,
|
| 2459 |
+
"eval_loss": 0.603872537612915,
|
| 2460 |
+
"eval_runtime": 0.6105,
|
| 2461 |
+
"eval_samples_per_second": 232.606,
|
| 2462 |
+
"eval_steps_per_second": 4.914,
|
| 2463 |
+
"step": 420
|
| 2464 |
+
},
|
| 2465 |
+
{
|
| 2466 |
+
"epoch": 240.57142857142858,
|
| 2467 |
+
"eval_accuracy": 0.7535211267605634,
|
| 2468 |
+
"eval_loss": 0.6003507375717163,
|
| 2469 |
+
"eval_runtime": 0.7014,
|
| 2470 |
+
"eval_samples_per_second": 202.45,
|
| 2471 |
+
"eval_steps_per_second": 4.277,
|
| 2472 |
+
"step": 421
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 241.71428571428572,
|
| 2476 |
+
"eval_accuracy": 0.7535211267605634,
|
| 2477 |
+
"eval_loss": 0.6000791788101196,
|
| 2478 |
+
"eval_runtime": 0.622,
|
| 2479 |
+
"eval_samples_per_second": 228.292,
|
| 2480 |
+
"eval_steps_per_second": 4.823,
|
| 2481 |
+
"step": 423
|
| 2482 |
+
},
|
| 2483 |
+
{
|
| 2484 |
+
"epoch": 242.85714285714286,
|
| 2485 |
+
"eval_accuracy": 0.7464788732394366,
|
| 2486 |
+
"eval_loss": 0.594083309173584,
|
| 2487 |
+
"eval_runtime": 0.628,
|
| 2488 |
+
"eval_samples_per_second": 226.107,
|
| 2489 |
+
"eval_steps_per_second": 4.777,
|
| 2490 |
+
"step": 425
|
| 2491 |
+
},
|
| 2492 |
+
{
|
| 2493 |
+
"epoch": 244.0,
|
| 2494 |
+
"eval_accuracy": 0.7394366197183099,
|
| 2495 |
+
"eval_loss": 0.5941977500915527,
|
| 2496 |
+
"eval_runtime": 0.6784,
|
| 2497 |
+
"eval_samples_per_second": 209.316,
|
| 2498 |
+
"eval_steps_per_second": 4.422,
|
| 2499 |
+
"step": 427
|
| 2500 |
+
},
|
| 2501 |
+
{
|
| 2502 |
+
"epoch": 244.57142857142858,
|
| 2503 |
+
"eval_accuracy": 0.7464788732394366,
|
| 2504 |
+
"eval_loss": 0.5971840023994446,
|
| 2505 |
+
"eval_runtime": 0.6368,
|
| 2506 |
+
"eval_samples_per_second": 222.979,
|
| 2507 |
+
"eval_steps_per_second": 4.711,
|
| 2508 |
+
"step": 428
|
| 2509 |
+
},
|
| 2510 |
+
{
|
| 2511 |
+
"epoch": 245.71428571428572,
|
| 2512 |
+
"grad_norm": 7.927194595336914,
|
| 2513 |
+
"learning_rate": 1.75e-06,
|
| 2514 |
+
"loss": 0.2529,
|
| 2515 |
+
"step": 430
|
| 2516 |
+
},
|
| 2517 |
+
{
|
| 2518 |
+
"epoch": 245.71428571428572,
|
| 2519 |
+
"eval_accuracy": 0.7535211267605634,
|
| 2520 |
+
"eval_loss": 0.6077085733413696,
|
| 2521 |
+
"eval_runtime": 0.6879,
|
| 2522 |
+
"eval_samples_per_second": 206.42,
|
| 2523 |
+
"eval_steps_per_second": 4.361,
|
| 2524 |
+
"step": 430
|
| 2525 |
+
},
|
| 2526 |
+
{
|
| 2527 |
+
"epoch": 246.85714285714286,
|
| 2528 |
+
"eval_accuracy": 0.7464788732394366,
|
| 2529 |
+
"eval_loss": 0.6173303127288818,
|
| 2530 |
+
