Commit
•
e0f2216
1
Parent(s):
fc3a4d2
End of training
Browse files- README.md +1 -1
- all_results.json +15 -0
- eval_results.json +9 -0
- runs/Oct29_13-03-52_443023c04f1e/events.out.tfevents.1730266679.443023c04f1e.36.1 +3 -0
- train_results.json +9 -0
- trainer_state.json +1487 -0
README.md
CHANGED
|
@@ -15,7 +15,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 15 |
|
| 16 |
# whisper-large-v3-ft-btb-ca-cy
|
| 17 |
|
| 18 |
-
This model is a fine-tuned version of [openai/whisper-large-v3](https://huggingface.co/openai/whisper-large-v3) on
|
| 19 |
It achieves the following results on the evaluation set:
|
| 20 |
- Loss: 0.4037
|
| 21 |
- Wer: 0.2791
|
|
|
|
| 15 |
|
| 16 |
# whisper-large-v3-ft-btb-ca-cy
|
| 17 |
|
| 18 |
+
This model is a fine-tuned version of [openai/whisper-large-v3](https://huggingface.co/openai/whisper-large-v3) on the DewiBrynJones/banc-trawsgrifiadau-bangor-clean train main, cymen-arfor/15awr train+dev+test main dataset.
|
| 19 |
It achieves the following results on the evaluation set:
|
| 20 |
- Loss: 0.4037
|
| 21 |
- Wer: 0.2791
|
all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.1938677738741617,
|
| 3 |
+
"eval_loss": 0.40369176864624023,
|
| 4 |
+
"eval_runtime": 1635.5488,
|
| 5 |
+
"eval_samples": 3901,
|
| 6 |
+
"eval_samples_per_second": 2.385,
|
| 7 |
+
"eval_steps_per_second": 0.149,
|
| 8 |
+
"eval_wer": 0.27909784603496574,
|
| 9 |
+
"total_flos": 5.435895365546803e+20,
|
| 10 |
+
"train_loss": 0.3537434986591339,
|
| 11 |
+
"train_runtime": 57744.1557,
|
| 12 |
+
"train_samples": 50095,
|
| 13 |
+
"train_samples_per_second": 2.771,
|
| 14 |
+
"train_steps_per_second": 0.087
|
| 15 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.1938677738741617,
|
| 3 |
+
"eval_loss": 0.40369176864624023,
|
| 4 |
+
"eval_runtime": 1635.5488,
|
| 5 |
+
"eval_samples": 3901,
|
| 6 |
+
"eval_samples_per_second": 2.385,
|
| 7 |
+
"eval_steps_per_second": 0.149,
|
| 8 |
+
"eval_wer": 0.27909784603496574
|
| 9 |
+
}
|
runs/Oct29_13-03-52_443023c04f1e/events.out.tfevents.1730266679.443023c04f1e.36.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f7e733abdd8dffbbbf172dd82e60ec570d7c09c5639c4c1d7551b2971f92437b
|
| 3 |
+
size 406
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.1938677738741617,
|
| 3 |
+
"total_flos": 5.435895365546803e+20,
|
| 4 |
+
"train_loss": 0.3537434986591339,
|
| 5 |
+
"train_runtime": 57744.1557,
|
| 6 |
+
"train_samples": 50095,
|
| 7 |
+
"train_samples_per_second": 2.771,
|
| 8 |
+
"train_steps_per_second": 0.087
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1487 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.1938677738741617,
|
| 5 |
+
"eval_steps": 1000,
|
| 6 |
+
"global_step": 5000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.015969338869370808,
|
| 13 |
+
"grad_norm": 9.112746238708496,
|
| 14 |
+
"learning_rate": 5.000000000000001e-07,
|
| 15 |
+
"loss": 1.6703,
|
| 16 |
+
"step": 25
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.031938677738741615,
|
| 20 |
+
"grad_norm": 5.510941505432129,
|
| 21 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 22 |
+
"loss": 1.2963,
|
| 23 |
+
"step": 50
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.047908016608112426,
|
| 27 |
+
"grad_norm": 5.158988952636719,
|
| 28 |
+
"learning_rate": 1.5e-06,
|
| 29 |
+
"loss": 1.0164,
|
| 30 |
+
"step": 75
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.06387735547748323,
|
| 34 |
+
"grad_norm": 5.4585957527160645,
|
| 35 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 36 |
+
"loss": 0.8123,
|
| 37 |
+
"step": 100
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.07984669434685404,
|
| 41 |
+
"grad_norm": 5.480216979980469,
|
| 42 |
+
"learning_rate": 2.5e-06,
|
| 43 |
+
"loss": 0.7918,
|
| 44 |
+
"step": 125
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.09581603321622485,
|
| 48 |
+
"grad_norm": 5.493738651275635,
|
| 49 |
+
"learning_rate": 3e-06,
|
| 50 |
+
"loss": 0.7452,
|
| 51 |
+
"step": 150
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.11178537208559565,
|
| 55 |
+
"grad_norm": 4.575730800628662,
|
| 56 |
+
"learning_rate": 3.5e-06,
|
| 57 |
+
"loss": 0.7583,
|
| 58 |
+
"step": 175
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.12775471095496646,
|
| 62 |
+
"grad_norm": 4.527775764465332,
|
| 63 |
+
"learning_rate": 4.000000000000001e-06,
|
| 64 |
+
"loss": 0.7108,
|
| 65 |
+
"step": 200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.14372404982433729,
|
| 69 |
+
"grad_norm": 4.807627201080322,
|
| 70 |
+
"learning_rate": 4.5e-06,
|
| 71 |
+
"loss": 0.7146,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.15969338869370808,
|
| 76 |
+
"grad_norm": 4.733044147491455,
|
| 77 |
+
"learning_rate": 5e-06,
|
| 78 |
+
"loss": 0.659,
|
| 79 |
+
"step": 250
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.17566272756307888,
|
| 83 |
+
"grad_norm": 5.310563564300537,
|
| 84 |
+
"learning_rate": 5.500000000000001e-06,
|
| 85 |
+
"loss": 0.7,
|
| 86 |
+
"step": 275
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.1916320664324497,
|
| 90 |
+
"grad_norm": 4.584811687469482,
|
| 91 |
+
"learning_rate": 6e-06,
|
| 92 |
+
"loss": 0.6643,
|
| 93 |
+
"step": 300
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.2076014053018205,
|
| 97 |
+
"grad_norm": 5.011808395385742,
|
| 98 |
+
"learning_rate": 6.5000000000000004e-06,
|
| 99 |
+
"loss": 0.6696,
|
| 100 |
+
"step": 325
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.2235707441711913,
|
| 104 |
+
"grad_norm": 4.633641719818115,
|
| 105 |
+
"learning_rate": 7e-06,
|
| 106 |
+
"loss": 0.6692,
|
| 107 |
+
"step": 350
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.23954008304056212,
|
| 111 |
+
"grad_norm": 3.665330648422241,
|
| 112 |
+
"learning_rate": 7.500000000000001e-06,
|
| 113 |
+
"loss": 0.6285,
|
| 114 |
+
"step": 375
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.2555094219099329,
|
| 118 |
+
"grad_norm": 5.0646185874938965,
|
| 119 |
+
"learning_rate": 8.000000000000001e-06,
|
| 120 |
+
"loss": 0.6313,
|
| 121 |
+
"step": 400
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.2714787607793037,
|
| 125 |
+
"grad_norm": 3.9207818508148193,
|
| 126 |
+
"learning_rate": 8.5e-06,
