End of training
Browse files- README.md +4 -3
- all_results.json +12 -0
- eval_results.json +8 -0
- runs/Jun22_12-01-17_ml6.hpc.uio.no/events.out.tfevents.1719060593.ml6.hpc.uio.no.238353.1 +3 -0
- train_results.json +7 -0
- trainer_state.json +2158 -0
README.md
CHANGED
|
@@ -1,6 +1,7 @@
|
|
| 1 |
---
|
| 2 |
base_model: anderloh/Hugginhface-master-wav2vec-pretreined-5-class-train-test
|
| 3 |
tags:
|
|
|
|
| 4 |
- generated_from_trainer
|
| 5 |
metrics:
|
| 6 |
- accuracy
|
|
@@ -14,10 +15,10 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 14 |
|
| 15 |
# wav2vec2-5Class-train-test-finetune-V7
|
| 16 |
|
| 17 |
-
This model is a fine-tuned version of [anderloh/Hugginhface-master-wav2vec-pretreined-5-class-train-test](https://huggingface.co/anderloh/Hugginhface-master-wav2vec-pretreined-5-class-train-test) on
|
| 18 |
It achieves the following results on the evaluation set:
|
| 19 |
-
- Loss: 0.
|
| 20 |
-
- Accuracy: 0.
|
| 21 |
|
| 22 |
## Model description
|
| 23 |
|
|
|
|
| 1 |
---
|
| 2 |
base_model: anderloh/Hugginhface-master-wav2vec-pretreined-5-class-train-test
|
| 3 |
tags:
|
| 4 |
+
- audio-classification
|
| 5 |
- generated_from_trainer
|
| 6 |
metrics:
|
| 7 |
- accuracy
|
|
|
|
| 15 |
|
| 16 |
# wav2vec2-5Class-train-test-finetune-V7
|
| 17 |
|
| 18 |
+
This model is a fine-tuned version of [anderloh/Hugginhface-master-wav2vec-pretreined-5-class-train-test](https://huggingface.co/anderloh/Hugginhface-master-wav2vec-pretreined-5-class-train-test) on the anderloh/Master5Class dataset.
|
| 19 |
It achieves the following results on the evaluation set:
|
| 20 |
+
- Loss: 0.9663
|
| 21 |
+
- Accuracy: 0.6538
|
| 22 |
|
| 23 |
## Model description
|
| 24 |
|
all_results.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 230.77,
|
| 3 |
+
"eval_accuracy": 0.6538461538461539,
|
| 4 |
+
"eval_loss": 0.9662555456161499,
|
| 5 |
+
"eval_runtime": 12.2333,
|
| 6 |
+
"eval_samples_per_second": 23.379,
|
| 7 |
+
"eval_steps_per_second": 0.245,
|
| 8 |
+
"train_loss": 0.9558865051269532,
|
| 9 |
+
"train_runtime": 10052.9365,
|
| 10 |
+
"train_samples_per_second": 38.496,
|
| 11 |
+
"train_steps_per_second": 0.075
|
| 12 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 230.77,
|
| 3 |
+
"eval_accuracy": 0.6538461538461539,
|
| 4 |
+
"eval_loss": 0.9662555456161499,
|
| 5 |
+
"eval_runtime": 12.2333,
|
| 6 |
+
"eval_samples_per_second": 23.379,
|
| 7 |
+
"eval_steps_per_second": 0.245
|
| 8 |
+
}
|
runs/Jun22_12-01-17_ml6.hpc.uio.no/events.out.tfevents.1719060593.ml6.hpc.uio.no.238353.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4eddfe22897295ca46493a3faff688e42d69c3a1fb780760dd448767b0d918de
|
| 3 |
+
size 411
|
train_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 230.77,
|
| 3 |
+
"train_loss": 0.9558865051269532,
|
| 4 |
+
"train_runtime": 10052.9365,
|
| 5 |
+
"train_samples_per_second": 38.496,
|
| 6 |
+
"train_steps_per_second": 0.075
|
| 7 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2158 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.6538461538461539,
|
| 3 |
+
"best_model_checkpoint": "wav2vec2-5Class-train-test-finetune-V7/checkpoint-552",
|
| 4 |
+
"epoch": 230.76923076923077,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 750,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.92,
|
| 13 |
+
"eval_accuracy": 0.16083916083916083,
|
| 14 |
+
"eval_loss": 1.6025639772415161,
|
| 15 |
+
"eval_runtime": 25.1775,
|
| 16 |
+
"eval_samples_per_second": 11.359,
|
| 17 |
+
"eval_steps_per_second": 0.119,
|
| 18 |
+
"step": 3
|
| 19 |
+
},
|
| 20 |
+
{
|
| 21 |
+
"epoch": 1.85,
|
| 22 |
+
"eval_accuracy": 0.16083916083916083,
|
| 23 |
+
"eval_loss": 1.6023768186569214,
|
| 24 |
+
"eval_runtime": 13.2244,
|
| 25 |
+
"eval_samples_per_second": 21.627,
|
| 26 |
+
"eval_steps_per_second": 0.227,
|
| 27 |
+
"step": 6
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 2.77,
|
| 31 |
+
"eval_accuracy": 0.16083916083916083,
|
| 32 |
+
"eval_loss": 1.6020615100860596,
|
| 33 |
+
"eval_runtime": 19.1459,
|
| 34 |
+
"eval_samples_per_second": 14.938,
|
| 35 |
+
"eval_steps_per_second": 0.157,
|
| 36 |
+
"step": 9
|
| 37 |
+
},
|
| 38 |
+
{
|
| 39 |
+
"epoch": 4.0,
|
| 40 |
+
"eval_accuracy": 0.16083916083916083,
|
| 41 |
+
"eval_loss": 1.6014316082000732,
|
| 42 |
+
"eval_runtime": 24.6871,
|
| 43 |
+
"eval_samples_per_second": 11.585,
|
| 44 |
+
"eval_steps_per_second": 0.122,
|
| 45 |
+
"step": 13
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 4.92,
|
| 49 |
+
"eval_accuracy": 0.16083916083916083,
|
| 50 |
+
"eval_loss": 1.6007977724075317,
|
| 51 |
+
"eval_runtime": 14.5871,
|
| 52 |
+
"eval_samples_per_second": 19.606,
|
| 53 |
+
"eval_steps_per_second": 0.206,
|
| 54 |
+
"step": 16
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 5.85,
|
| 58 |
+
"eval_accuracy": 0.16083916083916083,
|
| 59 |
+
"eval_loss": 1.600056529045105,
|
| 60 |
+
"eval_runtime": 15.3427,
|
| 61 |
+
"eval_samples_per_second": 18.641,
|
| 62 |
+
"eval_steps_per_second": 0.196,
|
| 63 |
+
"step": 19
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 6.77,
|
| 67 |
+
"eval_accuracy": 0.16083916083916083,
|
| 68 |
+
"eval_loss": 1.599173903465271,
|
| 69 |
+
"eval_runtime": 23.1969,
|
| 70 |
+
"eval_samples_per_second": 12.329,
|
| 71 |
+
"eval_steps_per_second": 0.129,
|
| 72 |
+
"step": 22
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 8.0,
|
| 76 |
+
"eval_accuracy": 0.17482517482517482,
|
| 77 |
+
"eval_loss": 1.5977518558502197,
|
| 78 |
+
"eval_runtime": 24.249,
|
| 79 |
+
"eval_samples_per_second": 11.794,
|
| 80 |
+
"eval_steps_per_second": 0.124,
|
| 81 |
+
"step": 26
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 8.92,
|
| 85 |
+
"eval_accuracy": 0.1888111888111888,
|
| 86 |
+
"eval_loss": 1.5965358018875122,
|
| 87 |
+
"eval_runtime": 14.7479,
|
| 88 |
+
"eval_samples_per_second": 19.393,
|
| 89 |
+
"eval_steps_per_second": 0.203,
|
| 90 |
+
"step": 29
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 9.85,
|
| 94 |
+
"eval_accuracy": 0.2097902097902098,
|
| 95 |
+
"eval_loss": 1.5952231884002686,
|
| 96 |
+
"eval_runtime": 18.4251,
|
| 97 |
+
"eval_samples_per_second": 15.522,
|
| 98 |
+
"eval_steps_per_second": 0.163,
|
| 99 |
+
"step": 32
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 10.77,
|
| 103 |
+
"eval_accuracy": 0.22727272727272727,
|
| 104 |
+
"eval_loss": 1.593780517578125,
|
| 105 |
+
"eval_runtime": 20.5455,
|
| 106 |
+
"eval_samples_per_second": 13.92,
|
| 107 |
+
"eval_steps_per_second": 0.146,
|
| 108 |
+
"step": 35
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 12.0,
|
| 112 |
+
"eval_accuracy": 0.23426573426573427,
|
| 113 |
+
"eval_loss": 1.5916367769241333,
|
| 114 |
+
"eval_runtime": 16.0385,
|
| 115 |
+
"eval_samples_per_second": 17.832,
|
| 116 |
+
"eval_steps_per_second": 0.187,
|
| 117 |
+
"step": 39
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 12.92,
|
| 121 |
+
"eval_accuracy": 0.2692307692307692,
|
| 122 |
+
"eval_loss": 1.5898847579956055,
|
| 123 |
+
"eval_runtime": 21.1768,
|
| 124 |
+
"eval_samples_per_second": 13.505,
|
| 125 |
+
"eval_steps_per_second": 0.142,
|
| 126 |
+
"step": 42
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 13.85,
|
| 130 |
+
"eval_accuracy": 0.2727272727272727,
|
| 131 |
+
"eval_loss": 1.5879966020584106,
|
| 132 |
+
"eval_runtime": 13.5248,
|
| 133 |
+
"eval_samples_per_second": 21.146,
|
| 134 |
+
"eval_steps_per_second": 0.222,
|
| 135 |
+
"step": 45
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 14.77,
|
| 139 |
+
"eval_accuracy": 0.3076923076923077,
|
| 140 |
+
"eval_loss": 1.5860111713409424,
|
| 141 |
+
"eval_runtime": 12.9444,
|
| 142 |
+
"eval_samples_per_second": 22.094,
|
| 143 |
+
"eval_steps_per_second": 0.232,
|
| 144 |
+
"step": 48
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 16.0,
|
| 148 |
+
"eval_accuracy": 0.35664335664335667,
|
| 149 |
+
"eval_loss": 1.583256483078003,
|
| 150 |
+
"eval_runtime": 25.0161,
|
| 151 |
+
"eval_samples_per_second": 11.433,
|
| 152 |
+
"eval_steps_per_second": 0.12,
|
| 153 |
+
"step": 52
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 16.92,
|
| 157 |
+
"eval_accuracy": 0.3881118881118881,
|
| 158 |
+
"eval_loss": 1.5810612440109253,
|
| 159 |
+
"eval_runtime": 14.5666,
|
| 160 |
+
"eval_samples_per_second": 19.634,
|
| 161 |
+
"eval_steps_per_second": 0.206,
|
| 162 |
+
"step": 55
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 17.85,
|
| 166 |
+
"eval_accuracy": 0.3811188811188811,
|
| 167 |
+
"eval_loss": 1.578792929649353,
|
| 168 |
+
"eval_runtime": 13.0625,
|
| 169 |
+
"eval_samples_per_second": 21.895,
|
| 170 |
+
"eval_steps_per_second": 0.23,
|
| 171 |
+
"step": 58
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 18.77,
|
| 175 |
+
"eval_accuracy": 0.36713286713286714,
|
| 176 |
+
"eval_loss": 1.576446294784546,
|
| 177 |
+
"eval_runtime": 18.6648,
|
| 178 |
+
"eval_samples_per_second": 15.323,
|
| 179 |
+
"eval_steps_per_second": 0.161,
|
| 180 |
+
"step": 61
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 20.0,
|
| 184 |
+
"eval_accuracy": 0.34965034965034963,
|
| 185 |
+
"eval_loss": 1.5730595588684082,
|
| 186 |
+
"eval_runtime": 11.5386,
|
| 187 |
+
"eval_samples_per_second": 24.786,
|
| 188 |
+
"eval_steps_per_second": 0.26,
|
| 189 |
+
"step": 65
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 20.92,
|
| 193 |
+
"eval_accuracy": 0.32867132867132864,
|
| 194 |
+
"eval_loss": 1.5702308416366577,
|
| 195 |
+
"eval_runtime": 15.7041,
|
| 196 |
+
"eval_samples_per_second": 18.212,
|
| 197 |
+
"eval_steps_per_second": 0.191,
|
| 198 |
+
"step": 68
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 21.85,
|
| 202 |
+