"eval_runtime": 0.6034,
|
| 2531 |
+
"eval_samples_per_second": 235.336,
|
| 2532 |
+
"eval_steps_per_second": 4.972,
|
| 2533 |
+
"step": 432
|
| 2534 |
+
},
|
| 2535 |
+
{
|
| 2536 |
+
"epoch": 248.0,
|
| 2537 |
+
"eval_accuracy": 0.7605633802816901,
|
| 2538 |
+
"eval_loss": 0.6129435300827026,
|
| 2539 |
+
"eval_runtime": 0.6618,
|
| 2540 |
+
"eval_samples_per_second": 214.555,
|
| 2541 |
+
"eval_steps_per_second": 4.533,
|
| 2542 |
+
"step": 434
|
| 2543 |
+
},
|
| 2544 |
+
{
|
| 2545 |
+
"epoch": 248.57142857142858,
|
| 2546 |
+
"eval_accuracy": 0.7605633802816901,
|
| 2547 |
+
"eval_loss": 0.6099234819412231,
|
| 2548 |
+
"eval_runtime": 0.6902,
|
| 2549 |
+
"eval_samples_per_second": 205.729,
|
| 2550 |
+
"eval_steps_per_second": 4.346,
|
| 2551 |
+
"step": 435
|
| 2552 |
+
},
|
| 2553 |
+
{
|
| 2554 |
+
"epoch": 249.71428571428572,
|
| 2555 |
+
"eval_accuracy": 0.7605633802816901,
|
| 2556 |
+
"eval_loss": 0.6004898548126221,
|
| 2557 |
+
"eval_runtime": 0.6677,
|
| 2558 |
+
"eval_samples_per_second": 212.685,
|
| 2559 |
+
"eval_steps_per_second": 4.493,
|
| 2560 |
+
"step": 437
|
| 2561 |
+
},
|
| 2562 |
+
{
|
| 2563 |
+
"epoch": 250.85714285714286,
|
| 2564 |
+
"eval_accuracy": 0.7605633802816901,
|
| 2565 |
+
"eval_loss": 0.5919698476791382,
|
| 2566 |
+
"eval_runtime": 0.6164,
|
| 2567 |
+
"eval_samples_per_second": 230.363,
|
| 2568 |
+
"eval_steps_per_second": 4.867,
|
| 2569 |
+
"step": 439
|
| 2570 |
+
},
|
| 2571 |
+
{
|
| 2572 |
+
"epoch": 251.42857142857142,
|
| 2573 |
+
"grad_norm": 6.316997051239014,
|
| 2574 |
+
"learning_rate": 1.5e-06,
|
| 2575 |
+
"loss": 0.261,
|
| 2576 |
+
"step": 440
|
| 2577 |
+
},
|
| 2578 |
+
{
|
| 2579 |
+
"epoch": 252.0,
|
| 2580 |
+
"eval_accuracy": 0.7605633802816901,
|
| 2581 |
+
"eval_loss": 0.5945872664451599,
|
| 2582 |
+
"eval_runtime": 0.67,
|
| 2583 |
+
"eval_samples_per_second": 211.952,
|
| 2584 |
+
"eval_steps_per_second": 4.478,
|
| 2585 |
+
"step": 441
|
| 2586 |
+
},
|
| 2587 |
+
{
|
| 2588 |
+
"epoch": 252.57142857142858,
|
| 2589 |
+
"eval_accuracy": 0.7605633802816901,
|
| 2590 |
+
"eval_loss": 0.5991929769515991,
|
| 2591 |
+
"eval_runtime": 0.6342,
|
| 2592 |
+
"eval_samples_per_second": 223.921,
|
| 2593 |
+
"eval_steps_per_second": 4.731,
|
| 2594 |
+
"step": 442
|
| 2595 |
+
},
|
| 2596 |
+
{
|
| 2597 |
+
"epoch": 253.71428571428572,
|
| 2598 |
+
"eval_accuracy": 0.7605633802816901,
|
| 2599 |
+
"eval_loss": 0.6141655445098877,
|
| 2600 |
+
"eval_runtime": 0.6927,
|
| 2601 |
+
"eval_samples_per_second": 204.998,
|
| 2602 |
+