|
| 127 |
+
"loss": 0.6045,
|
| 128 |
+
"step": 425
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.28744809964867457,
|
| 132 |
+
"grad_norm": 4.316348552703857,
|
| 133 |
+
"learning_rate": 9e-06,
|
| 134 |
+
"loss": 0.6044,
|
| 135 |
+
"step": 450
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.30341743851804537,
|
| 139 |
+
"grad_norm": 3.9778010845184326,
|
| 140 |
+
"learning_rate": 9.5e-06,
|
| 141 |
+
"loss": 0.6015,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.31938677738741617,
|
| 146 |
+
"grad_norm": 3.8584372997283936,
|
| 147 |
+
"learning_rate": 1e-05,
|
| 148 |
+
"loss": 0.5793,
|
| 149 |
+
"step": 500
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.33535611625678696,
|
| 153 |
+
"grad_norm": 4.777307033538818,
|
| 154 |
+
"learning_rate": 9.944444444444445e-06,
|
| 155 |
+
"loss": 0.6289,
|
| 156 |
+
"step": 525
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.35132545512615776,
|
| 160 |
+
"grad_norm": 4.359010219573975,
|
| 161 |
+
"learning_rate": 9.88888888888889e-06,
|
| 162 |
+
"loss": 0.6047,
|
| 163 |
+
"step": 550
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.3672947939955286,
|
| 167 |
+
"grad_norm": 4.476827621459961,
|
| 168 |
+
"learning_rate": 9.833333333333333e-06,
|
| 169 |
+
"loss": 0.5868,
|
| 170 |
+
"step": 575
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.3832641328648994,
|
| 174 |
+
"grad_norm": 3.478997230529785,
|
| 175 |
+
"learning_rate": 9.777777777777779e-06,
|
| 176 |
+
"loss": 0.5877,
|
| 177 |
+
"step": 600
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.3992334717342702,
|
| 181 |
+
"grad_norm": 4.191431522369385,
|
| 182 |
+
"learning_rate": 9.722222222222223e-06,
|
| 183 |
+
"loss": 0.547,
|
| 184 |
+
"step": 625
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.415202810603641,
|
| 188 |
+
"grad_norm": 3.8496177196502686,
|
| 189 |
+
"learning_rate": 9.666666666666667e-06,
|
| 190 |
+
"loss": 0.5511,
|
| 191 |
+
"step": 650
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.4311721494730118,
|
| 195 |
+
"grad_norm": 3.78070330619812,
|
| 196 |
+
"learning_rate": 9.611111111111112e-06,
|
| 197 |
+
"loss": 0.5503,
|
| 198 |
+
"step": 675
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.4471414883423826,
|
| 202 |
+
"grad_norm": 3.223071813583374,
|
| 203 |
+
"learning_rate": 9.555555555555556e-06,
|
| 204 |
+
"loss": 0.5418,
|
| 205 |
+
"step": 700
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.46311082721175345,
|
| 209 |
+
"grad_norm": 4.323662281036377,
|
| 210 |
+
"learning_rate": 9.5e-06,
|
| 211 |
+
"loss": 0.5362,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.47908016608112425,
|
| 216 |
+
"grad_norm": 3.7754769325256348,
|
| 217 |
+
"learning_rate": 9.444444444444445e-06,
|
| 218 |
+
"loss": 0.5478,
|
| 219 |
+
"step": 750
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.49504950495049505,
|
| 223 |
+
"grad_norm": 4.361028671264648,
|
| 224 |
+
"learning_rate": 9.38888888888889e-06,
|
| 225 |
+
"loss": 0.5257,
|
| 226 |
+
"step": 775
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.5110188438198658,
|
| 230 |
+
"grad_norm": 3.5216853618621826,
|
| 231 |
+
"learning_rate": 9.333333333333334e-06,
|
| 232 |
+
"loss": 0.5046,
|
| 233 |
+
"step": 800
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.5269881826892366,
|
| 237 |
+
"grad_norm": 3.39201021194458,
|
| 238 |
+
"learning_rate": 9.277777777777778e-06,
|
| 239 |
+
"loss": 0.5209,
|
| 240 |
+
"step": 825
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.5429575215586074,
|
| 244 |
+
"grad_norm": 3.3704864978790283,
|
| 245 |
+
"learning_rate": 9.222222222222224e-06,
|
| 246 |
+
"loss": 0.5128,
|
| 247 |
+
"step": 850
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.5589268604279782,
|
| 251 |
+
"grad_norm": 3.4954028129577637,
|
| 252 |
+
"learning_rate": 9.166666666666666e-06,
|
| 253 |
+
"loss": 0.4954,
|
| 254 |
+
"step": 875
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.5748961992973491,
|
| 258 |
+
"grad_norm": 3.267204523086548,
|
| 259 |
+
"learning_rate": 9.111111111111112e-06,
|
| 260 |
+
"loss": 0.5204,
|
| 261 |
+
"step": 900
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.5908655381667199,
|
| 265 |
+
"grad_norm": 3.802452802658081,
|
| 266 |
+
"learning_rate": 9.055555555555556e-06,
|
| 267 |
+
"loss": 0.4988,
|
| 268 |
+
"step": 925
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.6068348770360907,
|
| 272 |
+
"grad_norm": 3.5180504322052,
|
| 273 |
+
"learning_rate": 9e-06,
|
| 274 |
+
"loss": 0.5068,
|
| 275 |
+
"step": 950
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.6228042159054615,
|
| 279 |
+
"grad_norm": 3.5314347743988037,
|
| 280 |
+
"learning_rate": 8.944444444444446e-06,
|
| 281 |
+
"loss": 0.5039,
|
| 282 |
+
"step": 975
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.6387735547748323,
|
| 286 |
+
"grad_norm": 3.1311240196228027,
|
| 287 |
+
"learning_rate": 8.888888888888888e-06,
|
| 288 |
+
"loss": 0.4739,
|
| 289 |
+
"step": 1000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.6387735547748323,
|
| 293 |
+
"eval_loss": 0.4840342402458191,
|
| 294 |
+
"eval_runtime": 1688.2428,
|
| 295 |
+
"eval_samples_per_second": 2.311,
|
| 296 |
+
"eval_steps_per_second": 0.145,
|
| 297 |
+
"eval_wer": 0.3517579445571332,
|
| 298 |
+
"step": 1000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.6547428936442031,
|
| 302 |
+
"grad_norm": 2.715566396713257,
|
| 303 |
+
"learning_rate": 8.833333333333334e-06,
|
| 304 |
+
"loss": 0.502,
|
| 305 |
+
"step": 1025
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.6707122325135739,
|
| 309 |
+
"grad_norm": 3.692412853240967,
|
| 310 |
+
"learning_rate": 8.777777777777778e-06,
|
| 311 |
+
"loss": 0.4733,
|
| 312 |
+
"step": 1050
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.6866815713829447,
|
| 316 |
+
"grad_norm": 3.414935350418091,
|
| 317 |
+
"learning_rate": 8.722222222222224e-06,
|
| 318 |
+
"loss": 0.4945,
|
| 319 |
+
"step": 1075
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.7026509102523155,
|
| 323 |
+
"grad_norm": 3.058530569076538,
|
| 324 |
+
"learning_rate": 8.666666666666668e-06,
|
| 325 |
+
"loss": 0.5033,
|
| 326 |
+
"step": 1100