"eval_accuracy": 0.32517482517482516,
|
| 203 |
+
"eval_loss": 1.5671954154968262,
|
| 204 |
+
"eval_runtime": 11.3961,
|
| 205 |
+
"eval_samples_per_second": 25.096,
|
| 206 |
+
"eval_steps_per_second": 0.263,
|
| 207 |
+
"step": 71
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 22.77,
|
| 211 |
+
"eval_accuracy": 0.3146853146853147,
|
| 212 |
+
"eval_loss": 1.5641109943389893,
|
| 213 |
+
"eval_runtime": 15.1673,
|
| 214 |
+
"eval_samples_per_second": 18.856,
|
| 215 |
+
"eval_steps_per_second": 0.198,
|
| 216 |
+
"step": 74
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 24.0,
|
| 220 |
+
"eval_accuracy": 0.3111888111888112,
|
| 221 |
+
"eval_loss": 1.5597317218780518,
|
| 222 |
+
"eval_runtime": 24.7249,
|
| 223 |
+
"eval_samples_per_second": 11.567,
|
| 224 |
+
"eval_steps_per_second": 0.121,
|
| 225 |
+
"step": 78
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 24.92,
|
| 229 |
+
"eval_accuracy": 0.3076923076923077,
|
| 230 |
+
"eval_loss": 1.5563873052597046,
|
| 231 |
+
"eval_runtime": 13.1713,
|
| 232 |
+
"eval_samples_per_second": 21.714,
|
| 233 |
+
"eval_steps_per_second": 0.228,
|
| 234 |
+
"step": 81
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 25.85,
|
| 238 |
+
"eval_accuracy": 0.3041958041958042,
|
| 239 |
+
"eval_loss": 1.553160548210144,
|
| 240 |
+
"eval_runtime": 12.194,
|
| 241 |
+
"eval_samples_per_second": 23.454,
|
| 242 |
+
"eval_steps_per_second": 0.246,
|
| 243 |
+
"step": 84
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 26.77,
|
| 247 |
+
"eval_accuracy": 0.2937062937062937,
|
| 248 |
+
"eval_loss": 1.549930214881897,
|
| 249 |
+
"eval_runtime": 25.3259,
|
| 250 |
+
"eval_samples_per_second": 11.293,
|
| 251 |
+
"eval_steps_per_second": 0.118,
|
| 252 |
+
"step": 87
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 28.0,
|
| 256 |
+
"eval_accuracy": 0.2902097902097902,
|
| 257 |
+
"eval_loss": 1.5454081296920776,
|
| 258 |
+
"eval_runtime": 10.4422,
|
| 259 |
+
"eval_samples_per_second": 27.389,
|
| 260 |
+
"eval_steps_per_second": 0.287,
|
| 261 |
+
"step": 91
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 28.92,
|
| 265 |
+
"eval_accuracy": 0.2867132867132867,
|
| 266 |
+
"eval_loss": 1.5419245958328247,
|
| 267 |
+
"eval_runtime": 16.2077,
|
| 268 |
+
"eval_samples_per_second": 17.646,
|
| 269 |
+
"eval_steps_per_second": 0.185,
|
| 270 |
+
"step": 94
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 29.85,
|
| 274 |
+
"eval_accuracy": 0.28321678321678323,
|
| 275 |
+
"eval_loss": 1.5383468866348267,
|
| 276 |
+
"eval_runtime": 27.5909,
|
| 277 |
+
"eval_samples_per_second": 10.366,
|
| 278 |
+
"eval_steps_per_second": 0.109,
|
| 279 |
+
"step": 97
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 30.77,
|
| 283 |
+
"grad_norm": 65491.41796875,
|
| 284 |
+
"learning_rate": 2.8888888888888888e-05,
|
| 285 |
+
"loss": 1.5563,
|
| 286 |
+
"step": 100
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 30.77,
|
| 290 |
+
"eval_accuracy": 0.2762237762237762,
|
| 291 |
+
"eval_loss": 1.5348902940750122,
|
| 292 |
+
"eval_runtime": 14.2201,
|
| 293 |
+
"eval_samples_per_second": 20.112,
|
| 294 |
+
"eval_steps_per_second": 0.211,
|
| 295 |
+
"step": 100
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 32.0,
|
| 299 |
+
"eval_accuracy": 0.27972027972027974,
|
| 300 |
+
"eval_loss": 1.5304443836212158,
|
| 301 |
+
"eval_runtime": 11.5662,
|
| 302 |
+
"eval_samples_per_second": 24.727,
|
| 303 |
+
"eval_steps_per_second": 0.259,
|
| 304 |
+
"step": 104
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 32.92,
|
| 308 |
+
"eval_accuracy": 0.2762237762237762,
|
| 309 |
+
"eval_loss": 1.5273348093032837,
|
| 310 |
+
"eval_runtime": 10.0608,
|
| 311 |
+
"eval_samples_per_second": 28.427,
|
| 312 |
+
"eval_steps_per_second": 0.298,
|
| 313 |
+
"step": 107
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 33.85,
|
| 317 |
+
"eval_accuracy": 0.26573426573426573,
|
| 318 |
+
"eval_loss": 1.524675965309143,
|
| 319 |
+
"eval_runtime": 16.3502,
|
| 320 |
+
"eval_samples_per_second": 17.492,
|
| 321 |
+
"eval_steps_per_second": 0.183,
|
| 322 |
+
"step": 110
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 34.77,
|
| 326 |
+
"eval_accuracy": 0.2517482517482518,
|
| 327 |
+
"eval_loss": 1.5223478078842163,
|
| 328 |
+
"eval_runtime": 11.4239,
|
| 329 |
+
"eval_samples_per_second": 25.035,
|
| 330 |
+
"eval_steps_per_second": 0.263,
|
| 331 |
+
"step": 113
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 36.0,
|
| 335 |
+
"eval_accuracy": 0.24825174825174826,
|
| 336 |
+
"eval_loss": 1.5194127559661865,
|
| 337 |
+
"eval_runtime": 12.5757,
|
| 338 |
+
"eval_samples_per_second": 22.742,
|
| 339 |
+
"eval_steps_per_second": 0.239,
|
| 340 |
+
"step": 117
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 36.92,
|
| 344 |
+
"eval_accuracy": 0.24125874125874125,
|
| 345 |
+
"eval_loss": 1.5177949666976929,
|
| 346 |
+
"eval_runtime": 29.9608,
|
| 347 |
+
"eval_samples_per_second": 9.546,
|
| 348 |
+
"eval_steps_per_second": 0.1,
|
| 349 |
+
"step": 120
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 37.85,
|
| 353 |
+
"eval_accuracy": 0.23776223776223776,
|
| 354 |
+
"eval_loss": 1.5167657136917114,
|
| 355 |
+
"eval_runtime": 11.3556,
|
| 356 |
+
"eval_samples_per_second": 25.186,
|
| 357 |
+
"eval_steps_per_second": 0.264,
|
| 358 |
+
"step": 123
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 38.77,
|
| 362 |
+
"eval_accuracy": 0.24475524475524477,
|
| 363 |
+
"eval_loss": 1.5161939859390259,
|
| 364 |
+
"eval_runtime": 8.3481,
|
| 365 |
+
"eval_samples_per_second": 34.259,
|
| 366 |
+
"eval_steps_per_second": 0.359,
|
| 367 |
+
"step": 126
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 40.0,
|
| 371 |
+
"eval_accuracy": 0.24475524475524477,
|
| 372 |
+
"eval_loss": 1.5161994695663452,
|
| 373 |
+
"eval_runtime": 11.5276,
|
| 374 |
+
"eval_samples_per_second": 24.81,
|
| 375 |
+
"eval_steps_per_second": 0.26,
|
| 376 |
+
"step": 130
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 40.92,
|
| 380 |
+
"eval_accuracy": 0.24825174825174826,
|
| 381 |
+
"eval_loss": 1.5166517496109009,
|
| 382 |
+
"eval_runtime": 16.6347,
|
| 383 |
+
"eval_samples_per_second": 17.193,
|
| 384 |
+
"eval_steps_per_second": 0.18,
|
| 385 |
+
"step": 133
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 41.85,
|
| 389 |
+
"eval_accuracy": 0.24825174825174826,
|
| 390 |
+
"eval_loss": 1.518052101135254,
|
| 391 |
+
"eval_runtime": 9.0049,
|
| 392 |
+
"eval_samples_per_second": 31.76,
|
| 393 |
+
"eval_steps_per_second": 0.333,
|
| 394 |
+
"step": 136
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 42.77,
|
| 398 |
+
"eval_accuracy": 0.25874125874125875,
|
| 399 |
+
"eval_loss": 1.520302414894104,
|
| 400 |
+
"eval_runtime": 9.9294,
|
| 401 |
+
"eval_samples_per_second": 28.803,
|
| 402 |
+
"eval_steps_per_second": 0.302,
|
| 403 |
+
"step": 139
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 44.0,
|
| 407 |
+
"eval_accuracy": 0.2692307692307692,
|
| 408 |
+
"eval_loss": 1.5227454900741577,
|
| 409 |
+
"eval_runtime": 11.2365,
|
| 410 |
+
"eval_samples_per_second": 25.453,
|
| 411 |
+
"eval_steps_per_second": 0.267,
|
| 412 |
+
"step": 143
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 44.92,
|
| 416 |
+
"eval_accuracy": 0.28321678321678323,
|
| 417 |
+
"eval_loss": 1.5243264436721802,
|
| 418 |
+
"eval_runtime": 19.5213,
|
| 419 |
+
"eval_samples_per_second": 14.651,
|
| 420 |
+
"eval_steps_per_second": 0.154,
|
| 421 |
+
"step": 146
|
| 422 |
+
},
|
| 423 |
+
{
|
| 424 |
+
"epoch": 45.85,
|
| 425 |
+
"eval_accuracy": 0.27972027972027974,
|
| 426 |
+
"eval_loss": 1.5238893032073975,
|
| 427 |
+
"eval_runtime": 10.1208,
|
| 428 |
+
"eval_samples_per_second": 28.259,
|
| 429 |
+
"eval_steps_per_second": 0.296,
|
| 430 |
+
"step": 149
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 46.77,
|
| 434 |
+
"eval_accuracy": 0.3006993006993007,
|
| 435 |
+
"eval_loss": 1.5224095582962036,
|
| 436 |
+
"eval_runtime": 11.1479,
|
| 437 |
+
"eval_samples_per_second": 25.655,
|
| 438 |
+
"eval_steps_per_second": 0.269,
|
| 439 |
+
"step": 152
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"epoch": 48.0,
|
| 443 |
+
"eval_accuracy": 0.3076923076923077,
|
| 444 |
+
"eval_loss": 1.5170344114303589,
|
| 445 |
+
"eval_runtime": 23.385,
|
| 446 |
+
"eval_samples_per_second": 12.23,
|
| 447 |
+
"eval_steps_per_second": 0.128,
|
| 448 |
+
"step": 156
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 48.92,
|
| 452 |
+
"eval_accuracy": 0.32867132867132864,
|
| 453 |
+
"eval_loss": 1.5102859735488892,
|
| 454 |
+
"eval_runtime": 13.363,
|
| 455 |
+
"eval_samples_per_second": 21.402,
|
| 456 |
+
"eval_steps_per_second": 0.225,
|
| 457 |
+
"step": 159
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 49.85,
|
| 461 |
+
"eval_accuracy": 0.34965034965034963,
|
| 462 |
+
"eval_loss": 1.5032000541687012,
|
| 463 |
+
"eval_runtime": 10.5466,
|
| 464 |
+
"eval_samples_per_second": 27.118,
|
| 465 |
+
"eval_steps_per_second": 0.284,
|
| 466 |
+
"step": 162
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 50.77,
|
| 470 |
+
"eval_accuracy": 0.36013986013986016,
|
| 471 |
+
"eval_loss": 1.4958622455596924,
|
| 472 |
+
"eval_runtime": 13.1319,
|
| 473 |
+
"eval_samples_per_second": 21.779,
|
| 474 |
+
"eval_steps_per_second": 0.228,
|
| 475 |
+
"step": 165
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 52.0,
|
| 479 |
+
"eval_accuracy": 0.36363636363636365,
|
| 480 |
+
"eval_loss": 1.4857254028320312,
|
| 481 |
+
"eval_runtime": 11.3199,
|
| 482 |
+
"eval_samples_per_second": 25.265,
|
| 483 |
+
"eval_steps_per_second": 0.265,
|
| 484 |
+
"step": 169
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 52.92,