"eval_steps_per_second": 4.331,
|
| 2603 |
+
"step": 444
|
| 2604 |
+
},
|
| 2605 |
+
{
|
| 2606 |
+
"epoch": 254.85714285714286,
|
| 2607 |
+
"eval_accuracy": 0.7464788732394366,
|
| 2608 |
+
"eval_loss": 0.628886342048645,
|
| 2609 |
+
"eval_runtime": 0.7291,
|
| 2610 |
+
"eval_samples_per_second": 194.758,
|
| 2611 |
+
"eval_steps_per_second": 4.115,
|
| 2612 |
+
"step": 446
|
| 2613 |
+
},
|
| 2614 |
+
{
|
| 2615 |
+
"epoch": 256.0,
|
| 2616 |
+
"eval_accuracy": 0.7464788732394366,
|
| 2617 |
+
"eval_loss": 0.6316113471984863,
|
| 2618 |
+
"eval_runtime": 0.6453,
|
| 2619 |
+
"eval_samples_per_second": 220.039,
|
| 2620 |
+
"eval_steps_per_second": 4.649,
|
| 2621 |
+
"step": 448
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 256.57142857142856,
|
| 2625 |
+
"eval_accuracy": 0.7535211267605634,
|
| 2626 |
+
"eval_loss": 0.6301937699317932,
|
| 2627 |
+
"eval_runtime": 0.6641,
|
| 2628 |
+
"eval_samples_per_second": 213.816,
|
| 2629 |
+
"eval_steps_per_second": 4.517,
|
| 2630 |
+
"step": 449
|
| 2631 |
+
},
|
| 2632 |
+
{
|
| 2633 |
+
"epoch": 257.14285714285717,
|
| 2634 |
+
"grad_norm": 8.127288818359375,
|
| 2635 |
+
"learning_rate": 1.25e-06,
|
| 2636 |
+
"loss": 0.2675,
|
| 2637 |
+
"step": 450
|
| 2638 |
+
},
|
| 2639 |
+
{
|
| 2640 |
+
"epoch": 257.7142857142857,
|
| 2641 |
+
"eval_accuracy": 0.7535211267605634,
|
| 2642 |
+
"eval_loss": 0.6241323351860046,
|
| 2643 |
+
"eval_runtime": 0.7238,
|
| 2644 |
+
"eval_samples_per_second": 196.181,
|
| 2645 |
+
"eval_steps_per_second": 4.145,
|
| 2646 |
+
"step": 451
|
| 2647 |
+
},
|
| 2648 |
+
{
|
| 2649 |
+
"epoch": 258.85714285714283,
|
| 2650 |
+
"eval_accuracy": 0.7535211267605634,
|
| 2651 |
+
"eval_loss": 0.6129297614097595,
|
| 2652 |
+
"eval_runtime": 0.6069,
|
| 2653 |
+
"eval_samples_per_second": 233.994,
|
| 2654 |
+
"eval_steps_per_second": 4.944,
|
| 2655 |
+
"step": 453
|
| 2656 |
+
},
|
| 2657 |
+
{
|
| 2658 |
+
"epoch": 260.0,
|
| 2659 |
+
"eval_accuracy": 0.7464788732394366,
|
| 2660 |
+
"eval_loss": 0.6065521240234375,
|
| 2661 |
+
"eval_runtime": 0.703,
|
| 2662 |
+
"eval_samples_per_second": 201.991,
|
| 2663 |
+
"eval_steps_per_second": 4.267,
|
| 2664 |
+
"step": 455
|
| 2665 |
+
},
|
| 2666 |
+
{
|
| 2667 |
+
"epoch": 260.57142857142856,
|
| 2668 |
+
"eval_accuracy": 0.7464788732394366,
|
| 2669 |
+
"eval_loss": 0.6061440110206604,
|
| 2670 |
+
"eval_runtime": 0.6863,
|
| 2671 |
+
"eval_samples_per_second": 206.913,
|
| 2672 |
+
"eval_steps_per_second": 4.371,
|
| 2673 |
+
"step": 456
|
| 2674 |
+
},
|
| 2675 |