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.7186202491216863,
|
| 330 |
+
"grad_norm": 3.8033292293548584,
|
| 331 |
+
"learning_rate": 8.611111111111112e-06,
|
| 332 |
+
"loss": 0.483,
|
| 333 |
+
"step": 1125
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.7345895879910572,
|
| 337 |
+
"grad_norm": 3.238865852355957,
|
| 338 |
+
"learning_rate": 8.555555555555556e-06,
|
| 339 |
+
"loss": 0.4809,
|
| 340 |
+
"step": 1150
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.750558926860428,
|
| 344 |
+
"grad_norm": 2.665280342102051,
|
| 345 |
+
"learning_rate": 8.5e-06,
|
| 346 |
+
"loss": 0.4792,
|
| 347 |
+
"step": 1175
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 0.7665282657297988,
|
| 351 |
+
"grad_norm": 2.9003748893737793,
|
| 352 |
+
"learning_rate": 8.444444444444446e-06,
|
| 353 |
+
"loss": 0.4662,
|
| 354 |
+
"step": 1200
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.7824976045991696,
|
| 358 |
+
"grad_norm": 2.7720658779144287,
|
| 359 |
+
"learning_rate": 8.38888888888889e-06,
|
| 360 |
+
"loss": 0.4532,
|
| 361 |
+
"step": 1225
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 0.7984669434685404,
|
| 365 |
+
"grad_norm": 3.0096704959869385,
|
| 366 |
+
"learning_rate": 8.333333333333334e-06,
|
| 367 |
+
"loss": 0.4719,
|
| 368 |
+
"step": 1250
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 0.8144362823379112,
|
| 372 |
+
"grad_norm": 3.1118383407592773,
|
| 373 |
+
"learning_rate": 8.277777777777778e-06,
|
| 374 |
+
"loss": 0.4567,
|
| 375 |
+
"step": 1275
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.830405621207282,
|
| 379 |
+
"grad_norm": 2.854698896408081,
|
| 380 |
+
"learning_rate": 8.222222222222222e-06,
|
| 381 |
+
"loss": 0.4729,
|
| 382 |
+
"step": 1300
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 0.8463749600766528,
|
| 386 |
+
"grad_norm": 3.1824917793273926,
|
| 387 |
+
"learning_rate": 8.166666666666668e-06,
|
| 388 |
+
"loss": 0.5073,
|
| 389 |
+
"step": 1325
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 0.8623442989460236,
|
| 393 |
+
"grad_norm": 3.554313898086548,
|
| 394 |
+
"learning_rate": 8.111111111111112e-06,
|
| 395 |
+
"loss": 0.4663,
|
| 396 |
+
"step": 1350
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 0.8783136378153944,
|
| 400 |
+
"grad_norm": 4.958135604858398,
|
| 401 |
+
"learning_rate": 8.055555555555557e-06,
|
| 402 |
+
"loss": 0.4614,
|
| 403 |
+
"step": 1375
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.8942829766847652,
|
| 407 |
+
"grad_norm": 3.5128700733184814,
|
| 408 |
+
"learning_rate": 8.000000000000001e-06,
|
| 409 |
+
"loss": 0.4522,
|
| 410 |
+
"step": 1400
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.9102523155541361,
|
| 414 |
+
"grad_norm": 4.640254020690918,
|
| 415 |
+
"learning_rate": 7.944444444444445e-06,
|
| 416 |
+
"loss": 0.4488,
|
| 417 |
+
"step": 1425
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.9262216544235069,
|
| 421 |
+
"grad_norm": 2.928645610809326,
|
| 422 |
+
"learning_rate": 7.88888888888889e-06,
|
| 423 |
+
"loss": 0.4352,
|
| 424 |
+
"step": 1450
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 0.9421909932928777,
|
| 428 |
+
"grad_norm": 3.42668080329895,
|
| 429 |
+
"learning_rate": 7.833333333333333e-06,
|
| 430 |
+
"loss": 0.4648,
|
| 431 |
+
"step": 1475
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.9581603321622485,
|
| 435 |
+
"grad_norm": 3.204360008239746,
|
| 436 |
+
"learning_rate": 7.77777777777778e-06,
|
| 437 |
+
"loss": 0.4318,
|
| 438 |
+
"step": 1500
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 0.9741296710316193,
|
| 442 |
+
"grad_norm": 2.9323625564575195,
|
| 443 |
+
"learning_rate": 7.722222222222223e-06,
|
| 444 |
+
"loss": 0.4464,
|
| 445 |
+
"step": 1525
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 0.9900990099009901,
|
| 449 |
+
"grad_norm": 3.103804349899292,
|
| 450 |
+
"learning_rate": 7.666666666666667e-06,
|
| 451 |
+
"loss": 0.4599,
|
| 452 |
+
"step": 1550
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 1.006068348770361,
|
| 456 |
+
"grad_norm": 2.95595121383667,
|
| 457 |
+
"learning_rate": 7.611111111111111e-06,
|
| 458 |
+
"loss": 0.4017,
|
| 459 |
+
"step": 1575
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 1.0220376876397317,
|
| 463 |
+
"grad_norm": 2.5740716457366943,
|
| 464 |
+
"learning_rate": 7.555555555555556e-06,
|
| 465 |
+
"loss": 0.3208,
|
| 466 |
+
"step": 1600
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 1.0380070265091026,
|
| 470 |
+
"grad_norm": 3.569032907485962,
|
| 471 |
+
"learning_rate": 7.500000000000001e-06,
|
| 472 |
+
"loss": 0.3206,
|
| 473 |
+
"step": 1625
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 1.0539763653784733,
|
| 477 |
+
"grad_norm": 3.01520037651062,
|
| 478 |
+
"learning_rate": 7.444444444444445e-06,
|
| 479 |
+
"loss": 0.325,
|
| 480 |
+
"step": 1650
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 1.0699457042478442,
|
| 484 |
+
"grad_norm": 2.9796643257141113,
|
| 485 |
+
"learning_rate": 7.38888888888889e-06,
|
| 486 |
+
"loss": 0.3197,
|
| 487 |
+
"step": 1675
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 1.0859150431172149,
|
| 491 |
+
"grad_norm": 2.679849624633789,
|
| 492 |
+
"learning_rate": 7.333333333333333e-06,
|
| 493 |
+
"loss": 0.3305,
|
| 494 |
+
"step": 1700
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 1.1018843819865858,
|
| 498 |
+
"grad_norm": 2.760922908782959,
|
| 499 |
+
"learning_rate": 7.277777777777778e-06,
|
| 500 |
+
"loss": 0.3333,
|
| 501 |
+
"step": 1725
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 1.1178537208559565,
|
| 505 |
+
"grad_norm": 2.7702419757843018,
|
| 506 |
+
"learning_rate": 7.222222222222223e-06,
|
| 507 |
+
"loss": 0.3452,
|
| 508 |
+
"step": 1750
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 1.1338230597253274,
|
| 512 |
+
"grad_norm": 3.0105628967285156,
|
| 513 |
+
"learning_rate": 7.166666666666667e-06,
|
| 514 |
+
"loss": 0.3318,
|
| 515 |
+
"step": 1775
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 1.1497923985946983,
|
| 519 |
+
"grad_norm": 3.0848121643066406,
|
| 520 |
+
"learning_rate": 7.111111111111112e-06,
|
| 521 |
+
"loss": 0.314,
|
| 522 |
+
"step": 1800