|
| 488 |
+
"eval_accuracy": 0.36713286713286714,
|
| 489 |
+
"eval_loss": 1.4788074493408203,
|
| 490 |
+
"eval_runtime": 10.1044,
|
| 491 |
+
"eval_samples_per_second": 28.304,
|
| 492 |
+
"eval_steps_per_second": 0.297,
|
| 493 |
+
"step": 172
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 53.85,
|
| 497 |
+
"eval_accuracy": 0.3741258741258741,
|
| 498 |
+
"eval_loss": 1.4713162183761597,
|
| 499 |
+
"eval_runtime": 21.314,
|
| 500 |
+
"eval_samples_per_second": 13.418,
|
| 501 |
+
"eval_steps_per_second": 0.141,
|
| 502 |
+
"step": 175
|
| 503 |
+
},
|
| 504 |
+
{
|
| 505 |
+
"epoch": 54.77,
|
| 506 |
+
"eval_accuracy": 0.3811188811188811,
|
| 507 |
+
"eval_loss": 1.4641575813293457,
|
| 508 |
+
"eval_runtime": 11.6546,
|
| 509 |
+
"eval_samples_per_second": 24.54,
|
| 510 |
+
"eval_steps_per_second": 0.257,
|
| 511 |
+
"step": 178
|
| 512 |
+
},
|
| 513 |
+
{
|
| 514 |
+
"epoch": 56.0,
|
| 515 |
+
"eval_accuracy": 0.3881118881118881,
|
| 516 |
+
"eval_loss": 1.455277442932129,
|
| 517 |
+
"eval_runtime": 11.9334,
|
| 518 |
+
"eval_samples_per_second": 23.966,
|
| 519 |
+
"eval_steps_per_second": 0.251,
|
| 520 |
+
"step": 182
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 56.92,
|
| 524 |
+
"eval_accuracy": 0.3986013986013986,
|
| 525 |
+
"eval_loss": 1.4481216669082642,
|
| 526 |
+
"eval_runtime": 10.9879,
|
| 527 |
+
"eval_samples_per_second": 26.029,
|
| 528 |
+
"eval_steps_per_second": 0.273,
|
| 529 |
+
"step": 185
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 57.85,
|
| 533 |
+
"eval_accuracy": 0.4020979020979021,
|
| 534 |
+
"eval_loss": 1.4420974254608154,
|
| 535 |
+
"eval_runtime": 8.8556,
|
| 536 |
+
"eval_samples_per_second": 32.296,
|
| 537 |
+
"eval_steps_per_second": 0.339,
|
| 538 |
+
"step": 188
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 58.77,
|
| 542 |
+
"eval_accuracy": 0.4125874125874126,
|
| 543 |
+
"eval_loss": 1.4357017278671265,
|
| 544 |
+
"eval_runtime": 9.9507,
|
| 545 |
+
"eval_samples_per_second": 28.742,
|
| 546 |
+
"eval_steps_per_second": 0.301,
|
| 547 |
+
"step": 191
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 60.0,
|
| 551 |
+
"eval_accuracy": 0.4195804195804196,
|
| 552 |
+
"eval_loss": 1.4284459352493286,
|
| 553 |
+
"eval_runtime": 10.9332,
|
| 554 |
+
"eval_samples_per_second": 26.159,
|
| 555 |
+
"eval_steps_per_second": 0.274,
|
| 556 |
+
"step": 195
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 60.92,
|
| 560 |
+
"eval_accuracy": 0.4195804195804196,
|
| 561 |
+
"eval_loss": 1.4218381643295288,
|
| 562 |
+
"eval_runtime": 15.3761,
|
| 563 |
+
"eval_samples_per_second": 18.6,
|
| 564 |
+
"eval_steps_per_second": 0.195,
|
| 565 |
+
"step": 198
|
| 566 |
+
},
|
| 567 |
+
{
|
| 568 |
+
"epoch": 61.54,
|
| 569 |
+
"grad_norm": 26744.056640625,
|
| 570 |
+
"learning_rate": 2.4444444444444445e-05,
|
| 571 |
+
"loss": 1.3138,
|
| 572 |
+
"step": 200
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 61.85,
|
| 576 |
+
"eval_accuracy": 0.43006993006993005,
|
| 577 |
+
"eval_loss": 1.4166817665100098,
|
| 578 |
+
"eval_runtime": 13.1906,
|
| 579 |
+
"eval_samples_per_second": 21.682,
|
| 580 |
+
"eval_steps_per_second": 0.227,
|
| 581 |
+
"step": 201
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 62.77,
|
| 585 |
+
"eval_accuracy": 0.43006993006993005,
|
| 586 |
+
"eval_loss": 1.409144639968872,
|
| 587 |
+
"eval_runtime": 21.121,
|
| 588 |
+
"eval_samples_per_second": 13.541,
|
| 589 |
+
"eval_steps_per_second": 0.142,
|
| 590 |
+
"step": 204
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 64.0,
|
| 594 |
+
"eval_accuracy": 0.4370629370629371,
|
| 595 |
+
"eval_loss": 1.3994969129562378,
|
| 596 |
+
"eval_runtime": 11.3707,
|
| 597 |
+
"eval_samples_per_second": 25.152,
|
| 598 |
+
"eval_steps_per_second": 0.264,
|
| 599 |
+
"step": 208
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 64.92,
|
| 603 |
+
"eval_accuracy": 0.4405594405594406,
|
| 604 |
+
"eval_loss": 1.3911248445510864,
|
| 605 |
+
"eval_runtime": 15.8386,
|
| 606 |
+
"eval_samples_per_second": 18.057,
|
| 607 |
+
"eval_steps_per_second": 0.189,
|
| 608 |
+
"step": 211
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 65.85,
|
| 612 |
+
"eval_accuracy": 0.43356643356643354,
|
| 613 |
+
"eval_loss": 1.382477045059204,
|
| 614 |
+
"eval_runtime": 15.9417,
|
| 615 |
+
"eval_samples_per_second": 17.94,
|
| 616 |
+
"eval_steps_per_second": 0.188,
|
| 617 |
+
"step": 214
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 66.77,
|
| 621 |
+
"eval_accuracy": 0.44405594405594406,
|
| 622 |
+
"eval_loss": 1.373460054397583,
|
| 623 |
+
"eval_runtime": 22.8178,
|
| 624 |
+
"eval_samples_per_second": 12.534,
|
| 625 |
+
"eval_steps_per_second": 0.131,
|
| 626 |
+
"step": 217
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 68.0,
|
| 630 |
+
"eval_accuracy": 0.44755244755244755,
|
| 631 |
+
"eval_loss": 1.3632004261016846,
|
| 632 |
+
"eval_runtime": 8.8406,
|
| 633 |
+
"eval_samples_per_second": 32.351,
|
| 634 |
+
"eval_steps_per_second": 0.339,
|
| 635 |
+
"step": 221
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 68.92,
|
| 639 |
+
"eval_accuracy": 0.45104895104895104,
|
| 640 |
+
"eval_loss": 1.355635166168213,
|
| 641 |
+
"eval_runtime": 22.4517,
|
| 642 |
+
"eval_samples_per_second": 12.738,
|
| 643 |
+
"eval_steps_per_second": 0.134,
|
| 644 |
+
"step": 224
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 69.85,
|
| 648 |
+
"eval_accuracy": 0.45104895104895104,
|
| 649 |
+
"eval_loss": 1.349207878112793,
|
| 650 |
+
"eval_runtime": 19.129,
|
| 651 |
+
"eval_samples_per_second": 14.951,
|
| 652 |
+
"eval_steps_per_second": 0.157,
|
| 653 |
+
"step": 227
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 70.77,
|
| 657 |
+
"eval_accuracy": 0.45104895104895104,
|
| 658 |
+
"eval_loss": 1.3441168069839478,
|
| 659 |
+
"eval_runtime": 10.9281,
|
| 660 |
+
"eval_samples_per_second": 26.171,
|
| 661 |
+
"eval_steps_per_second": 0.275,
|
| 662 |
+
"step": 230
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 72.0,
|
| 666 |
+
"eval_accuracy": 0.458041958041958,
|
| 667 |
+
"eval_loss": 1.3352017402648926,
|
| 668 |
+
"eval_runtime": 9.8412,
|
| 669 |
+
"eval_samples_per_second": 29.062,
|
| 670 |
+
"eval_steps_per_second": 0.305,
|
| 671 |
+
"step": 234
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 72.92,
|
| 675 |
+
"eval_accuracy": 0.46153846153846156,
|
| 676 |
+
"eval_loss": 1.326931357383728,
|
| 677 |
+
"eval_runtime": 9.1291,
|
| 678 |
+
"eval_samples_per_second": 31.329,
|
| 679 |
+
"eval_steps_per_second": 0.329,
|
| 680 |
+
"step": 237
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 73.85,
|
| 684 |
+
"eval_accuracy": 0.4755244755244755,
|
| 685 |
+
"eval_loss": 1.3185527324676514,
|
| 686 |
+
"eval_runtime": 8.2766,
|
| 687 |
+
"eval_samples_per_second": 34.555,
|
| 688 |
+
"eval_steps_per_second": 0.362,
|
| 689 |
+
"step": 240
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 74.77,
|
| 693 |
+
"eval_accuracy": 0.4755244755244755,
|
| 694 |
+
"eval_loss": 1.3105268478393555,
|
| 695 |
+
"eval_runtime": 11.5807,
|
| 696 |
+
"eval_samples_per_second": 24.696,
|
| 697 |
+
"eval_steps_per_second": 0.259,
|
| 698 |
+
"step": 243
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 76.0,
|
| 702 |
+
"eval_accuracy": 0.479020979020979,
|
| 703 |
+
"eval_loss": 1.299223780632019,
|
| 704 |
+
"eval_runtime": 26.8999,
|
| 705 |
+
"eval_samples_per_second": 10.632,
|
| 706 |
+
"eval_steps_per_second": 0.112,
|
| 707 |
+
"step": 247
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 76.92,
|
| 711 |
+
"eval_accuracy": 0.4825174825174825,
|
| 712 |
+
"eval_loss": 1.2896299362182617,
|
| 713 |
+
"eval_runtime": 22.9313,
|
| 714 |
+
"eval_samples_per_second": 12.472,
|
| 715 |
+
"eval_steps_per_second": 0.131,
|
| 716 |
+
"step": 250
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 77.85,
|
| 720 |
+
"eval_accuracy": 0.4825174825174825,
|
| 721 |
+
"eval_loss": 1.2797133922576904,
|
| 722 |
+
"eval_runtime": 12.0479,
|
| 723 |
+
"eval_samples_per_second": 23.739,
|
| 724 |
+
"eval_steps_per_second": 0.249,
|
| 725 |
+
"step": 253
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 78.77,
|
| 729 |
+
"eval_accuracy": 0.486013986013986,
|
| 730 |
+
"eval_loss": 1.270691156387329,
|
| 731 |
+
"eval_runtime": 21.8623,
|
| 732 |
+
"eval_samples_per_second": 13.082,
|
| 733 |
+
"eval_steps_per_second": 0.137,
|
| 734 |
+
"step": 256
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 80.0,
|
| 738 |
+
"eval_accuracy": 0.493006993006993,
|
| 739 |
+
"eval_loss": 1.2586532831192017,
|
| 740 |
+
"eval_runtime": 21.4963,
|
| 741 |
+
"eval_samples_per_second": 13.305,
|
| 742 |
+
"eval_steps_per_second": 0.14,
|
| 743 |
+
"step": 260
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 80.92,
|
| 747 |
+
"eval_accuracy": 0.493006993006993,
|
| 748 |
+
"eval_loss": 1.2493596076965332,
|
| 749 |
+
"eval_runtime": 16.0914,
|
| 750 |
+
"eval_samples_per_second": 17.774,
|
| 751 |
+
"eval_steps_per_second": 0.186,
|
| 752 |
+
"step": 263
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 81.85,
|
| 756 |
+
"eval_accuracy": 0.493006993006993,
|
| 757 |
+
"eval_loss": 1.2407375574111938,
|
| 758 |
+
"eval_runtime": 16.0788,
|
| 759 |
+
"eval_samples_per_second": 17.787,
|
| 760 |
+
"eval_steps_per_second": 0.187,
|
| 761 |
+
"step": 266
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 82.77,
|
| 765 |
+
"eval_accuracy": 0.5104895104895105,
|
| 766 |
+
"eval_loss": 1.2313759326934814,
|
| 767 |
+
"eval_runtime": 13.8028,
|
| 768 |
+
"eval_samples_per_second": 20.72,
|
| 769 |
+
"eval_steps_per_second": 0.217,
|
| 770 |
+
"step": 269