+
{
|
| 2676 |
+
"epoch": 261.7142857142857,
|
| 2677 |
+
"eval_accuracy": 0.7535211267605634,
|
| 2678 |
+
"eval_loss": 0.6098330616950989,
|
| 2679 |
+
"eval_runtime": 0.6103,
|
| 2680 |
+
"eval_samples_per_second": 232.689,
|
| 2681 |
+
"eval_steps_per_second": 4.916,
|
| 2682 |
+
"step": 458
|
| 2683 |
+
},
|
| 2684 |
+
{
|
| 2685 |
+
"epoch": 262.85714285714283,
|
| 2686 |
+
"grad_norm": 8.799203872680664,
|
| 2687 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 2688 |
+
"loss": 0.2737,
|
| 2689 |
+
"step": 460
|
| 2690 |
+
},
|
| 2691 |
+
{
|
| 2692 |
+
"epoch": 262.85714285714283,
|
| 2693 |
+
"eval_accuracy": 0.7394366197183099,
|
| 2694 |
+
"eval_loss": 0.6171931028366089,
|
| 2695 |
+
"eval_runtime": 0.6263,
|
| 2696 |
+
"eval_samples_per_second": 226.74,
|
| 2697 |
+
"eval_steps_per_second": 4.79,
|
| 2698 |
+
"step": 460
|
| 2699 |
+
},
|
| 2700 |
+
{
|
| 2701 |
+
"epoch": 264.0,
|
| 2702 |
+
"eval_accuracy": 0.7323943661971831,
|
| 2703 |
+
"eval_loss": 0.6274363398551941,
|
| 2704 |
+
"eval_runtime": 0.6469,
|
| 2705 |
+
"eval_samples_per_second": 219.51,
|
| 2706 |
+
"eval_steps_per_second": 4.638,
|
| 2707 |
+
"step": 462
|
| 2708 |
+
},
|
| 2709 |
+
{
|
| 2710 |
+
"epoch": 264.57142857142856,
|
| 2711 |
+
"eval_accuracy": 0.7323943661971831,
|
| 2712 |
+
"eval_loss": 0.6298220753669739,
|
| 2713 |
+
"eval_runtime": 0.6375,
|
| 2714 |
+
"eval_samples_per_second": 222.743,
|
| 2715 |
+
"eval_steps_per_second": 4.706,
|
| 2716 |
+
"step": 463
|
| 2717 |
+
},
|
| 2718 |
+
{
|
| 2719 |
+
"epoch": 265.7142857142857,
|
| 2720 |
+
"eval_accuracy": 0.7323943661971831,
|
| 2721 |
+
"eval_loss": 0.629562497138977,
|
| 2722 |
+
"eval_runtime": 0.6336,
|
| 2723 |
+
"eval_samples_per_second": 224.099,
|
| 2724 |
+
"eval_steps_per_second": 4.734,
|
| 2725 |
+
"step": 465
|
| 2726 |
+
},
|
| 2727 |
+
{
|
| 2728 |
+
"epoch": 266.85714285714283,
|
| 2729 |
+
"eval_accuracy": 0.7323943661971831,
|
| 2730 |
+
"eval_loss": 0.6285203695297241,
|
| 2731 |
+
"eval_runtime": 0.6921,
|
| 2732 |
+
"eval_samples_per_second": 205.168,
|
| 2733 |
+
"eval_steps_per_second": 4.335,
|
| 2734 |
+
"step": 467
|
| 2735 |
+
},
|
| 2736 |
+
{
|
| 2737 |
+
"epoch": 268.0,
|
| 2738 |
+
"eval_accuracy": 0.7323943661971831,
|
| 2739 |
+
"eval_loss": 0.6264737248420715,
|
| 2740 |
+
"eval_runtime": 0.6231,
|
| 2741 |
+
"eval_samples_per_second": 227.888,
|
| 2742 |
+
"eval_steps_per_second": 4.815,
|
| 2743 |
+
"step": 469
|
| 2744 |
+
},
|
| 2745 |
+
{
|
| 2746 |
+
"epoch": 268.57142857142856,
|
| 2747 |
+