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 1.165761737464069,
|
| 526 |
+
"grad_norm": 3.0407936573028564,
|
| 527 |
+
"learning_rate": 7.055555555555557e-06,
|
| 528 |
+
"loss": 0.339,
|
| 529 |
+
"step": 1825
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 1.1817310763334399,
|
| 533 |
+
"grad_norm": 2.692070722579956,
|
| 534 |
+
"learning_rate": 7e-06,
|
| 535 |
+
"loss": 0.3301,
|
| 536 |
+
"step": 1850
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 1.1977004152028106,
|
| 540 |
+
"grad_norm": 2.7906270027160645,
|
| 541 |
+
"learning_rate": 6.944444444444445e-06,
|
| 542 |
+
"loss": 0.3243,
|
| 543 |
+
"step": 1875
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 1.2136697540721815,
|
| 547 |
+
"grad_norm": 3.005059003829956,
|
| 548 |
+
"learning_rate": 6.88888888888889e-06,
|
| 549 |
+
"loss": 0.3348,
|
| 550 |
+
"step": 1900
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 1.2296390929415522,
|
| 554 |
+
"grad_norm": 2.512002944946289,
|
| 555 |
+
"learning_rate": 6.833333333333334e-06,
|
| 556 |
+
"loss": 0.3282,
|
| 557 |
+
"step": 1925
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.245608431810923,
|
| 561 |
+
"grad_norm": 2.2834126949310303,
|
| 562 |
+
"learning_rate": 6.777777777777779e-06,
|
| 563 |
+
"loss": 0.3184,
|
| 564 |
+
"step": 1950
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.2615777706802938,
|
| 568 |
+
"grad_norm": 2.6210882663726807,
|
| 569 |
+
"learning_rate": 6.7222222222222235e-06,
|
| 570 |
+
"loss": 0.3223,
|
| 571 |
+
"step": 1975
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.2775471095496647,
|
| 575 |
+
"grad_norm": 2.5650439262390137,
|
| 576 |
+
"learning_rate": 6.666666666666667e-06,
|
| 577 |
+
"loss": 0.3335,
|
| 578 |
+
"step": 2000
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.2775471095496647,
|
| 582 |
+
"eval_loss": 0.41682446002960205,
|
| 583 |
+
"eval_runtime": 1686.343,
|
| 584 |
+
"eval_samples_per_second": 2.313,
|
| 585 |
+
"eval_steps_per_second": 0.145,
|
| 586 |
+
"eval_wer": 0.31297691490389257,
|
| 587 |
+
"step": 2000
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 1.2935164484190356,
|
| 591 |
+
"grad_norm": 3.5779247283935547,
|
| 592 |
+
"learning_rate": 6.6111111111111115e-06,
|
| 593 |
+
"loss": 0.3204,
|
| 594 |
+
"step": 2025
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 1.3094857872884063,
|
| 598 |
+
"grad_norm": 2.3297035694122314,
|
| 599 |
+
"learning_rate": 6.555555555555556e-06,
|
| 600 |
+
"loss": 0.3237,
|
| 601 |
+
"step": 2050
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 1.325455126157777,
|
| 605 |
+
"grad_norm": 2.9575600624084473,
|
| 606 |
+
"learning_rate": 6.5000000000000004e-06,
|
| 607 |
+
"loss": 0.3333,
|
| 608 |
+
"step": 2075
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 1.3414244650271478,
|
| 612 |
+
"grad_norm": 2.6735305786132812,
|
| 613 |
+
"learning_rate": 6.444444444444445e-06,
|
| 614 |
+
"loss": 0.3199,
|
| 615 |
+
"step": 2100
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 1.3573938038965188,
|
| 619 |
+
"grad_norm": 3.5485923290252686,
|
| 620 |
+
"learning_rate": 6.3888888888888885e-06,
|
| 621 |
+
"loss": 0.3148,
|
| 622 |
+
"step": 2125
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 1.3733631427658894,
|
| 626 |
+
"grad_norm": 2.704557180404663,
|
| 627 |
+
"learning_rate": 6.333333333333333e-06,
|
| 628 |
+
"loss": 0.3268,
|
| 629 |
+
"step": 2150
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 1.3893324816352604,
|
| 633 |
+
"grad_norm": 2.417282819747925,
|
| 634 |
+
"learning_rate": 6.277777777777778e-06,
|
| 635 |
+
"loss": 0.3286,
|
| 636 |
+
"step": 2175
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 1.405301820504631,
|
| 640 |
+
"grad_norm": 2.5613949298858643,
|
| 641 |
+
"learning_rate": 6.222222222222223e-06,
|
| 642 |
+
"loss": 0.3182,
|
| 643 |
+
"step": 2200
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 1.421271159374002,
|
| 647 |
+
"grad_norm": 3.0761849880218506,
|
| 648 |
+
"learning_rate": 6.166666666666667e-06,
|
| 649 |
+
"loss": 0.329,
|
| 650 |
+
"step": 2225
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 1.4372404982433729,
|
| 654 |
+
"grad_norm": 2.764951229095459,
|
| 655 |
+
"learning_rate": 6.111111111111112e-06,
|
| 656 |
+
"loss": 0.3157,
|
| 657 |
+
"step": 2250
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.4532098371127435,
|
| 661 |
+
"grad_norm": 2.909294605255127,
|
| 662 |
+
"learning_rate": 6.055555555555555e-06,
|
| 663 |
+
"loss": 0.3291,
|
| 664 |
+
"step": 2275
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 1.4691791759821142,
|
| 668 |
+
"grad_norm": 3.374781370162964,
|
| 669 |
+
"learning_rate": 6e-06,
|
| 670 |
+
"loss": 0.3325,
|
| 671 |
+
"step": 2300
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 1.4851485148514851,
|
| 675 |
+
"grad_norm": 2.8524813652038574,
|
| 676 |
+
"learning_rate": 5.944444444444445e-06,
|
| 677 |
+
"loss": 0.3182,
|
| 678 |
+
"step": 2325
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 1.501117853720856,
|
| 682 |
+
"grad_norm": 3.066960573196411,
|
| 683 |
+
"learning_rate": 5.88888888888889e-06,
|
| 684 |
+
"loss": 0.3125,
|
| 685 |
+
"step": 2350
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 1.5170871925902267,
|
| 689 |
+
"grad_norm": 2.604304790496826,
|
| 690 |
+
"learning_rate": 5.833333333333334e-06,
|
| 691 |
+
"loss": 0.335,
|
| 692 |
+
"step": 2375
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 1.5330565314595974,
|
| 696 |
+
"grad_norm": 2.784276008605957,
|
| 697 |
+
"learning_rate": 5.777777777777778e-06,
|
| 698 |
+
"loss": 0.3318,
|
| 699 |
+
"step": 2400
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 1.5490258703289683,
|
| 703 |
+
"grad_norm": 3.203049898147583,
|
| 704 |
+
"learning_rate": 5.722222222222222e-06,
|
| 705 |
+
"loss": 0.3208,
|
| 706 |
+
"step": 2425
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 1.5649952091983392,
|
| 710 |
+
"grad_norm": 2.73185658454895,
|
| 711 |
+
"learning_rate": 5.666666666666667e-06,
|
| 712 |
+
"loss": 0.2981,
|
| 713 |
+
"step": 2450
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 1.5809645480677101,
|
| 717 |
+
"grad_norm": 2.6829750537872314,
|
| 718 |
+
"learning_rate": 5.611111111111112e-06,