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 84.0,
|
| 774 |
+
"eval_accuracy": 0.513986013986014,
|
| 775 |
+
"eval_loss": 1.2205084562301636,
|
| 776 |
+
"eval_runtime": 15.155,
|
| 777 |
+
"eval_samples_per_second": 18.872,
|
| 778 |
+
"eval_steps_per_second": 0.198,
|
| 779 |
+
"step": 273
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 84.92,
|
| 783 |
+
"eval_accuracy": 0.5209790209790209,
|
| 784 |
+
"eval_loss": 1.2124475240707397,
|
| 785 |
+
"eval_runtime": 24.3109,
|
| 786 |
+
"eval_samples_per_second": 11.764,
|
| 787 |
+
"eval_steps_per_second": 0.123,
|
| 788 |
+
"step": 276
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 85.85,
|
| 792 |
+
"eval_accuracy": 0.5314685314685315,
|
| 793 |
+
"eval_loss": 1.2043241262435913,
|
| 794 |
+
"eval_runtime": 12.0961,
|
| 795 |
+
"eval_samples_per_second": 23.644,
|
| 796 |
+
"eval_steps_per_second": 0.248,
|
| 797 |
+
"step": 279
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 86.77,
|
| 801 |
+
"eval_accuracy": 0.534965034965035,
|
| 802 |
+
"eval_loss": 1.1973276138305664,
|
| 803 |
+
"eval_runtime": 18.6151,
|
| 804 |
+
"eval_samples_per_second": 15.364,
|
| 805 |
+
"eval_steps_per_second": 0.161,
|
| 806 |
+
"step": 282
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 88.0,
|
| 810 |
+
"eval_accuracy": 0.5524475524475524,
|
| 811 |
+
"eval_loss": 1.1870229244232178,
|
| 812 |
+
"eval_runtime": 19.514,
|
| 813 |
+
"eval_samples_per_second": 14.656,
|
| 814 |
+
"eval_steps_per_second": 0.154,
|
| 815 |
+
"step": 286
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 88.92,
|
| 819 |
+
"eval_accuracy": 0.5629370629370629,
|
| 820 |
+
"eval_loss": 1.178816318511963,
|
| 821 |
+
"eval_runtime": 17.8959,
|
| 822 |
+
"eval_samples_per_second": 15.981,
|
| 823 |
+
"eval_steps_per_second": 0.168,
|
| 824 |
+
"step": 289
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 89.85,
|
| 828 |
+
"eval_accuracy": 0.5629370629370629,
|
| 829 |
+
"eval_loss": 1.1699650287628174,
|
| 830 |
+
"eval_runtime": 14.6752,
|
| 831 |
+
"eval_samples_per_second": 19.489,
|
| 832 |
+
"eval_steps_per_second": 0.204,
|
| 833 |
+
"step": 292
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 90.77,
|
| 837 |
+
"eval_accuracy": 0.5699300699300699,
|
| 838 |
+
"eval_loss": 1.1613417863845825,
|
| 839 |
+
"eval_runtime": 23.2377,
|
| 840 |
+
"eval_samples_per_second": 12.308,
|
| 841 |
+
"eval_steps_per_second": 0.129,
|
| 842 |
+
"step": 295
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 92.0,
|
| 846 |
+
"eval_accuracy": 0.583916083916084,
|
| 847 |
+
"eval_loss": 1.1498184204101562,
|
| 848 |
+
"eval_runtime": 16.8037,
|
| 849 |
+
"eval_samples_per_second": 17.02,
|
| 850 |
+
"eval_steps_per_second": 0.179,
|
| 851 |
+
"step": 299
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 92.31,
|
| 855 |
+
"grad_norm": 27893.916015625,
|
| 856 |
+
"learning_rate": 1.9999999999999998e-05,
|
| 857 |
+
"loss": 1.047,
|
| 858 |
+
"step": 300
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 92.92,
|
| 862 |
+
"eval_accuracy": 0.5874125874125874,
|
| 863 |
+
"eval_loss": 1.1410824060440063,
|
| 864 |
+
"eval_runtime": 17.7012,
|
| 865 |
+
"eval_samples_per_second": 16.157,
|
| 866 |
+
"eval_steps_per_second": 0.169,
|
| 867 |
+
"step": 302
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 93.85,
|
| 871 |
+
"eval_accuracy": 0.5944055944055944,
|
| 872 |
+
"eval_loss": 1.133009433746338,
|
| 873 |
+
"eval_runtime": 21.3624,
|
| 874 |
+
"eval_samples_per_second": 13.388,
|
| 875 |
+
"eval_steps_per_second": 0.14,
|
| 876 |
+
"step": 305
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 94.77,
|
| 880 |
+
"eval_accuracy": 0.5944055944055944,
|
| 881 |
+
"eval_loss": 1.1261355876922607,
|
| 882 |
+
"eval_runtime": 12.5743,
|
| 883 |
+
"eval_samples_per_second": 22.745,
|
| 884 |
+
"eval_steps_per_second": 0.239,
|
| 885 |
+
"step": 308
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 96.0,
|
| 889 |
+
"eval_accuracy": 0.6013986013986014,
|
| 890 |
+
"eval_loss": 1.1161199808120728,
|
| 891 |
+
"eval_runtime": 10.714,
|
| 892 |
+
"eval_samples_per_second": 26.694,
|
| 893 |
+
"eval_steps_per_second": 0.28,
|
| 894 |
+
"step": 312
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 96.92,
|
| 898 |
+
"eval_accuracy": 0.6013986013986014,
|
| 899 |
+
"eval_loss": 1.1083568334579468,
|
| 900 |
+
"eval_runtime": 23.7041,
|
| 901 |
+
"eval_samples_per_second": 12.065,
|
| 902 |
+
"eval_steps_per_second": 0.127,
|
| 903 |
+
"step": 315
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 97.85,
|
| 907 |
+
"eval_accuracy": 0.6048951048951049,
|
| 908 |
+
"eval_loss": 1.1002540588378906,
|
| 909 |
+
"eval_runtime": 18.2723,
|
| 910 |
+
"eval_samples_per_second": 15.652,
|
| 911 |
+
"eval_steps_per_second": 0.164,
|
| 912 |
+
"step": 318
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 98.77,
|
| 916 |
+
"eval_accuracy": 0.6048951048951049,
|
| 917 |
+
"eval_loss": 1.0926333665847778,
|
| 918 |
+
"eval_runtime": 15.4521,
|
| 919 |
+
"eval_samples_per_second": 18.509,
|
| 920 |
+
"eval_steps_per_second": 0.194,
|
| 921 |
+
"step": 321
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 100.0,
|
| 925 |
+
"eval_accuracy": 0.6083916083916084,
|
| 926 |
+
"eval_loss": 1.0821202993392944,
|
| 927 |
+
"eval_runtime": 14.9648,
|
| 928 |
+
"eval_samples_per_second": 19.112,
|
| 929 |
+
"eval_steps_per_second": 0.2,
|
| 930 |
+
"step": 325
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 100.92,
|
| 934 |
+
"eval_accuracy": 0.6083916083916084,
|
| 935 |
+
"eval_loss": 1.075362205505371,
|
| 936 |
+
"eval_runtime": 18.8741,
|
| 937 |
+
"eval_samples_per_second": 15.153,
|
| 938 |
+
"eval_steps_per_second": 0.159,
|
| 939 |
+
"step": 328
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 101.85,
|
| 943 |
+
"eval_accuracy": 0.6083916083916084,
|
| 944 |
+
"eval_loss": 1.06904935836792,
|
| 945 |
+
"eval_runtime": 12.1488,
|
| 946 |
+
"eval_samples_per_second": 23.541,
|
| 947 |
+
"eval_steps_per_second": 0.247,
|
| 948 |
+
"step": 331
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 102.77,
|
| 952 |
+
"eval_accuracy": 0.6153846153846154,
|
| 953 |
+
"eval_loss": 1.0637034177780151,
|
| 954 |
+
"eval_runtime": 21.2582,
|
| 955 |
+
"eval_samples_per_second": 13.454,
|
| 956 |
+
"eval_steps_per_second": 0.141,
|
| 957 |
+
"step": 334
|
| 958 |
+
},
|
| 959 |
+
{
|
| 960 |
+
"epoch": 104.0,
|
| 961 |
+
"eval_accuracy": 0.6188811188811189,
|
| 962 |
+
"eval_loss": 1.0549243688583374,
|
| 963 |
+
"eval_runtime": 19.5513,
|
| 964 |
+
"eval_samples_per_second": 14.628,
|
| 965 |
+
"eval_steps_per_second": 0.153,
|
| 966 |
+
"step": 338
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 104.92,
|
| 970 |
+
"eval_accuracy": 0.6223776223776224,
|
| 971 |
+
"eval_loss": 1.047833800315857,
|
| 972 |
+
"eval_runtime": 17.2191,
|
| 973 |
+
"eval_samples_per_second": 16.609,
|
| 974 |
+
"eval_steps_per_second": 0.174,
|
| 975 |
+
"step": 341
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 105.85,
|
| 979 |
+
"eval_accuracy": 0.6258741258741258,
|
| 980 |
+
"eval_loss": 1.0420035123825073,
|
| 981 |
+
"eval_runtime": 25.1625,
|
| 982 |
+
"eval_samples_per_second": 11.366,
|
| 983 |
+
"eval_steps_per_second": 0.119,
|
| 984 |
+
"step": 344
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 106.77,
|
| 988 |
+
"eval_accuracy": 0.6293706293706294,
|
| 989 |
+
"eval_loss": 1.0369622707366943,
|
| 990 |
+
"eval_runtime": 14.0552,
|
| 991 |
+
"eval_samples_per_second": 20.348,
|
| 992 |
+
"eval_steps_per_second": 0.213,
|
| 993 |
+
"step": 347
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 108.0,
|
| 997 |
+
"eval_accuracy": 0.6293706293706294,
|
| 998 |
+
"eval_loss": 1.0308306217193604,
|
| 999 |
+
"eval_runtime": 15.7213,
|
| 1000 |
+
"eval_samples_per_second": 18.192,
|
| 1001 |
+
"eval_steps_per_second": 0.191,
|
| 1002 |
+
"step": 351
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 108.92,
|
| 1006 |
+
"eval_accuracy": 0.6258741258741258,
|
| 1007 |
+
"eval_loss": 1.0263006687164307,
|
| 1008 |
+
"eval_runtime": 19.1306,
|
| 1009 |
+
"eval_samples_per_second": 14.95,
|
| 1010 |
+
"eval_steps_per_second": 0.157,
|
| 1011 |
+
"step": 354
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 109.85,
|
| 1015 |
+
"eval_accuracy": 0.6258741258741258,
|
| 1016 |
+
"eval_loss": 1.0230927467346191,
|
| 1017 |
+
"eval_runtime": 15.401,
|
| 1018 |
+
"eval_samples_per_second": 18.57,
|
| 1019 |
+
"eval_steps_per_second": 0.195,
|
| 1020 |
+
"step": 357
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 110.77,
|
| 1024 |
+
"eval_accuracy": 0.6328671328671329,
|
| 1025 |
+
"eval_loss": 1.0204286575317383,
|
| 1026 |
+
"eval_runtime": 11.4397,
|
| 1027 |
+
"eval_samples_per_second": 25.001,
|
| 1028 |
+
"eval_steps_per_second": 0.262,
|
| 1029 |
+
"step": 360
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 112.0,
|
| 1033 |
+
"eval_accuracy": 0.6293706293706294,
|
| 1034 |
+
"eval_loss": 1.0166879892349243,
|
| 1035 |
+
"eval_runtime": 14.7547,
|
| 1036 |
+
"eval_samples_per_second": 19.384,
|
| 1037 |
+
"eval_steps_per_second": 0.203,
|
| 1038 |
+
"step": 364
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 112.92,
|
| 1042 |
+
"eval_accuracy": 0.6293706293706294,
|
| 1043 |
+
"eval_loss": 1.0145463943481445,
|
| 1044 |
+
"eval_runtime": 12.2752,
|
| 1045 |
+
"eval_samples_per_second": 23.299,
|
| 1046 |
+
"eval_steps_per_second": 0.244,
|
| 1047 |
+
"step": 367
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 113.85,
|
| 1051 |
+
"eval_accuracy": 0.6328671328671329,
|
| 1052 |
+
"eval_loss": 1.0118980407714844,
|
| 1053 |
+
"eval_runtime": 29.1643,
|
| 1054 |
+
"eval_samples_per_second": 9.807,
|
| 1055 |