"grad_norm": 7.2211222648620605,
|
| 2748 |
+
"learning_rate": 7.5e-07,
|
| 2749 |
+
"loss": 0.2504,
|
| 2750 |
+
"step": 470
|
| 2751 |
+
},
|
| 2752 |
+
{
|
| 2753 |
+
"epoch": 268.57142857142856,
|
| 2754 |
+
"eval_accuracy": 0.7464788732394366,
|
| 2755 |
+
"eval_loss": 0.6274013519287109,
|
| 2756 |
+
"eval_runtime": 0.6682,
|
| 2757 |
+
"eval_samples_per_second": 212.519,
|
| 2758 |
+
"eval_steps_per_second": 4.49,
|
| 2759 |
+
"step": 470
|
| 2760 |
+
},
|
| 2761 |
+
{
|
| 2762 |
+
"epoch": 269.7142857142857,
|
| 2763 |
+
"eval_accuracy": 0.7394366197183099,
|
| 2764 |
+
"eval_loss": 0.6285874843597412,
|
| 2765 |
+
"eval_runtime": 0.683,
|
| 2766 |
+
"eval_samples_per_second": 207.896,
|
| 2767 |
+
"eval_steps_per_second": 4.392,
|
| 2768 |
+
"step": 472
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 270.85714285714283,
|
| 2772 |
+
"eval_accuracy": 0.7464788732394366,
|
| 2773 |
+
"eval_loss": 0.6235743761062622,
|
| 2774 |
+
"eval_runtime": 0.6396,
|
| 2775 |
+
"eval_samples_per_second": 221.999,
|
| 2776 |
+
"eval_steps_per_second": 4.69,
|
| 2777 |
+
"step": 474
|
| 2778 |
+
},
|
| 2779 |
+
{
|
| 2780 |
+
"epoch": 272.0,
|
| 2781 |
+
"eval_accuracy": 0.7464788732394366,
|
| 2782 |
+
"eval_loss": 0.617820680141449,
|
| 2783 |
+
"eval_runtime": 0.6619,
|
| 2784 |
+
"eval_samples_per_second": 214.54,
|
| 2785 |
+
"eval_steps_per_second": 4.533,
|
| 2786 |
+
"step": 476
|
| 2787 |
+
},
|
| 2788 |
+
{
|
| 2789 |
+
"epoch": 272.57142857142856,
|
| 2790 |
+
"eval_accuracy": 0.7464788732394366,
|
| 2791 |
+
"eval_loss": 0.6164286732673645,
|
| 2792 |
+
"eval_runtime": 0.6361,
|
| 2793 |
+
"eval_samples_per_second": 223.244,
|
| 2794 |
+
"eval_steps_per_second": 4.716,
|
| 2795 |
+
"step": 477
|
| 2796 |
+
},
|
| 2797 |
+
{
|
| 2798 |
+
"epoch": 273.7142857142857,
|
| 2799 |
+
"eval_accuracy": 0.7464788732394366,
|
| 2800 |
+
"eval_loss": 0.6160867214202881,
|
| 2801 |
+
"eval_runtime": 0.6616,
|
| 2802 |
+
"eval_samples_per_second": 214.643,
|
| 2803 |
+
"eval_steps_per_second": 4.535,
|
| 2804 |
+
"step": 479
|
| 2805 |
+
},
|
| 2806 |
+
{
|
| 2807 |
+
"epoch": 274.2857142857143,
|
| 2808 |
+
"grad_norm": 9.972301483154297,
|
| 2809 |
+
"learning_rate": 5.000000000000001e-07,
|
| 2810 |
+
"loss": 0.2539,
|
| 2811 |
+
"step": 480
|
| 2812 |
+
},
|
| 2813 |
+
{
|
| 2814 |
+
"epoch": 274.85714285714283,
|
| 2815 |
+
"eval_accuracy": 0.7464788732394366,
|
| 2816 |
+
"eval_loss": 0.6192934513092041,
|
| 2817 |
+
"eval_runtime": 0.694,
|
| 2818 |
+
"eval_samples_per_second": 204.62,
|
| 2819 |