|
| 719 |
+
"loss": 0.3191,
|
| 720 |
+
"step": 2475
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 1.5969338869370808,
|
| 724 |
+
"grad_norm": 3.1976850032806396,
|
| 725 |
+
"learning_rate": 5.555555555555557e-06,
|
| 726 |
+
"loss": 0.3335,
|
| 727 |
+
"step": 2500
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 1.6129032258064515,
|
| 731 |
+
"grad_norm": 3.003782033920288,
|
| 732 |
+
"learning_rate": 5.500000000000001e-06,
|
| 733 |
+
"loss": 0.3013,
|
| 734 |
+
"step": 2525
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 1.6288725646758224,
|
| 738 |
+
"grad_norm": 2.897247076034546,
|
| 739 |
+
"learning_rate": 5.444444444444445e-06,
|
| 740 |
+
"loss": 0.3123,
|
| 741 |
+
"step": 2550
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 1.6448419035451933,
|
| 745 |
+
"grad_norm": 2.996609687805176,
|
| 746 |
+
"learning_rate": 5.388888888888889e-06,
|
| 747 |
+
"loss": 0.3182,
|
| 748 |
+
"step": 2575
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 1.660811242414564,
|
| 752 |
+
"grad_norm": 2.700706720352173,
|
| 753 |
+
"learning_rate": 5.333333333333334e-06,
|
| 754 |
+
"loss": 0.312,
|
| 755 |
+
"step": 2600
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 1.6767805812839347,
|
| 759 |
+
"grad_norm": 2.901888370513916,
|
| 760 |
+
"learning_rate": 5.2777777777777785e-06,
|
| 761 |
+
"loss": 0.3088,
|
| 762 |
+
"step": 2625
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 1.6927499201533056,
|
| 766 |
+
"grad_norm": 2.958951711654663,
|
| 767 |
+
"learning_rate": 5.2222222222222226e-06,
|
| 768 |
+
"loss": 0.3283,
|
| 769 |
+
"step": 2650
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 1.7087192590226765,
|
| 773 |
+
"grad_norm": 3.610682964324951,
|
| 774 |
+
"learning_rate": 5.1666666666666675e-06,
|
| 775 |
+
"loss": 0.3332,
|
| 776 |
+
"step": 2675
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 1.7246885978920474,
|
| 780 |
+
"grad_norm": 2.292649745941162,
|
| 781 |
+
"learning_rate": 5.1111111111111115e-06,
|
| 782 |
+
"loss": 0.2918,
|
| 783 |
+
"step": 2700
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 1.7406579367614181,
|
| 787 |
+
"grad_norm": 2.41726016998291,
|
| 788 |
+
"learning_rate": 5.0555555555555555e-06,
|
| 789 |
+
"loss": 0.3215,
|
| 790 |
+
"step": 2725
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 1.7566272756307888,
|
| 794 |
+
"grad_norm": 2.4622721672058105,
|
| 795 |
+
"learning_rate": 5e-06,
|
| 796 |
+
"loss": 0.3025,
|
| 797 |
+
"step": 2750
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 1.7725966145001597,
|
| 801 |
+
"grad_norm": 2.802022695541382,
|
| 802 |
+
"learning_rate": 4.944444444444445e-06,
|
| 803 |
+
"loss": 0.3008,
|
| 804 |
+
"step": 2775
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 1.7885659533695306,
|
| 808 |
+
"grad_norm": 2.8401265144348145,
|
| 809 |
+
"learning_rate": 4.888888888888889e-06,
|
| 810 |
+
"loss": 0.3065,
|
| 811 |
+
"step": 2800
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 1.8045352922389013,
|
| 815 |
+
"grad_norm": 2.6890196800231934,
|
| 816 |
+
"learning_rate": 4.833333333333333e-06,
|
| 817 |
+
"loss": 0.3052,
|
| 818 |
+
"step": 2825
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 1.820504631108272,
|
| 822 |
+
"grad_norm": 3.257754325866699,
|
| 823 |
+
"learning_rate": 4.777777777777778e-06,
|
| 824 |
+
"loss": 0.3243,
|
| 825 |
+
"step": 2850
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 1.836473969977643,
|
| 829 |
+
"grad_norm": 2.595125675201416,
|
| 830 |
+
"learning_rate": 4.722222222222222e-06,
|
| 831 |
+
"loss": 0.316,
|
| 832 |
+
"step": 2875
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 1.8524433088470138,
|
| 836 |
+
"grad_norm": 2.713214159011841,
|
| 837 |
+
"learning_rate": 4.666666666666667e-06,
|
| 838 |
+
"loss": 0.3162,
|
| 839 |
+
"step": 2900
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 1.8684126477163845,
|
| 843 |
+
"grad_norm": 2.4735445976257324,
|
| 844 |
+
"learning_rate": 4.611111111111112e-06,
|
| 845 |
+
"loss": 0.2955,
|
| 846 |
+
"step": 2925
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 1.8843819865857554,
|
| 850 |
+
"grad_norm": 3.026921033859253,
|
| 851 |
+
"learning_rate": 4.557777777777778e-06,
|
| 852 |
+
"loss": 0.3058,
|
| 853 |
+
"step": 2950
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 1.900351325455126,
|
| 857 |
+
"grad_norm": 2.5658116340637207,
|
| 858 |
+
"learning_rate": 4.502222222222223e-06,
|
| 859 |
+
"loss": 0.3107,
|
| 860 |
+
"step": 2975
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 1.916320664324497,
|
| 864 |
+
"grad_norm": 2.6040449142456055,
|
| 865 |
+
"learning_rate": 4.446666666666667e-06,
|
| 866 |
+
"loss": 0.3281,
|
| 867 |
+
"step": 3000
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 1.916320664324497,
|
| 871 |
+
"eval_loss": 0.3877977430820465,
|
| 872 |
+
"eval_runtime": 1665.3573,
|
| 873 |
+
"eval_samples_per_second": 2.342,
|
| 874 |
+
"eval_steps_per_second": 0.147,
|
| 875 |
+
"eval_wer": 0.2965324060658746,
|
| 876 |
+
"step": 3000
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 1.932290003193868,
|
| 880 |
+
"grad_norm": 2.7531864643096924,
|
| 881 |
+
"learning_rate": 4.391111111111112e-06,
|
| 882 |
+
"loss": 0.3129,
|
| 883 |
+
"step": 3025
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 1.9482593420632386,
|
| 887 |
+
"grad_norm": 2.1554644107818604,
|
| 888 |
+
"learning_rate": 4.3355555555555565e-06,
|
| 889 |
+
"loss": 0.2996,
|
| 890 |
+
"step": 3050
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 1.9642286809326093,
|
| 894 |
+
"grad_norm": 2.7987213134765625,
|
| 895 |
+
"learning_rate": 4.2800000000000005e-06,
|
| 896 |
+
"loss": 0.3048,
|
| 897 |
+
"step": 3075
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 1.9801980198019802,
|
| 901 |
+
"grad_norm": 2.697014093399048,
|
| 902 |
+
"learning_rate": 4.2244444444444446e-06,
|
| 903 |
+
"loss": 0.3101,
|
| 904 |
+
"step": 3100
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 1.996167358671351,
|
| 908 |
+
"grad_norm": 2.2209935188293457,
|
| 909 |
+
"learning_rate": 4.168888888888889e-06,
|
| 910 |
+
"loss": 0.2935,
|
| 911 |
+
"step": 3125
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 2.012136697540722,