+
"eval_steps_per_second": 0.103,
|
| 1056 |
+
"step": 370
|
| 1057 |
+
},
|
| 1058 |
+
{
|
| 1059 |
+
"epoch": 114.77,
|
| 1060 |
+
"eval_accuracy": 0.6293706293706294,
|
| 1061 |
+
"eval_loss": 1.0077061653137207,
|
| 1062 |
+
"eval_runtime": 15.736,
|
| 1063 |
+
"eval_samples_per_second": 18.175,
|
| 1064 |
+
"eval_steps_per_second": 0.191,
|
| 1065 |
+
"step": 373
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 116.0,
|
| 1069 |
+
"eval_accuracy": 0.6363636363636364,
|
| 1070 |
+
"eval_loss": 1.001206398010254,
|
| 1071 |
+
"eval_runtime": 13.5478,
|
| 1072 |
+
"eval_samples_per_second": 21.11,
|
| 1073 |
+
"eval_steps_per_second": 0.221,
|
| 1074 |
+
"step": 377
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 116.92,
|
| 1078 |
+
"eval_accuracy": 0.6363636363636364,
|
| 1079 |
+
"eval_loss": 0.9975122809410095,
|
| 1080 |
+
"eval_runtime": 12.3425,
|
| 1081 |
+
"eval_samples_per_second": 23.172,
|
| 1082 |
+
"eval_steps_per_second": 0.243,
|
| 1083 |
+
"step": 380
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 117.85,
|
| 1087 |
+
"eval_accuracy": 0.6363636363636364,
|
| 1088 |
+
"eval_loss": 0.993752121925354,
|
| 1089 |
+
"eval_runtime": 16.3699,
|
| 1090 |
+
"eval_samples_per_second": 17.471,
|
| 1091 |
+
"eval_steps_per_second": 0.183,
|
| 1092 |
+
"step": 383
|
| 1093 |
+
},
|
| 1094 |
+
{
|
| 1095 |
+
"epoch": 118.77,
|
| 1096 |
+
"eval_accuracy": 0.6398601398601399,
|
| 1097 |
+
"eval_loss": 0.9913222193717957,
|
| 1098 |
+
"eval_runtime": 15.9587,
|
| 1099 |
+
"eval_samples_per_second": 17.921,
|
| 1100 |
+
"eval_steps_per_second": 0.188,
|
| 1101 |
+
"step": 386
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 120.0,
|
| 1105 |
+
"eval_accuracy": 0.6468531468531469,
|
| 1106 |
+
"eval_loss": 0.9886361360549927,
|
| 1107 |
+
"eval_runtime": 14.2728,
|
| 1108 |
+
"eval_samples_per_second": 20.038,
|
| 1109 |
+
"eval_steps_per_second": 0.21,
|
| 1110 |
+
"step": 390
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 120.92,
|
| 1114 |
+
"eval_accuracy": 0.6468531468531469,
|
| 1115 |
+
"eval_loss": 0.987022340297699,
|
| 1116 |
+
"eval_runtime": 28.4357,
|
| 1117 |
+
"eval_samples_per_second": 10.058,
|
| 1118 |
+
"eval_steps_per_second": 0.106,
|
| 1119 |
+
"step": 393
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 121.85,
|
| 1123 |
+
"eval_accuracy": 0.6398601398601399,
|
| 1124 |
+
"eval_loss": 0.9861236810684204,
|
| 1125 |
+
"eval_runtime": 12.6739,
|
| 1126 |
+
"eval_samples_per_second": 22.566,
|
| 1127 |
+
"eval_steps_per_second": 0.237,
|
| 1128 |
+
"step": 396
|
| 1129 |
+
},
|
| 1130 |
+
{
|
| 1131 |
+
"epoch": 122.77,
|
| 1132 |
+
"eval_accuracy": 0.6433566433566433,
|
| 1133 |
+
"eval_loss": 0.9857375025749207,
|
| 1134 |
+
"eval_runtime": 14.4905,
|
| 1135 |
+
"eval_samples_per_second": 19.737,
|
| 1136 |
+
"eval_steps_per_second": 0.207,
|
| 1137 |
+
"step": 399
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 123.08,
|
| 1141 |
+
"grad_norm": 28786.830078125,
|
| 1142 |
+
"learning_rate": 1.5555555555555555e-05,
|
| 1143 |
+
"loss": 0.8183,
|
| 1144 |
+
"step": 400
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 124.0,
|
| 1148 |
+
"eval_accuracy": 0.6398601398601399,
|
| 1149 |
+
"eval_loss": 0.9855088591575623,
|
| 1150 |
+
"eval_runtime": 23.5291,
|
| 1151 |
+
"eval_samples_per_second": 12.155,
|
| 1152 |
+
"eval_steps_per_second": 0.128,
|
| 1153 |
+
"step": 403
|
| 1154 |
+
},
|
| 1155 |
+
{
|
| 1156 |
+
"epoch": 124.92,
|
| 1157 |
+
"eval_accuracy": 0.6398601398601399,
|
| 1158 |
+
"eval_loss": 0.986443817615509,
|
| 1159 |
+
"eval_runtime": 11.267,
|
| 1160 |
+
"eval_samples_per_second": 25.384,
|
| 1161 |
+
"eval_steps_per_second": 0.266,
|
| 1162 |
+
"step": 406
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 125.85,
|
| 1166 |
+
"eval_accuracy": 0.6398601398601399,
|
| 1167 |
+
"eval_loss": 0.9856669902801514,
|
| 1168 |
+
"eval_runtime": 15.102,
|
| 1169 |
+
"eval_samples_per_second": 18.938,
|
| 1170 |
+
"eval_steps_per_second": 0.199,
|
| 1171 |
+
"step": 409
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 126.77,
|
| 1175 |
+
"eval_accuracy": 0.6398601398601399,
|
| 1176 |
+
"eval_loss": 0.9817618131637573,
|
| 1177 |
+
"eval_runtime": 13.7486,
|
| 1178 |
+
"eval_samples_per_second": 20.802,
|
| 1179 |
+
"eval_steps_per_second": 0.218,
|
| 1180 |
+
"step": 412
|
| 1181 |
+
},
|
| 1182 |
+
{
|
| 1183 |
+
"epoch": 128.0,
|
| 1184 |
+
"eval_accuracy": 0.6398601398601399,
|
| 1185 |
+
"eval_loss": 0.9765364527702332,
|
| 1186 |
+
"eval_runtime": 15.0977,
|
| 1187 |
+
"eval_samples_per_second": 18.943,
|
| 1188 |
+
"eval_steps_per_second": 0.199,
|
| 1189 |
+
"step": 416
|
| 1190 |
+
},
|
| 1191 |
+
{
|
| 1192 |
+
"epoch": 128.92,
|
| 1193 |
+
"eval_accuracy": 0.6398601398601399,
|
| 1194 |
+
"eval_loss": 0.9739550352096558,
|
| 1195 |
+
"eval_runtime": 9.9261,
|
| 1196 |
+
"eval_samples_per_second": 28.813,
|
| 1197 |
+
"eval_steps_per_second": 0.302,
|
| 1198 |
+
"step": 419
|
| 1199 |
+
},
|
| 1200 |
+
{
|
| 1201 |
+
"epoch": 129.85,
|
| 1202 |
+
"eval_accuracy": 0.6433566433566433,
|
| 1203 |
+
"eval_loss": 0.9736567139625549,
|
| 1204 |
+
"eval_runtime": 23.6705,
|
| 1205 |
+
"eval_samples_per_second": 12.083,
|
| 1206 |
+
"eval_steps_per_second": 0.127,
|
| 1207 |
+
"step": 422
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 130.77,
|
| 1211 |
+
"eval_accuracy": 0.6468531468531469,
|
| 1212 |
+
"eval_loss": 0.9753543734550476,
|
| 1213 |
+
"eval_runtime": 13.1489,
|
| 1214 |
+
"eval_samples_per_second": 21.751,
|
| 1215 |
+
"eval_steps_per_second": 0.228,
|
| 1216 |
+
"step": 425
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"epoch": 132.0,
|
| 1220 |
+
"eval_accuracy": 0.6468531468531469,
|
| 1221 |
+
"eval_loss": 0.975328803062439,
|
| 1222 |
+
"eval_runtime": 14.1291,
|
| 1223 |
+
"eval_samples_per_second": 20.242,
|
| 1224 |
+
"eval_steps_per_second": 0.212,
|
| 1225 |
+
"step": 429
|
| 1226 |
+
},
|
| 1227 |
+
{
|
| 1228 |
+
"epoch": 132.92,
|
| 1229 |
+
"eval_accuracy": 0.6468531468531469,
|
| 1230 |
+
"eval_loss": 0.9739886522293091,
|
| 1231 |
+
"eval_runtime": 23.3295,
|
| 1232 |
+
"eval_samples_per_second": 12.259,
|
| 1233 |
+
"eval_steps_per_second": 0.129,
|
| 1234 |
+
"step": 432
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 133.85,
|
| 1238 |
+
"eval_accuracy": 0.6468531468531469,
|
| 1239 |
+
"eval_loss": 0.9709957242012024,
|
| 1240 |
+
"eval_runtime": 12.6523,
|
| 1241 |
+
"eval_samples_per_second": 22.605,
|
| 1242 |
+
"eval_steps_per_second": 0.237,
|
| 1243 |
+
"step": 435
|
| 1244 |
+
},
|
| 1245 |
+
{
|
| 1246 |
+
"epoch": 134.77,
|
| 1247 |
+
"eval_accuracy": 0.6468531468531469,
|
| 1248 |
+
"eval_loss": 0.9685706496238708,
|
| 1249 |
+
"eval_runtime": 12.6672,
|
| 1250 |
+
"eval_samples_per_second": 22.578,
|
| 1251 |
+
"eval_steps_per_second": 0.237,
|
| 1252 |
+
"step": 438
|
| 1253 |
+
},
|
| 1254 |
+
{
|
| 1255 |
+
"epoch": 136.0,
|
| 1256 |
+
"eval_accuracy": 0.6468531468531469,
|
| 1257 |
+
"eval_loss": 0.9670514464378357,
|
| 1258 |
+
"eval_runtime": 14.1037,
|
| 1259 |
+
"eval_samples_per_second": 20.278,
|
| 1260 |
+
"eval_steps_per_second": 0.213,
|
| 1261 |
+
"step": 442
|
| 1262 |
+
},
|
| 1263 |
+
{
|
| 1264 |
+
"epoch": 136.92,
|
| 1265 |
+
"eval_accuracy": 0.6433566433566433,
|
| 1266 |
+
"eval_loss": 0.966910719871521,
|
| 1267 |
+
"eval_runtime": 17.8781,
|
| 1268 |
+
"eval_samples_per_second": 15.997,
|
| 1269 |
+
"eval_steps_per_second": 0.168,
|
| 1270 |
+
"step": 445
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 137.85,
|
| 1274 |
+
"eval_accuracy": 0.6398601398601399,
|
| 1275 |
+
"eval_loss": 0.9659475088119507,
|
| 1276 |
+
"eval_runtime": 14.4986,
|
| 1277 |
+
"eval_samples_per_second": 19.726,
|
| 1278 |
+
"eval_steps_per_second": 0.207,
|
| 1279 |
+
"step": 448
|
| 1280 |
+
},
|
| 1281 |
+
{
|
| 1282 |
+
"epoch": 138.77,
|
| 1283 |
+
"eval_accuracy": 0.6433566433566433,
|
| 1284 |
+
"eval_loss": 0.9662479758262634,
|
| 1285 |
+
"eval_runtime": 9.0453,
|
| 1286 |
+
"eval_samples_per_second": 31.619,
|
| 1287 |
+
"eval_steps_per_second": 0.332,
|
| 1288 |
+
"step": 451
|
| 1289 |
+
},
|
| 1290 |
+
{
|
| 1291 |
+
"epoch": 140.0,
|
| 1292 |
+
"eval_accuracy": 0.6433566433566433,
|
| 1293 |
+
"eval_loss": 0.967424750328064,
|
| 1294 |
+
"eval_runtime": 15.45,
|
| 1295 |
+
"eval_samples_per_second": 18.511,
|
| 1296 |
+
"eval_steps_per_second": 0.194,
|
| 1297 |
+
"step": 455
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 140.92,
|
| 1301 |
+
"eval_accuracy": 0.6398601398601399,
|
| 1302 |
+
"eval_loss": 0.9694240689277649,
|
| 1303 |
+
"eval_runtime": 25.3846,
|
| 1304 |
+
"eval_samples_per_second": 11.267,
|
| 1305 |
+
"eval_steps_per_second": 0.118,
|
| 1306 |
+
"step": 458
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 141.85,
|
| 1310 |
+
"eval_accuracy": 0.6468531468531469,
|
| 1311 |
+
"eval_loss": 0.9716474413871765,
|
| 1312 |
+
"eval_runtime": 13.4283,
|
| 1313 |
+
"eval_samples_per_second": 21.298,
|
| 1314 |
+
"eval_steps_per_second": 0.223,
|
| 1315 |
+
"step": 461
|
| 1316 |
+
},
|
| 1317 |
+
{
|
| 1318 |
+
"epoch": 142.77,
|
| 1319 |
+
"eval_accuracy": 0.6433566433566433,
|
| 1320 |
+
"eval_loss": 0.9738869667053223,
|
| 1321 |
+
"eval_runtime": 14.4683,
|
| 1322 |
+
"eval_samples_per_second": 19.767,
|
| 1323 |
+
"eval_steps_per_second": 0.207,
|
| 1324 |
+
"step": 464
|
| 1325 |
+
},
|
| 1326 |
+
{
|
| 1327 |
+
"epoch": 144.0,
|
| 1328 |
+
"eval_accuracy": 0.6468531468531469,
|
| 1329 |
+
"eval_loss": 0.9711639881134033,
|
| 1330 |
+
"eval_runtime": 16.7189,
|
| 1331 |
+
"eval_samples_per_second": 17.106,
|
| 1332 |
+
"eval_steps_per_second": 0.179,
|
| 1333 |
+
"step": 468