+
"eval_steps_per_second": 4.323,
|
| 2820 |
+
"step": 481
|
| 2821 |
+
},
|
| 2822 |
+
{
|
| 2823 |
+
"epoch": 276.0,
|
| 2824 |
+
"eval_accuracy": 0.7394366197183099,
|
| 2825 |
+
"eval_loss": 0.6235812306404114,
|
| 2826 |
+
"eval_runtime": 0.6625,
|
| 2827 |
+
"eval_samples_per_second": 214.337,
|
| 2828 |
+
"eval_steps_per_second": 4.528,
|
| 2829 |
+
"step": 483
|
| 2830 |
+
},
|
| 2831 |
+
{
|
| 2832 |
+
"epoch": 276.57142857142856,
|
| 2833 |
+
"eval_accuracy": 0.7394366197183099,
|
| 2834 |
+
"eval_loss": 0.6258184909820557,
|
| 2835 |
+
"eval_runtime": 0.6311,
|
| 2836 |
+
"eval_samples_per_second": 224.993,
|
| 2837 |
+
"eval_steps_per_second": 4.753,
|
| 2838 |
+
"step": 484
|
| 2839 |
+
},
|
| 2840 |
+
{
|
| 2841 |
+
"epoch": 277.7142857142857,
|
| 2842 |
+
"eval_accuracy": 0.7394366197183099,
|
| 2843 |
+
"eval_loss": 0.6307951211929321,
|
| 2844 |
+
"eval_runtime": 0.6308,
|
| 2845 |
+
"eval_samples_per_second": 225.107,
|
| 2846 |
+
"eval_steps_per_second": 4.756,
|
| 2847 |
+
"step": 486
|
| 2848 |
+
},
|
| 2849 |
+
{
|
| 2850 |
+
"epoch": 278.85714285714283,
|
| 2851 |
+
"eval_accuracy": 0.7394366197183099,
|
| 2852 |
+
"eval_loss": 0.6349059343338013,
|
| 2853 |
+
"eval_runtime": 0.6686,
|
| 2854 |
+
"eval_samples_per_second": 212.392,
|
| 2855 |
+
"eval_steps_per_second": 4.487,
|
| 2856 |
+
"step": 488
|
| 2857 |
+
},
|
| 2858 |
+
{
|
| 2859 |
+
"epoch": 280.0,
|
| 2860 |
+
"grad_norm": 7.056405067443848,
|
| 2861 |
+
"learning_rate": 2.5000000000000004e-07,
|
| 2862 |
+
"loss": 0.2508,
|
| 2863 |
+
"step": 490
|
| 2864 |
+
},
|
| 2865 |
+
{
|
| 2866 |
+
"epoch": 280.0,
|
| 2867 |
+
"eval_accuracy": 0.7394366197183099,
|
| 2868 |
+
"eval_loss": 0.6352339386940002,
|
| 2869 |
+
"eval_runtime": 0.6086,
|
| 2870 |
+
"eval_samples_per_second": 233.309,
|
| 2871 |
+
"eval_steps_per_second": 4.929,
|
| 2872 |
+
"step": 490
|
| 2873 |
+
},
|
| 2874 |
+
{
|
| 2875 |
+
"epoch": 280.57142857142856,
|
| 2876 |
+
"eval_accuracy": 0.7394366197183099,
|
| 2877 |
+
"eval_loss": 0.6346058249473572,
|
| 2878 |
+
"eval_runtime": 0.6835,
|
| 2879 |
+
"eval_samples_per_second": 207.769,
|
| 2880 |
+
"eval_steps_per_second": 4.389,
|
| 2881 |
+
"step": 491
|
| 2882 |
+
},
|
| 2883 |
+
{
|
| 2884 |
+
"epoch": 281.7142857142857,
|
| 2885 |
+
"eval_accuracy": 0.7394366197183099,
|
| 2886 |
+
"eval_loss": 0.6336230635643005,
|
| 2887 |
+
"eval_runtime": 0.653,
|
| 2888 |
+
"eval_samples_per_second": 217.473,
|
| 2889 |
+
"eval_steps_per_second": 4.595,
|
| 2890 |
+
"step": 493
|
| 2891 |
+
},
|