|
| 915 |
+
"grad_norm": 2.877319574356079,
|
| 916 |
+
"learning_rate": 4.1133333333333335e-06,
|
| 917 |
+
"loss": 0.2273,
|
| 918 |
+
"step": 3150
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 2.0281060364100925,
|
| 922 |
+
"grad_norm": 2.122941732406616,
|
| 923 |
+
"learning_rate": 4.057777777777778e-06,
|
| 924 |
+
"loss": 0.1945,
|
| 925 |
+
"step": 3175
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 2.0440753752794634,
|
| 929 |
+
"grad_norm": 2.2523446083068848,
|
| 930 |
+
"learning_rate": 4.002222222222222e-06,
|
| 931 |
+
"loss": 0.198,
|
| 932 |
+
"step": 3200
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 2.0600447141488343,
|
| 936 |
+
"grad_norm": 2.1842360496520996,
|
| 937 |
+
"learning_rate": 3.946666666666667e-06,
|
| 938 |
+
"loss": 0.1964,
|
| 939 |
+
"step": 3225
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 2.076014053018205,
|
| 943 |
+
"grad_norm": 2.5835633277893066,
|
| 944 |
+
"learning_rate": 3.891111111111111e-06,
|
| 945 |
+
"loss": 0.2093,
|
| 946 |
+
"step": 3250
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 2.0919833918875756,
|
| 950 |
+
"grad_norm": 2.534207344055176,
|
| 951 |
+
"learning_rate": 3.835555555555555e-06,
|
| 952 |
+
"loss": 0.193,
|
| 953 |
+
"step": 3275
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 2.1079527307569466,
|
| 957 |
+
"grad_norm": 3.2382678985595703,
|
| 958 |
+
"learning_rate": 3.7800000000000002e-06,
|
| 959 |
+
"loss": 0.2066,
|
| 960 |
+
"step": 3300
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 2.1239220696263175,
|
| 964 |
+
"grad_norm": 1.761076807975769,
|
| 965 |
+
"learning_rate": 3.724444444444445e-06,
|
| 966 |
+
"loss": 0.2032,
|
| 967 |
+
"step": 3325
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 2.1398914084956884,
|
| 971 |
+
"grad_norm": 2.6212522983551025,
|
| 972 |
+
"learning_rate": 3.668888888888889e-06,
|
| 973 |
+
"loss": 0.2018,
|
| 974 |
+
"step": 3350
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 2.1558607473650593,
|
| 978 |
+
"grad_norm": 2.240907669067383,
|
| 979 |
+
"learning_rate": 3.6133333333333336e-06,
|
| 980 |
+
"loss": 0.203,
|
| 981 |
+
"step": 3375
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 2.1718300862344297,
|
| 985 |
+
"grad_norm": 2.5074503421783447,
|
| 986 |
+
"learning_rate": 3.5577777777777785e-06,
|
| 987 |
+
"loss": 0.2029,
|
| 988 |
+
"step": 3400
|
| 989 |
+
},
|
| 990 |
+
{
|
| 991 |
+
"epoch": 2.1877994251038007,
|
| 992 |
+
"grad_norm": 2.3992538452148438,
|
| 993 |
+
"learning_rate": 3.5022222222222225e-06,
|
| 994 |
+
"loss": 0.1975,
|
| 995 |
+
"step": 3425
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 2.2037687639731716,
|
| 999 |
+
"grad_norm": 2.3788444995880127,
|
| 1000 |
+
"learning_rate": 3.446666666666667e-06,
|
| 1001 |
+
"loss": 0.2004,
|
| 1002 |
+
"step": 3450
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 2.2197381028425425,
|
| 1006 |
+
"grad_norm": 2.7057225704193115,
|
| 1007 |
+
"learning_rate": 3.391111111111111e-06,
|
| 1008 |
+
"loss": 0.1956,
|
| 1009 |
+
"step": 3475
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 2.235707441711913,
|
| 1013 |
+
"grad_norm": 2.399021863937378,
|
| 1014 |
+
"learning_rate": 3.335555555555556e-06,
|
| 1015 |
+
"loss": 0.2033,
|
| 1016 |
+
"step": 3500
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 2.251676780581284,
|
| 1020 |
+
"grad_norm": 2.3665523529052734,
|
| 1021 |
+
"learning_rate": 3.2800000000000004e-06,
|
| 1022 |
+
"loss": 0.203,
|
| 1023 |
+
"step": 3525
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 2.2676461194506548,
|
| 1027 |
+
"grad_norm": 2.4822895526885986,
|
| 1028 |
+
"learning_rate": 3.2244444444444444e-06,
|
| 1029 |
+
"loss": 0.1991,
|
| 1030 |
+
"step": 3550
|
| 1031 |
+
},
|
| 1032 |
+
{
|
| 1033 |
+
"epoch": 2.2836154583200257,
|
| 1034 |
+
"grad_norm": 2.2787749767303467,
|
| 1035 |
+
"learning_rate": 3.1688888888888893e-06,
|
| 1036 |
+
"loss": 0.2024,
|
| 1037 |
+
"step": 3575
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 2.2995847971893966,
|
| 1041 |
+
"grad_norm": 2.3927159309387207,
|
| 1042 |
+
"learning_rate": 3.1133333333333337e-06,
|
| 1043 |
+
"loss": 0.1928,
|
| 1044 |
+
"step": 3600
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 2.315554136058767,
|
| 1048 |
+
"grad_norm": 2.311298370361328,
|
| 1049 |
+
"learning_rate": 3.0577777777777778e-06,
|
| 1050 |
+
"loss": 0.2231,
|
| 1051 |
+
"step": 3625
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 2.331523474928138,
|
| 1055 |
+
"grad_norm": 2.4807193279266357,
|
| 1056 |
+
"learning_rate": 3.0022222222222227e-06,
|
| 1057 |
+
"loss": 0.1938,
|
| 1058 |
+
"step": 3650
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 2.347492813797509,
|
| 1062 |
+
"grad_norm": 2.698840856552124,
|
| 1063 |
+
"learning_rate": 2.946666666666667e-06,
|
| 1064 |
+
"loss": 0.1954,
|
| 1065 |
+
"step": 3675
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 2.3634621526668798,
|
| 1069 |
+
"grad_norm": 2.193194627761841,
|
| 1070 |
+
"learning_rate": 2.891111111111111e-06,
|
| 1071 |
+
"loss": 0.2013,
|
| 1072 |
+
"step": 3700
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"epoch": 2.37943149153625,
|
| 1076 |
+
"grad_norm": 2.5082945823669434,
|
| 1077 |
+
"learning_rate": 2.835555555555556e-06,
|
| 1078 |
+
"loss": 0.1963,
|
| 1079 |
+
"step": 3725
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 2.395400830405621,
|
| 1083 |
+
"grad_norm": 2.7638180255889893,
|
| 1084 |
+
"learning_rate": 2.7800000000000005e-06,
|
| 1085 |
+
"loss": 0.1919,
|
| 1086 |
+
"step": 3750
|
| 1087 |
+
},
|
| 1088 |
+
{
|
| 1089 |
+
"epoch": 2.411370169274992,
|
| 1090 |
+
"grad_norm": 2.6804022789001465,
|
| 1091 |
+
"learning_rate": 2.7244444444444445e-06,
|
| 1092 |
+
"loss": 0.2008,
|
| 1093 |
+
"step": 3775
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"epoch": 2.427339508144363,
|
| 1097 |
+
"grad_norm": 2.131471872329712,
|
| 1098 |
+
"learning_rate": 2.6688888888888894e-06,
|
| 1099 |
+
"loss": 0.1912,
|
| 1100 |
+
"step": 3800
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 2.443308847013734,
|
| 1104 |
+
"grad_norm": 2.964355707168579,
|
| 1105 |
+
"learning_rate": 2.6133333333333334e-06,