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 144.92,
|
| 1337 |
+
"eval_accuracy": 0.6433566433566433,
|
| 1338 |
+
"eval_loss": 0.9670231938362122,
|
| 1339 |
+
"eval_runtime": 17.2531,
|
| 1340 |
+
"eval_samples_per_second": 16.577,
|
| 1341 |
+
"eval_steps_per_second": 0.174,
|
| 1342 |
+
"step": 471
|
| 1343 |
+
},
|
| 1344 |
+
{
|
| 1345 |
+
"epoch": 145.85,
|
| 1346 |
+
"eval_accuracy": 0.6433566433566433,
|
| 1347 |
+
"eval_loss": 0.9637404084205627,
|
| 1348 |
+
"eval_runtime": 7.2246,
|
| 1349 |
+
"eval_samples_per_second": 39.587,
|
| 1350 |
+
"eval_steps_per_second": 0.415,
|
| 1351 |
+
"step": 474
|
| 1352 |
+
},
|
| 1353 |
+
{
|
| 1354 |
+
"epoch": 146.77,
|
| 1355 |
+
"eval_accuracy": 0.6468531468531469,
|
| 1356 |
+
"eval_loss": 0.9625027775764465,
|
| 1357 |
+
"eval_runtime": 22.2959,
|
| 1358 |
+
"eval_samples_per_second": 12.827,
|
| 1359 |
+
"eval_steps_per_second": 0.135,
|
| 1360 |
+
"step": 477
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 148.0,
|
| 1364 |
+
"eval_accuracy": 0.6468531468531469,
|
| 1365 |
+
"eval_loss": 0.9634200930595398,
|
| 1366 |
+
"eval_runtime": 13.0089,
|
| 1367 |
+
"eval_samples_per_second": 21.985,
|
| 1368 |
+
"eval_steps_per_second": 0.231,
|
| 1369 |
+
"step": 481
|
| 1370 |
+
},
|
| 1371 |
+
{
|
| 1372 |
+
"epoch": 148.92,
|
| 1373 |
+
"eval_accuracy": 0.6468531468531469,
|
| 1374 |
+
"eval_loss": 0.9658861756324768,
|
| 1375 |
+
"eval_runtime": 13.9249,
|
| 1376 |
+
"eval_samples_per_second": 20.539,
|
| 1377 |
+
"eval_steps_per_second": 0.215,
|
| 1378 |
+
"step": 484
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 149.85,
|
| 1382 |
+
"eval_accuracy": 0.6468531468531469,
|
| 1383 |
+
"eval_loss": 0.9662710428237915,
|
| 1384 |
+
"eval_runtime": 26.5045,
|
| 1385 |
+
"eval_samples_per_second": 10.791,
|
| 1386 |
+
"eval_steps_per_second": 0.113,
|
| 1387 |
+
"step": 487
|
| 1388 |
+
},
|
| 1389 |
+
{
|
| 1390 |
+
"epoch": 150.77,
|
| 1391 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1392 |
+
"eval_loss": 0.9648845195770264,
|
| 1393 |
+
"eval_runtime": 13.8454,
|
| 1394 |
+
"eval_samples_per_second": 20.657,
|
| 1395 |
+
"eval_steps_per_second": 0.217,
|
| 1396 |
+
"step": 490
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 152.0,
|
| 1400 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1401 |
+
"eval_loss": 0.9654809832572937,
|
| 1402 |
+
"eval_runtime": 11.8579,
|
| 1403 |
+
"eval_samples_per_second": 24.119,
|
| 1404 |
+
"eval_steps_per_second": 0.253,
|
| 1405 |
+
"step": 494
|
| 1406 |
+
},
|
| 1407 |
+
{
|
| 1408 |
+
"epoch": 152.92,
|
| 1409 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1410 |
+
"eval_loss": 0.9648337364196777,
|
| 1411 |
+
"eval_runtime": 13.0859,
|
| 1412 |
+
"eval_samples_per_second": 21.856,
|
| 1413 |
+
"eval_steps_per_second": 0.229,
|
| 1414 |
+
"step": 497
|
| 1415 |
+
},
|
| 1416 |
+
{
|
| 1417 |
+
"epoch": 153.85,
|
| 1418 |
+
"grad_norm": 27909.80078125,
|
| 1419 |
+
"learning_rate": 1.111111111111111e-05,
|
| 1420 |
+
"loss": 0.7321,
|
| 1421 |
+
"step": 500
|
| 1422 |
+
},
|
| 1423 |
+
{
|
| 1424 |
+
"epoch": 153.85,
|
| 1425 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1426 |
+
"eval_loss": 0.963828980922699,
|
| 1427 |
+
"eval_runtime": 8.9676,
|
| 1428 |
+
"eval_samples_per_second": 31.892,
|
| 1429 |
+
"eval_steps_per_second": 0.335,
|
| 1430 |
+
"step": 500
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 154.77,
|
| 1434 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1435 |
+
"eval_loss": 0.9631070494651794,
|
| 1436 |
+
"eval_runtime": 32.863,
|
| 1437 |
+
"eval_samples_per_second": 8.703,
|
| 1438 |
+
"eval_steps_per_second": 0.091,
|
| 1439 |
+
"step": 503
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 156.0,
|
| 1443 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1444 |
+
"eval_loss": 0.964668333530426,
|
| 1445 |
+
"eval_runtime": 11.8735,
|
| 1446 |
+
"eval_samples_per_second": 24.087,
|
| 1447 |
+
"eval_steps_per_second": 0.253,
|
| 1448 |
+
"step": 507
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 156.92,
|
| 1452 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1453 |
+
"eval_loss": 0.9652780294418335,
|
| 1454 |
+
"eval_runtime": 12.5571,
|
| 1455 |
+
"eval_samples_per_second": 22.776,
|
| 1456 |
+
"eval_steps_per_second": 0.239,
|
| 1457 |
+
"step": 510
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 157.85,
|
| 1461 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1462 |
+
"eval_loss": 0.9662439227104187,
|
| 1463 |
+
"eval_runtime": 22.8916,
|
| 1464 |
+
"eval_samples_per_second": 12.494,
|
| 1465 |
+
"eval_steps_per_second": 0.131,
|
| 1466 |
+
"step": 513
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 158.77,
|
| 1470 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1471 |
+
"eval_loss": 0.9679471254348755,
|
| 1472 |
+
"eval_runtime": 13.713,
|
| 1473 |
+
"eval_samples_per_second": 20.856,
|
| 1474 |
+
"eval_steps_per_second": 0.219,
|
| 1475 |
+
"step": 516
|
| 1476 |
+
},
|
| 1477 |
+
{
|
| 1478 |
+
"epoch": 160.0,
|
| 1479 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1480 |
+
"eval_loss": 0.9674729704856873,
|
| 1481 |
+
"eval_runtime": 13.3059,
|
| 1482 |
+
"eval_samples_per_second": 21.494,
|
| 1483 |
+
"eval_steps_per_second": 0.225,
|
| 1484 |
+
"step": 520
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 160.92,
|
| 1488 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1489 |
+
"eval_loss": 0.9663581252098083,
|
| 1490 |
+
"eval_runtime": 14.0564,
|
| 1491 |
+
"eval_samples_per_second": 20.347,
|
| 1492 |
+
"eval_steps_per_second": 0.213,
|
| 1493 |
+
"step": 523
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 161.85,
|
| 1497 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1498 |
+
"eval_loss": 0.9654582738876343,
|
| 1499 |
+
"eval_runtime": 17.3488,
|
| 1500 |
+
"eval_samples_per_second": 16.485,
|
| 1501 |
+
"eval_steps_per_second": 0.173,
|
| 1502 |
+
"step": 526
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 162.77,
|
| 1506 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1507 |
+
"eval_loss": 0.9642163515090942,
|
| 1508 |
+
"eval_runtime": 28.2809,
|
| 1509 |
+
"eval_samples_per_second": 10.113,
|
| 1510 |
+
"eval_steps_per_second": 0.106,
|
| 1511 |
+
"step": 529
|
| 1512 |
+
},
|
| 1513 |
+
{
|
| 1514 |
+
"epoch": 164.0,
|
| 1515 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1516 |
+
"eval_loss": 0.9635214805603027,
|
| 1517 |
+
"eval_runtime": 13.2297,
|
| 1518 |
+
"eval_samples_per_second": 21.618,
|
| 1519 |
+
"eval_steps_per_second": 0.227,
|
| 1520 |
+
"step": 533
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 164.92,
|
| 1524 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1525 |
+
"eval_loss": 0.9632530808448792,
|
| 1526 |
+
"eval_runtime": 14.959,
|
| 1527 |
+
"eval_samples_per_second": 19.119,
|
| 1528 |
+
"eval_steps_per_second": 0.201,
|
| 1529 |
+
"step": 536
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 165.85,
|
| 1533 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1534 |
+
"eval_loss": 0.9644965529441833,
|
| 1535 |
+
"eval_runtime": 27.3169,
|
| 1536 |
+
"eval_samples_per_second": 10.47,
|
| 1537 |
+
"eval_steps_per_second": 0.11,
|
| 1538 |
+
"step": 539
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 166.77,
|
| 1542 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1543 |
+
"eval_loss": 0.9649432897567749,
|
| 1544 |
+
"eval_runtime": 16.2467,
|
| 1545 |
+
"eval_samples_per_second": 17.604,
|
| 1546 |
+
"eval_steps_per_second": 0.185,
|
| 1547 |
+
"step": 542
|
| 1548 |
+
},
|
| 1549 |
+
{
|
| 1550 |
+
"epoch": 168.0,
|
| 1551 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1552 |
+
"eval_loss": 0.9651181101799011,
|
| 1553 |
+
"eval_runtime": 14.8641,
|
| 1554 |
+
"eval_samples_per_second": 19.241,
|
| 1555 |
+
"eval_steps_per_second": 0.202,
|
| 1556 |
+
"step": 546
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 168.92,
|
| 1560 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1561 |
+
"eval_loss": 0.9657481908798218,
|
| 1562 |
+
"eval_runtime": 14.6637,
|
| 1563 |
+
"eval_samples_per_second": 19.504,
|
| 1564 |
+
"eval_steps_per_second": 0.205,
|
| 1565 |
+
"step": 549
|
| 1566 |
+
},
|
| 1567 |
+
{
|
| 1568 |
+
"epoch": 169.85,
|
| 1569 |
+
"eval_accuracy": 0.6538461538461539,
|
| 1570 |
+
"eval_loss": 0.9662555456161499,
|
| 1571 |
+
"eval_runtime": 16.1862,
|
| 1572 |
+
"eval_samples_per_second": 17.669,
|
| 1573 |
+
"eval_steps_per_second": 0.185,
|
| 1574 |
+
"step": 552
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 170.77,
|
| 1578 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1579 |
+
"eval_loss": 0.9652681946754456,
|
| 1580 |
+
"eval_runtime": 13.8601,
|
| 1581 |
+
"eval_samples_per_second": 20.635,
|
| 1582 |
+
"eval_steps_per_second": 0.216,
|
| 1583 |
+
"step": 555
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 172.0,
|
| 1587 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1588 |
+
"eval_loss": 0.9637671113014221,
|
| 1589 |
+
"eval_runtime": 14.0763,
|
| 1590 |
+
"eval_samples_per_second": 20.318,
|
| 1591 |
+
"eval_steps_per_second": 0.213,
|
| 1592 |
+
"step": 559
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 172.92,
|
| 1596 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1597 |
+
"eval_loss": 0.9616447687149048,
|
| 1598 |
+
"eval_runtime": 14.3281,
|
| 1599 |
+
"eval_samples_per_second": 19.961,
|
| 1600 |
+
"eval_steps_per_second": 0.209,
|
| 1601 |
+
"step": 562
|
| 1602 |
+
},
|
| 1603 |
+
{
|
| 1604 |
+
"epoch": 173.85,
|
| 1605 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1606 |
+
"eval_loss": 0.9601487517356873,
|
| 1607 |
+
"eval_runtime": 14.8622,