| 2892 |
+
{
|
| 2893 |
+
"epoch": 282.85714285714283,
|
| 2894 |
+
"eval_accuracy": 0.7394366197183099,
|
| 2895 |
+
"eval_loss": 0.6331290602684021,
|
| 2896 |
+
"eval_runtime": 0.6688,
|
| 2897 |
+
"eval_samples_per_second": 212.305,
|
| 2898 |
+
"eval_steps_per_second": 4.485,
|
| 2899 |
+
"step": 495
|
| 2900 |
+
},
|
| 2901 |
+
{
|
| 2902 |
+
"epoch": 284.0,
|
| 2903 |
+
"eval_accuracy": 0.7394366197183099,
|
| 2904 |
+
"eval_loss": 0.6323853731155396,
|
| 2905 |
+
"eval_runtime": 0.6328,
|
| 2906 |
+
"eval_samples_per_second": 224.389,
|
| 2907 |
+
"eval_steps_per_second": 4.741,
|
| 2908 |
+
"step": 497
|
| 2909 |
+
},
|
| 2910 |
+
{
|
| 2911 |
+
"epoch": 284.57142857142856,
|
| 2912 |
+
"eval_accuracy": 0.7394366197183099,
|
| 2913 |
+
"eval_loss": 0.6318850517272949,
|
| 2914 |
+
"eval_runtime": 0.7105,
|
| 2915 |
+
"eval_samples_per_second": 199.86,
|
| 2916 |
+
"eval_steps_per_second": 4.222,
|
| 2917 |
+
"step": 498
|
| 2918 |
+
},
|
| 2919 |
+
{
|
| 2920 |
+
"epoch": 285.7142857142857,
|
| 2921 |
+
"grad_norm": 8.08874225616455,
|
| 2922 |
+
"learning_rate": 0.0,
|
| 2923 |
+
"loss": 0.2393,
|
| 2924 |
+
"step": 500
|
| 2925 |
+
},
|
| 2926 |
+
{
|
| 2927 |
+
"epoch": 285.7142857142857,
|
| 2928 |
+
"eval_accuracy": 0.7394366197183099,
|
| 2929 |
+
"eval_loss": 0.6315631866455078,
|
| 2930 |
+
"eval_runtime": 0.6449,
|
| 2931 |
+
"eval_samples_per_second": 220.199,
|
| 2932 |
+
"eval_steps_per_second": 4.652,
|
| 2933 |
+
"step": 500
|
| 2934 |
+
},
|
| 2935 |
+
{
|
| 2936 |
+
"epoch": 285.7142857142857,
|
| 2937 |
+
"step": 500,
|
| 2938 |
+
"total_flos": 2.387684396739963e+18,
|
| 2939 |
+
"train_loss": 0.3142274961471558,
|
| 2940 |
+
"train_runtime": 1304.9715,
|
| 2941 |
+
"train_samples_per_second": 163.605,
|
| 2942 |
+
"train_steps_per_second": 0.383
|
| 2943 |
+
}
|
| 2944 |
+
],
|
| 2945 |
+
"logging_steps": 10,
|
| 2946 |
+
"max_steps": 500,
|
| 2947 |
+
"num_input_tokens_seen": 0,
|
| 2948 |
+
"num_train_epochs": 500,
|
| 2949 |
+
"save_steps": 500,
|
| 2950 |
+
"stateful_callbacks": {
|
| 2951 |
+
"TrainerControl": {
|
| 2952 |
+
"args": {
|
| 2953 |
+
"should_epoch_stop": false,
|
| 2954 |
+
"should_evaluate": false,
|
| 2955 |
+
"should_log": false,
|
| 2956 |
+
"should_save": true,
|
| 2957 |
+
"should_training_stop": true
|
| 2958 |
+
},
|
| 2959 |
+
"attributes": {}
|
| 2960 |
+
}
|
| 2961 |
+
},
|
| 2962 |
+
"total_flos": 2.387684396739963e+18,
|
| 2963 |
+
"train_batch_size": 64,
|
| 2964 |
+
"trial_name": null,
|
| 2965 |
+
"trial_params": null
|
| 2966 |
+
}
|