|
| 1106 |
+
"loss": 0.1943,
|
| 1107 |
+
"step": 3825
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 2.4592781858831043,
|
| 1111 |
+
"grad_norm": 2.4776132106781006,
|
| 1112 |
+
"learning_rate": 2.557777777777778e-06,
|
| 1113 |
+
"loss": 0.1916,
|
| 1114 |
+
"step": 3850
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 2.4752475247524752,
|
| 1118 |
+
"grad_norm": 2.556891441345215,
|
| 1119 |
+
"learning_rate": 2.5022222222222224e-06,
|
| 1120 |
+
"loss": 0.2041,
|
| 1121 |
+
"step": 3875
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 2.491216863621846,
|
| 1125 |
+
"grad_norm": 2.0888752937316895,
|
| 1126 |
+
"learning_rate": 2.446666666666667e-06,
|
| 1127 |
+
"loss": 0.2026,
|
| 1128 |
+
"step": 3900
|
| 1129 |
+
},
|
| 1130 |
+
{
|
| 1131 |
+
"epoch": 2.5071862024912166,
|
| 1132 |
+
"grad_norm": 2.438257932662964,
|
| 1133 |
+
"learning_rate": 2.3911111111111113e-06,
|
| 1134 |
+
"loss": 0.2017,
|
| 1135 |
+
"step": 3925
|
| 1136 |
+
},
|
| 1137 |
+
{
|
| 1138 |
+
"epoch": 2.5231555413605875,
|
| 1139 |
+
"grad_norm": 2.6804213523864746,
|
| 1140 |
+
"learning_rate": 2.3355555555555557e-06,
|
| 1141 |
+
"loss": 0.1957,
|
| 1142 |
+
"step": 3950
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 2.5391248802299584,
|
| 1146 |
+
"grad_norm": 2.409219264984131,
|
| 1147 |
+
"learning_rate": 2.28e-06,
|
| 1148 |
+
"loss": 0.201,
|
| 1149 |
+
"step": 3975
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 2.5550942190993293,
|
| 1153 |
+
"grad_norm": 2.4475369453430176,
|
| 1154 |
+
"learning_rate": 2.2244444444444447e-06,
|
| 1155 |
+
"loss": 0.1901,
|
| 1156 |
+
"step": 4000
|
| 1157 |
+
},
|
| 1158 |
+
{
|
| 1159 |
+
"epoch": 2.5550942190993293,
|
| 1160 |
+
"eval_loss": 0.392954021692276,
|
| 1161 |
+
"eval_runtime": 1661.0812,
|
| 1162 |
+
"eval_samples_per_second": 2.348,
|
| 1163 |
+
"eval_steps_per_second": 0.147,
|
| 1164 |
+
"eval_wer": 0.2848932676518883,
|
| 1165 |
+
"step": 4000
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 2.5710635579687002,
|
| 1169 |
+
"grad_norm": 2.4416608810424805,
|
| 1170 |
+
"learning_rate": 2.168888888888889e-06,
|
| 1171 |
+
"loss": 0.1886,
|
| 1172 |
+
"step": 4025
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 2.587032896838071,
|
| 1176 |
+
"grad_norm": 2.4266738891601562,
|
| 1177 |
+
"learning_rate": 2.1133333333333336e-06,
|
| 1178 |
+
"loss": 0.1964,
|
| 1179 |
+
"step": 4050
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 2.6030022357074416,
|
| 1183 |
+
"grad_norm": 2.2254934310913086,
|
| 1184 |
+
"learning_rate": 2.057777777777778e-06,
|
| 1185 |
+
"loss": 0.193,
|
| 1186 |
+
"step": 4075
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 2.6189715745768125,
|
| 1190 |
+
"grad_norm": 2.269014596939087,
|
| 1191 |
+
"learning_rate": 2.0022222222222225e-06,
|
| 1192 |
+
"loss": 0.1911,
|
| 1193 |
+
"step": 4100
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 2.6349409134461834,
|
| 1197 |
+
"grad_norm": 2.8170831203460693,
|
| 1198 |
+
"learning_rate": 1.9466666666666665e-06,
|
| 1199 |
+
"loss": 0.1891,
|
| 1200 |
+
"step": 4125
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 2.650910252315554,
|
| 1204 |
+
"grad_norm": 2.5246517658233643,
|
| 1205 |
+
"learning_rate": 1.8911111111111114e-06,
|
| 1206 |
+
"loss": 0.2021,
|
| 1207 |
+
"step": 4150
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 2.666879591184925,
|
| 1211 |
+
"grad_norm": 2.7256410121917725,
|
| 1212 |
+
"learning_rate": 1.8355555555555557e-06,
|
| 1213 |
+
"loss": 0.1924,
|
| 1214 |
+
"step": 4175
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 2.6828489300542957,
|
| 1218 |
+
"grad_norm": 2.446582555770874,
|
| 1219 |
+
"learning_rate": 1.7800000000000001e-06,
|
| 1220 |
+
"loss": 0.1934,
|
| 1221 |
+
"step": 4200
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 2.6988182689236666,
|
| 1225 |
+
"grad_norm": 2.602579116821289,
|
| 1226 |
+
"learning_rate": 1.7244444444444448e-06,
|
| 1227 |
+
"loss": 0.188,
|
| 1228 |
+
"step": 4225
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 2.7147876077930375,
|
| 1232 |
+
"grad_norm": 2.2500104904174805,
|
| 1233 |
+
"learning_rate": 1.668888888888889e-06,
|
| 1234 |
+
"loss": 0.1951,
|
| 1235 |
+
"step": 4250
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 2.7307569466624084,
|
| 1239 |
+
"grad_norm": 2.7052624225616455,
|
| 1240 |
+
"learning_rate": 1.6133333333333335e-06,
|
| 1241 |
+
"loss": 0.2039,
|
| 1242 |
+
"step": 4275
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 2.746726285531779,
|
| 1246 |
+
"grad_norm": 2.827786922454834,
|
| 1247 |
+
"learning_rate": 1.5577777777777777e-06,
|
| 1248 |
+
"loss": 0.1859,
|
| 1249 |
+
"step": 4300
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 2.76269562440115,
|
| 1253 |
+
"grad_norm": 2.8331093788146973,
|
| 1254 |
+
"learning_rate": 1.5022222222222224e-06,
|
| 1255 |
+
"loss": 0.2028,
|
| 1256 |
+
"step": 4325
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 2.7786649632705207,
|
| 1260 |
+
"grad_norm": 2.4881205558776855,
|
| 1261 |
+
"learning_rate": 1.4466666666666669e-06,
|
| 1262 |
+
"loss": 0.2013,
|
| 1263 |
+
"step": 4350
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 2.794634302139891,
|
| 1267 |
+
"grad_norm": 2.725170612335205,
|
| 1268 |
+
"learning_rate": 1.3911111111111111e-06,
|
| 1269 |
+
"loss": 0.202,
|
| 1270 |
+
"step": 4375
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 2.810603641009262,
|
| 1274 |
+
"grad_norm": 2.9598212242126465,
|
| 1275 |
+
"learning_rate": 1.3355555555555558e-06,
|
| 1276 |
+
"loss": 0.1957,
|
| 1277 |
+
"step": 4400
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 2.826572979878633,
|
| 1281 |
+
"grad_norm": 2.1554248332977295,
|
| 1282 |
+
"learning_rate": 1.28e-06,
|
| 1283 |
+
"loss": 0.185,
|
| 1284 |
+
"step": 4425
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 2.842542318748004,
|
| 1288 |
+
"grad_norm": 2.447131872177124,
|
| 1289 |
+
"learning_rate": 1.2244444444444445e-06,
|
| 1290 |
+
"loss": 0.197,
|
| 1291 |
+
"step": 4450
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 2.858511657617375,
|
| 1295 |
+
"grad_norm": 2.1197404861450195,
|
| 1296 |
+
"learning_rate": 1.168888888888889e-06,