|
| 1608 |
+
"eval_samples_per_second": 19.243,
|
| 1609 |
+
"eval_steps_per_second": 0.202,
|
| 1610 |
+
"step": 565
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 174.77,
|
| 1614 |
+
"eval_accuracy": 0.6538461538461539,
|
| 1615 |
+
"eval_loss": 0.9610276818275452,
|
| 1616 |
+
"eval_runtime": 11.3604,
|
| 1617 |
+
"eval_samples_per_second": 25.175,
|
| 1618 |
+
"eval_steps_per_second": 0.264,
|
| 1619 |
+
"step": 568
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 176.0,
|
| 1623 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1624 |
+
"eval_loss": 0.962989866733551,
|
| 1625 |
+
"eval_runtime": 33.9341,
|
| 1626 |
+
"eval_samples_per_second": 8.428,
|
| 1627 |
+
"eval_steps_per_second": 0.088,
|
| 1628 |
+
"step": 572
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 176.92,
|
| 1632 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1633 |
+
"eval_loss": 0.9633333683013916,
|
| 1634 |
+
"eval_runtime": 11.5929,
|
| 1635 |
+
"eval_samples_per_second": 24.67,
|
| 1636 |
+
"eval_steps_per_second": 0.259,
|
| 1637 |
+
"step": 575
|
| 1638 |
+
},
|
| 1639 |
+
{
|
| 1640 |
+
"epoch": 177.85,
|
| 1641 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1642 |
+
"eval_loss": 0.9645780324935913,
|
| 1643 |
+
"eval_runtime": 13.4891,
|
| 1644 |
+
"eval_samples_per_second": 21.202,
|
| 1645 |
+
"eval_steps_per_second": 0.222,
|
| 1646 |
+
"step": 578
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 178.77,
|
| 1650 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1651 |
+
"eval_loss": 0.9654965996742249,
|
| 1652 |
+
"eval_runtime": 28.6917,
|
| 1653 |
+
"eval_samples_per_second": 9.968,
|
| 1654 |
+
"eval_steps_per_second": 0.105,
|
| 1655 |
+
"step": 581
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 180.0,
|
| 1659 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1660 |
+
"eval_loss": 0.9672998785972595,
|
| 1661 |
+
"eval_runtime": 14.4786,
|
| 1662 |
+
"eval_samples_per_second": 19.753,
|
| 1663 |
+
"eval_steps_per_second": 0.207,
|
| 1664 |
+
"step": 585
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 180.92,
|
| 1668 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1669 |
+
"eval_loss": 0.9679505825042725,
|
| 1670 |
+
"eval_runtime": 13.6779,
|
| 1671 |
+
"eval_samples_per_second": 20.91,
|
| 1672 |
+
"eval_steps_per_second": 0.219,
|
| 1673 |
+
"step": 588
|
| 1674 |
+
},
|
| 1675 |
+
{
|
| 1676 |
+
"epoch": 181.85,
|
| 1677 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1678 |
+
"eval_loss": 0.9686614274978638,
|
| 1679 |
+
"eval_runtime": 21.421,
|
| 1680 |
+
"eval_samples_per_second": 13.351,
|
| 1681 |
+
"eval_steps_per_second": 0.14,
|
| 1682 |
+
"step": 591
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 182.77,
|
| 1686 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1687 |
+
"eval_loss": 0.9691935777664185,
|
| 1688 |
+
"eval_runtime": 12.6562,
|
| 1689 |
+
"eval_samples_per_second": 22.598,
|
| 1690 |
+
"eval_steps_per_second": 0.237,
|
| 1691 |
+
"step": 594
|
| 1692 |
+
},
|
| 1693 |
+
{
|
| 1694 |
+
"epoch": 184.0,
|
| 1695 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1696 |
+
"eval_loss": 0.9683831334114075,
|
| 1697 |
+
"eval_runtime": 18.1378,
|
| 1698 |
+
"eval_samples_per_second": 15.768,
|
| 1699 |
+
"eval_steps_per_second": 0.165,
|
| 1700 |
+
"step": 598
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 184.62,
|
| 1704 |
+
"grad_norm": 28079.01171875,
|
| 1705 |
+
"learning_rate": 6.666666666666667e-06,
|
| 1706 |
+
"loss": 0.6941,
|
| 1707 |
+
"step": 600
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 184.92,
|
| 1711 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1712 |
+
"eval_loss": 0.9676913022994995,
|
| 1713 |
+
"eval_runtime": 22.197,
|
| 1714 |
+
"eval_samples_per_second": 12.885,
|
| 1715 |
+
"eval_steps_per_second": 0.135,
|
| 1716 |
+
"step": 601
|
| 1717 |
+
},
|
| 1718 |
+
{
|
| 1719 |
+
"epoch": 185.85,
|
| 1720 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1721 |
+
"eval_loss": 0.9673991203308105,
|
| 1722 |
+
"eval_runtime": 13.8487,
|
| 1723 |
+
"eval_samples_per_second": 20.652,
|
| 1724 |
+
"eval_steps_per_second": 0.217,
|
| 1725 |
+
"step": 604
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 186.77,
|
| 1729 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1730 |
+
"eval_loss": 0.9671489000320435,
|
| 1731 |
+
"eval_runtime": 16.6344,
|
| 1732 |
+
"eval_samples_per_second": 17.193,
|
| 1733 |
+
"eval_steps_per_second": 0.18,
|
| 1734 |
+
"step": 607
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 188.0,
|
| 1738 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1739 |
+
"eval_loss": 0.96697998046875,
|
| 1740 |
+
"eval_runtime": 14.8813,
|
| 1741 |
+
"eval_samples_per_second": 19.219,
|
| 1742 |
+
"eval_steps_per_second": 0.202,
|
| 1743 |
+
"step": 611
|
| 1744 |
+
},
|
| 1745 |
+
{
|
| 1746 |
+
"epoch": 188.92,
|
| 1747 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1748 |
+
"eval_loss": 0.9661853313446045,
|
| 1749 |
+
"eval_runtime": 15.8068,
|
| 1750 |
+
"eval_samples_per_second": 18.094,
|
| 1751 |
+
"eval_steps_per_second": 0.19,
|
| 1752 |
+
"step": 614
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 189.85,
|
| 1756 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1757 |
+
"eval_loss": 0.9652778506278992,
|
| 1758 |
+
"eval_runtime": 27.4384,
|
| 1759 |
+
"eval_samples_per_second": 10.423,
|
| 1760 |
+
"eval_steps_per_second": 0.109,
|
| 1761 |
+
"step": 617
|
| 1762 |
+
},
|
| 1763 |
+
{
|
| 1764 |
+
"epoch": 190.77,
|
| 1765 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1766 |
+
"eval_loss": 0.9644954800605774,
|
| 1767 |
+
"eval_runtime": 12.0079,
|
| 1768 |
+
"eval_samples_per_second": 23.818,
|
| 1769 |
+
"eval_steps_per_second": 0.25,
|
| 1770 |
+
"step": 620
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 192.0,
|
| 1774 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1775 |
+
"eval_loss": 0.9648195505142212,
|
| 1776 |
+
"eval_runtime": 15.7494,
|
| 1777 |
+
"eval_samples_per_second": 18.159,
|
| 1778 |
+
"eval_steps_per_second": 0.19,
|
| 1779 |
+
"step": 624
|
| 1780 |
+
},
|
| 1781 |
+
{
|
| 1782 |
+
"epoch": 192.92,
|
| 1783 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1784 |
+
"eval_loss": 0.9652000665664673,
|
| 1785 |
+
"eval_runtime": 14.0294,
|
| 1786 |
+
"eval_samples_per_second": 20.386,
|
| 1787 |
+
"eval_steps_per_second": 0.214,
|
| 1788 |
+
"step": 627
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 193.85,
|
| 1792 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1793 |
+
"eval_loss": 0.9662671089172363,
|
| 1794 |
+
"eval_runtime": 16.6283,
|
| 1795 |
+
"eval_samples_per_second": 17.2,
|
| 1796 |
+
"eval_steps_per_second": 0.18,
|
| 1797 |
+
"step": 630
|
| 1798 |
+
},
|
| 1799 |
+
{
|
| 1800 |
+
"epoch": 194.77,
|
| 1801 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1802 |
+
"eval_loss": 0.9662363529205322,
|
| 1803 |
+
"eval_runtime": 12.434,
|
| 1804 |
+
"eval_samples_per_second": 23.001,
|
| 1805 |
+
"eval_steps_per_second": 0.241,
|
| 1806 |
+
"step": 633
|
| 1807 |
+
},
|
| 1808 |
+
{
|
| 1809 |
+
"epoch": 196.0,
|
| 1810 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1811 |
+
"eval_loss": 0.966452419757843,
|
| 1812 |
+
"eval_runtime": 27.5274,
|
| 1813 |
+
"eval_samples_per_second": 10.39,
|
| 1814 |
+
"eval_steps_per_second": 0.109,
|
| 1815 |
+
"step": 637
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 196.92,
|
| 1819 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1820 |
+
"eval_loss": 0.9667707085609436,
|
| 1821 |
+
"eval_runtime": 14.0792,
|
| 1822 |
+
"eval_samples_per_second": 20.314,
|
| 1823 |
+
"eval_steps_per_second": 0.213,
|
| 1824 |
+
"step": 640
|
| 1825 |
+
},
|
| 1826 |
+
{
|
| 1827 |
+
"epoch": 197.85,
|
| 1828 |
+
"eval_accuracy": 0.6468531468531469,
|
| 1829 |
+
"eval_loss": 0.9669297337532043,
|
| 1830 |
+
"eval_runtime": 11.4357,
|
| 1831 |
+
"eval_samples_per_second": 25.009,
|
| 1832 |
+
"eval_steps_per_second": 0.262,
|
| 1833 |
+
"step": 643
|
| 1834 |
+
},
|
| 1835 |
+
{
|
| 1836 |
+
"epoch": 198.77,
|
| 1837 |
+
"eval_accuracy": 0.6433566433566433,
|
| 1838 |
+
"eval_loss": 0.967424213886261,
|
| 1839 |
+
"eval_runtime": 19.968,
|
| 1840 |
+
"eval_samples_per_second": 14.323,
|
| 1841 |
+
"eval_steps_per_second": 0.15,
|
| 1842 |
+
"step": 646
|
| 1843 |
+
},
|
| 1844 |
+
{
|
| 1845 |
+
"epoch": 200.0,
|
| 1846 |
+
"eval_accuracy": 0.6468531468531469,
|
| 1847 |
+
"eval_loss": 0.9668986797332764,
|
| 1848 |
+
"eval_runtime": 11.3124,
|
| 1849 |
+
"eval_samples_per_second": 25.282,
|
| 1850 |
+
"eval_steps_per_second": 0.265,
|
| 1851 |
+
"step": 650
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 200.92,
|
| 1855 |
+
"eval_accuracy": 0.6468531468531469,
|
| 1856 |
+
"eval_loss": 0.9671657681465149,
|
| 1857 |
+
"eval_runtime": 15.4547,
|
| 1858 |
+
"eval_samples_per_second": 18.506,
|
| 1859 |
+
"eval_steps_per_second": 0.194,
|
| 1860 |
+
"step": 653
|
| 1861 |
+
},
|
| 1862 |
+
{
|
| 1863 |
+
"epoch": 201.85,
|
| 1864 |
+
"eval_accuracy": 0.6468531468531469,
|
| 1865 |
+
"eval_loss": 0.9671434760093689,
|
| 1866 |
+
"eval_runtime": 11.8747,
|
| 1867 |
+
"eval_samples_per_second": 24.085,
|
| 1868 |
+
"eval_steps_per_second": 0.253,
|
| 1869 |
+
"step": 656
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 202.77,
|
| 1873 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1874 |
+
"eval_loss": 0.967323362827301,
|
| 1875 |
+
"eval_runtime": 13.4706,
|
| 1876 |
+
"eval_samples_per_second": 21.231,
|
| 1877 |
+
"eval_steps_per_second": 0.223,
|
| 1878 |
+
"step": 659
|
| 1879 |
+
},
|
| 1880 |
+
{
|
| 1881 |
+
"epoch": 204.0,
|