|
| 1297 |
+
"loss": 0.189,
|
| 1298 |
+
"step": 4475
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 2.8744809964867457,
|
| 1302 |
+
"grad_norm": 2.2641892433166504,
|
| 1303 |
+
"learning_rate": 1.1133333333333334e-06,
|
| 1304 |
+
"loss": 0.1825,
|
| 1305 |
+
"step": 4500
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 2.890450335356116,
|
| 1309 |
+
"grad_norm": 2.380861759185791,
|
| 1310 |
+
"learning_rate": 1.0577777777777779e-06,
|
| 1311 |
+
"loss": 0.186,
|
| 1312 |
+
"step": 4525
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 2.906419674225487,
|
| 1316 |
+
"grad_norm": 2.575404167175293,
|
| 1317 |
+
"learning_rate": 1.0022222222222223e-06,
|
| 1318 |
+
"loss": 0.194,
|
| 1319 |
+
"step": 4550
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 2.922389013094858,
|
| 1323 |
+
"grad_norm": 1.9462391138076782,
|
| 1324 |
+
"learning_rate": 9.466666666666667e-07,
|
| 1325 |
+
"loss": 0.2058,
|
| 1326 |
+
"step": 4575
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 2.9383583519642285,
|
| 1330 |
+
"grad_norm": 2.332577705383301,
|
| 1331 |
+
"learning_rate": 8.911111111111112e-07,
|
| 1332 |
+
"loss": 0.1864,
|
| 1333 |
+
"step": 4600
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 2.9543276908335994,
|
| 1337 |
+
"grad_norm": 2.345797300338745,
|
| 1338 |
+
"learning_rate": 8.355555555555556e-07,
|
| 1339 |
+
"loss": 0.1955,
|
| 1340 |
+
"step": 4625
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 2.9702970297029703,
|
| 1344 |
+
"grad_norm": 2.41668438911438,
|
| 1345 |
+
"learning_rate": 7.8e-07,
|
| 1346 |
+
"loss": 0.1804,
|
| 1347 |
+
"step": 4650
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 2.986266368572341,
|
| 1351 |
+
"grad_norm": 2.3288943767547607,
|
| 1352 |
+
"learning_rate": 7.244444444444446e-07,
|
| 1353 |
+
"loss": 0.1911,
|
| 1354 |
+
"step": 4675
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 3.002235707441712,
|
| 1358 |
+
"grad_norm": 1.3726705312728882,
|
| 1359 |
+
"learning_rate": 6.68888888888889e-07,
|
| 1360 |
+
"loss": 0.1877,
|
| 1361 |
+
"step": 4700
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 3.0182050463110826,
|
| 1365 |
+
"grad_norm": 2.3931479454040527,
|
| 1366 |
+
"learning_rate": 6.133333333333333e-07,
|
| 1367 |
+
"loss": 0.1261,
|
| 1368 |
+
"step": 4725
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 3.0341743851804535,
|
| 1372 |
+
"grad_norm": 1.6263169050216675,
|
| 1373 |
+
"learning_rate": 5.577777777777779e-07,
|
| 1374 |
+
"loss": 0.1247,
|
| 1375 |
+
"step": 4750
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 3.0501437240498244,
|
| 1379 |
+
"grad_norm": 1.9789073467254639,
|
| 1380 |
+
"learning_rate": 5.022222222222222e-07,
|
| 1381 |
+
"loss": 0.1259,
|
| 1382 |
+
"step": 4775
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 3.0661130629191953,
|
| 1386 |
+
"grad_norm": 2.2936391830444336,
|
| 1387 |
+
"learning_rate": 4.466666666666667e-07,
|
| 1388 |
+
"loss": 0.1318,
|
| 1389 |
+
"step": 4800
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 3.0820824017885657,
|
| 1393 |
+
"grad_norm": 2.009859800338745,
|
| 1394 |
+
"learning_rate": 3.9111111111111115e-07,
|
| 1395 |
+
"loss": 0.1346,
|
| 1396 |
+
"step": 4825
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 3.0980517406579366,
|
| 1400 |
+
"grad_norm": 2.502535104751587,
|
| 1401 |
+
"learning_rate": 3.3555555555555556e-07,
|
| 1402 |
+
"loss": 0.1312,
|
| 1403 |
+
"step": 4850
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 3.1140210795273076,
|
| 1407 |
+
"grad_norm": 2.6293087005615234,
|
| 1408 |
+
"learning_rate": 2.8e-07,
|
| 1409 |
+
"loss": 0.1277,
|
| 1410 |
+
"step": 4875
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 3.1299904183966785,
|
| 1414 |
+
"grad_norm": 2.0969161987304688,
|
| 1415 |
+
"learning_rate": 2.2444444444444445e-07,
|
| 1416 |
+
"loss": 0.1301,
|
| 1417 |
+
"step": 4900
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 3.1459597572660494,
|
| 1421 |
+
"grad_norm": 1.9938933849334717,
|
| 1422 |
+
"learning_rate": 1.6888888888888888e-07,
|
| 1423 |
+
"loss": 0.129,
|
| 1424 |
+
"step": 4925
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 3.16192909613542,
|
| 1428 |
+
"grad_norm": 2.1169307231903076,
|
| 1429 |
+
"learning_rate": 1.1333333333333336e-07,
|
| 1430 |
+
"loss": 0.134,
|
| 1431 |
+
"step": 4950
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 3.1778984350047907,
|
| 1435 |
+
"grad_norm": 2.5867435932159424,
|
| 1436 |
+
"learning_rate": 5.777777777777778e-08,
|
| 1437 |
+
"loss": 0.1379,
|
| 1438 |
+
"step": 4975
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 3.1938677738741617,
|
| 1442 |
+
"grad_norm": 2.1530654430389404,
|
| 1443 |
+
"learning_rate": 2.2222222222222225e-09,
|
| 1444 |
+
"loss": 0.1306,
|
| 1445 |
+
"step": 5000
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 3.1938677738741617,
|
| 1449 |
+
"eval_loss": 0.40369176864624023,
|
| 1450 |
+
"eval_runtime": 1644.4554,
|
| 1451 |
+
"eval_samples_per_second": 2.372,
|
| 1452 |
+
"eval_steps_per_second": 0.148,
|
| 1453 |
+
"eval_wer": 0.27909784603496574,
|
| 1454 |
+
"step": 5000
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 3.1938677738741617,
|
| 1458 |
+
"step": 5000,
|
| 1459 |
+
"total_flos": 5.435895365546803e+20,
|
| 1460 |
+
"train_loss": 0.3537434986591339,
|
| 1461 |
+
"train_runtime": 57744.1557,
|
| 1462 |
+
"train_samples_per_second": 2.771,
|
| 1463 |
+
"train_steps_per_second": 0.087
|
| 1464 |
+
}
|
| 1465 |
+
],
|
| 1466 |
+
"logging_steps": 25,
|
| 1467 |
+
"max_steps": 5000,
|
| 1468 |
+
"num_input_tokens_seen": 0,
|
| 1469 |
+
"num_train_epochs": 4,
|
| 1470 |
+
"save_steps": 1000,
|
| 1471 |
+
"stateful_callbacks": {
|
| 1472 |
+
"TrainerControl": {
|
| 1473 |
+
"args": {
|
| 1474 |
+
"should_epoch_stop": false,
|
| 1475 |
+
"should_evaluate": false,
|
| 1476 |
+
"should_log": false,
|
| 1477 |
+
"should_save": true,
|
| 1478 |
+
"should_training_stop": true
|
| 1479 |
+
},
|
| 1480 |
+
"attributes": {}
|
| 1481 |
+
}
|
| 1482 |
+
},
|
| 1483 |
+
"total_flos": 5.435895365546803e+20,
|
| 1484 |
+
"train_batch_size": 16,
|
| 1485 |
+
"trial_name": null,
|
| 1486 |
+
"trial_params": null
|
| 1487 |
+
}
|