| 1882 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1883 |
+
"eval_loss": 0.9666022658348083,
|
| 1884 |
+
"eval_runtime": 11.809,
|
| 1885 |
+
"eval_samples_per_second": 24.219,
|
| 1886 |
+
"eval_steps_per_second": 0.254,
|
| 1887 |
+
"step": 663
|
| 1888 |
+
},
|
| 1889 |
+
{
|
| 1890 |
+
"epoch": 204.92,
|
| 1891 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1892 |
+
"eval_loss": 0.966048002243042,
|
| 1893 |
+
"eval_runtime": 30.3868,
|
| 1894 |
+
"eval_samples_per_second": 9.412,
|
| 1895 |
+
"eval_steps_per_second": 0.099,
|
| 1896 |
+
"step": 666
|
| 1897 |
+
},
|
| 1898 |
+
{
|
| 1899 |
+
"epoch": 205.85,
|
| 1900 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1901 |
+
"eval_loss": 0.9656386971473694,
|
| 1902 |
+
"eval_runtime": 10.0795,
|
| 1903 |
+
"eval_samples_per_second": 28.375,
|
| 1904 |
+
"eval_steps_per_second": 0.298,
|
| 1905 |
+
"step": 669
|
| 1906 |
+
},
|
| 1907 |
+
{
|
| 1908 |
+
"epoch": 206.77,
|
| 1909 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1910 |
+
"eval_loss": 0.9650998711585999,
|
| 1911 |
+
"eval_runtime": 11.1232,
|
| 1912 |
+
"eval_samples_per_second": 25.712,
|
| 1913 |
+
"eval_steps_per_second": 0.27,
|
| 1914 |
+
"step": 672
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 208.0,
|
| 1918 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1919 |
+
"eval_loss": 0.9660706520080566,
|
| 1920 |
+
"eval_runtime": 14.3325,
|
| 1921 |
+
"eval_samples_per_second": 19.955,
|
| 1922 |
+
"eval_steps_per_second": 0.209,
|
| 1923 |
+
"step": 676
|
| 1924 |
+
},
|
| 1925 |
+
{
|
| 1926 |
+
"epoch": 208.92,
|
| 1927 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1928 |
+
"eval_loss": 0.9666641354560852,
|
| 1929 |
+
"eval_runtime": 18.7486,
|
| 1930 |
+
"eval_samples_per_second": 15.254,
|
| 1931 |
+
"eval_steps_per_second": 0.16,
|
| 1932 |
+
"step": 679
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 209.85,
|
| 1936 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1937 |
+
"eval_loss": 0.9667859077453613,
|
| 1938 |
+
"eval_runtime": 11.9596,
|
| 1939 |
+
"eval_samples_per_second": 23.914,
|
| 1940 |
+
"eval_steps_per_second": 0.251,
|
| 1941 |
+
"step": 682
|
| 1942 |
+
},
|
| 1943 |
+
{
|
| 1944 |
+
"epoch": 210.77,
|
| 1945 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1946 |
+
"eval_loss": 0.9668753147125244,
|
| 1947 |
+
"eval_runtime": 21.049,
|
| 1948 |
+
"eval_samples_per_second": 13.587,
|
| 1949 |
+
"eval_steps_per_second": 0.143,
|
| 1950 |
+
"step": 685
|
| 1951 |
+
},
|
| 1952 |
+
{
|
| 1953 |
+
"epoch": 212.0,
|
| 1954 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1955 |
+
"eval_loss": 0.9665384292602539,
|
| 1956 |
+
"eval_runtime": 12.916,
|
| 1957 |
+
"eval_samples_per_second": 22.143,
|
| 1958 |
+
"eval_steps_per_second": 0.232,
|
| 1959 |
+
"step": 689
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 212.92,
|
| 1963 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1964 |
+
"eval_loss": 0.9665247797966003,
|
| 1965 |
+
"eval_runtime": 13.5797,
|
| 1966 |
+
"eval_samples_per_second": 21.061,
|
| 1967 |
+
"eval_steps_per_second": 0.221,
|
| 1968 |
+
"step": 692
|
| 1969 |
+
},
|
| 1970 |
+
{
|
| 1971 |
+
"epoch": 213.85,
|
| 1972 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1973 |
+
"eval_loss": 0.9663905501365662,
|
| 1974 |
+
"eval_runtime": 11.9452,
|
| 1975 |
+
"eval_samples_per_second": 23.943,
|
| 1976 |
+
"eval_steps_per_second": 0.251,
|
| 1977 |
+
"step": 695
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 214.77,
|
| 1981 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1982 |
+
"eval_loss": 0.9662503600120544,
|
| 1983 |
+
"eval_runtime": 11.3523,
|
| 1984 |
+
"eval_samples_per_second": 25.193,
|
| 1985 |
+
"eval_steps_per_second": 0.264,
|
| 1986 |
+
"step": 698
|
| 1987 |
+
},
|
| 1988 |
+
{
|
| 1989 |
+
"epoch": 215.38,
|
| 1990 |
+
"grad_norm": 25925.685546875,
|
| 1991 |
+
"learning_rate": 2.222222222222222e-06,
|
| 1992 |
+
"loss": 0.6696,
|
| 1993 |
+
"step": 700
|
| 1994 |
+
},
|
| 1995 |
+
{
|
| 1996 |
+
"epoch": 216.0,
|
| 1997 |
+
"eval_accuracy": 0.6503496503496503,
|
| 1998 |
+
"eval_loss": 0.9666140079498291,
|
| 1999 |
+
"eval_runtime": 11.4254,
|
| 2000 |
+
"eval_samples_per_second": 25.032,
|
| 2001 |
+
"eval_steps_per_second": 0.263,
|
| 2002 |
+
"step": 702
|
| 2003 |
+
},
|
| 2004 |
+
{
|
| 2005 |
+
"epoch": 216.92,
|
| 2006 |
+
"eval_accuracy": 0.6503496503496503,
|
| 2007 |
+
"eval_loss": 0.9667044878005981,
|
| 2008 |
+
"eval_runtime": 9.4078,
|
| 2009 |
+
"eval_samples_per_second": 30.4,
|
| 2010 |
+
"eval_steps_per_second": 0.319,
|
| 2011 |
+
"step": 705
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 217.85,
|
| 2015 |
+
"eval_accuracy": 0.6503496503496503,
|
| 2016 |
+
"eval_loss": 0.9665290713310242,
|
| 2017 |
+
"eval_runtime": 8.9463,
|
| 2018 |
+
"eval_samples_per_second": 31.969,
|
| 2019 |
+
"eval_steps_per_second": 0.335,
|
| 2020 |
+
"step": 708
|
| 2021 |
+
},
|
| 2022 |
+
{
|
| 2023 |
+
"epoch": 218.77,
|
| 2024 |
+
"eval_accuracy": 0.6503496503496503,
|
| 2025 |
+
"eval_loss": 0.9663181900978088,
|
| 2026 |
+
"eval_runtime": 12.4146,
|
| 2027 |
+
"eval_samples_per_second": 23.037,
|
| 2028 |
+
"eval_steps_per_second": 0.242,
|
| 2029 |
+
"step": 711
|
| 2030 |
+
},
|
| 2031 |
+
{
|
| 2032 |
+
"epoch": 220.0,
|
| 2033 |
+
"eval_accuracy": 0.6503496503496503,
|
| 2034 |
+
"eval_loss": 0.9661046862602234,
|
| 2035 |
+
"eval_runtime": 12.2235,
|
| 2036 |
+
"eval_samples_per_second": 23.398,
|
| 2037 |
+
"eval_steps_per_second": 0.245,
|
| 2038 |
+
"step": 715
|
| 2039 |
+
},
|
| 2040 |
+
{
|
| 2041 |
+
"epoch": 220.92,
|
| 2042 |
+
"eval_accuracy": 0.6503496503496503,
|
| 2043 |
+
"eval_loss": 0.966101348400116,
|
| 2044 |
+
"eval_runtime": 12.7423,
|
| 2045 |
+
"eval_samples_per_second": 22.445,
|
| 2046 |
+
"eval_steps_per_second": 0.235,
|
| 2047 |
+
"step": 718
|
| 2048 |
+
},
|
| 2049 |
+
{
|
| 2050 |
+
"epoch": 221.85,
|
| 2051 |
+
"eval_accuracy": 0.6503496503496503,
|
| 2052 |
+
"eval_loss": 0.9661813974380493,
|
| 2053 |
+
"eval_runtime": 11.077,
|
| 2054 |
+
"eval_samples_per_second": 25.819,
|
| 2055 |
+
"eval_steps_per_second": 0.271,
|
| 2056 |
+
"step": 721
|
| 2057 |
+
},
|
| 2058 |
+
{
|
| 2059 |
+
"epoch": 222.77,
|
| 2060 |
+
"eval_accuracy": 0.6503496503496503,
|
| 2061 |
+
"eval_loss": 0.9663574695587158,
|
| 2062 |
+
"eval_runtime": 13.5789,
|
| 2063 |
+
"eval_samples_per_second": 21.062,
|
| 2064 |
+
"eval_steps_per_second": 0.221,
|
| 2065 |
+
"step": 724
|
| 2066 |
+
},
|
| 2067 |
+
{
|
| 2068 |
+
"epoch": 224.0,
|
| 2069 |
+
"eval_accuracy": 0.6503496503496503,
|
| 2070 |
+
"eval_loss": 0.9663906693458557,
|
| 2071 |
+
"eval_runtime": 11.7803,
|
| 2072 |
+
"eval_samples_per_second": 24.278,
|
| 2073 |
+
"eval_steps_per_second": 0.255,
|
| 2074 |
+
"step": 728
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 224.92,
|
| 2078 |
+
"eval_accuracy": 0.6503496503496503,
|
| 2079 |
+
"eval_loss": 0.9664490222930908,
|
| 2080 |
+
"eval_runtime": 13.0369,
|
| 2081 |
+
"eval_samples_per_second": 21.938,
|
| 2082 |
+
"eval_steps_per_second": 0.23,
|
| 2083 |
+
"step": 731
|
| 2084 |
+
},
|
| 2085 |
+
{
|
| 2086 |
+
"epoch": 225.85,
|
| 2087 |
+
"eval_accuracy": 0.6503496503496503,
|
| 2088 |
+
"eval_loss": 0.9665610790252686,
|
| 2089 |
+
"eval_runtime": 10.8415,
|
| 2090 |
+
"eval_samples_per_second": 26.38,
|
| 2091 |
+
"eval_steps_per_second": 0.277,
|
| 2092 |
+
"step": 734
|
| 2093 |
+
},
|
| 2094 |
+
{
|
| 2095 |
+
"epoch": 226.77,
|
| 2096 |
+
"eval_accuracy": 0.6503496503496503,
|
| 2097 |
+
"eval_loss": 0.9665590524673462,
|
| 2098 |
+
"eval_runtime": 14.6312,
|
| 2099 |
+
"eval_samples_per_second": 19.547,
|
| 2100 |
+
"eval_steps_per_second": 0.205,
|
| 2101 |
+
"step": 737
|
| 2102 |
+
},
|
| 2103 |
+
{
|
| 2104 |
+
"epoch": 228.0,
|
| 2105 |
+
"eval_accuracy": 0.6503496503496503,
|
| 2106 |
+
"eval_loss": 0.9665370583534241,
|
| 2107 |
+
"eval_runtime": 11.8502,
|
| 2108 |
+
"eval_samples_per_second": 24.135,
|
| 2109 |
+
"eval_steps_per_second": 0.253,
|
| 2110 |
+
"step": 741
|
| 2111 |
+
},
|
| 2112 |
+
{
|
| 2113 |
+
"epoch": 228.92,
|
| 2114 |
+
"eval_accuracy": 0.6503496503496503,
|
| 2115 |
+
"eval_loss": 0.9665577411651611,
|
| 2116 |
+
"eval_runtime": 12.7745,
|
| 2117 |
+
"eval_samples_per_second": 22.388,
|
| 2118 |
+
"eval_steps_per_second": 0.235,
|
| 2119 |
+
"step": 744
|
| 2120 |
+
},
|
| 2121 |
+
{
|
| 2122 |
+
"epoch": 229.85,
|
| 2123 |
+
"eval_accuracy": 0.6503496503496503,
|
| 2124 |
+
"eval_loss": 0.9665866494178772,
|
| 2125 |
+
"eval_runtime": 13.6538,
|
| 2126 |
+
"eval_samples_per_second": 20.947,
|
| 2127 |
+
"eval_steps_per_second": 0.22,
|
| 2128 |
+
"step": 747
|
| 2129 |
+
},
|
| 2130 |
+
{
|
| 2131 |
+
"epoch": 230.77,
|
| 2132 |
+
"eval_accuracy": 0.6503496503496503,
|
| 2133 |
+
"eval_loss": 0.9665878415107727,
|
| 2134 |
+
"eval_runtime": 11.2039,
|
| 2135 |
+
"eval_samples_per_second": 25.527,
|
| 2136 |
+
"eval_steps_per_second": 0.268,
|
| 2137 |
+
"step": 750
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 230.77,
|
| 2141 |
+
"step": 750,
|
| 2142 |
+
"total_flos": 3.12964745557632e+18,
|
| 2143 |
+
"train_loss": 0.9558865051269532,
|
| 2144 |
+
"train_runtime": 10052.9365,
|
| 2145 |
+
"train_samples_per_second": 38.496,
|
| 2146 |
+
"train_steps_per_second": 0.075
|
| 2147 |
+
}
|
| 2148 |
+
],
|
| 2149 |
+
"logging_steps": 100,
|
| 2150 |
+
"max_steps": 750,
|
| 2151 |
+
"num_input_tokens_seen": 0,
|
| 2152 |
+
"num_train_epochs": 250,
|
| 2153 |
+
"save_steps": 500,
|
| 2154 |
+
"total_flos": 3.12964745557632e+18,
|
| 2155 |
+
"train_batch_size": 128,
|
| 2156 |
+
"trial_name": null,
|
| 2157 |
+
"trial_params": null
|
| 2158 |
+
}
|