End of training
Browse files- README.md +1 -1
- all_results.json +15 -0
- eval_results.json +9 -0
- runs/Oct04_07-16-17_a99ffc333f86/events.out.tfevents.1728085226.a99ffc333f86.35.1 +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-cv-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: 0.3838
|
| 21 |
- Wer: 0.2732
|
|
|
|
| 15 |
|
| 16 |
# whisper-large-v3-ft-btb-cv-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, DewiBrynJones/commonvoice_18_0_cy train+dev+test main dataset.
|
| 19 |
It achieves the following results on the evaluation set:
|
| 20 |
- Loss: 0.3838
|
| 21 |
- Wer: 0.2732
|
all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.8555111364934325,
|
| 3 |
+
"eval_loss": 0.38383349776268005,
|
| 4 |
+
"eval_runtime": 1673.49,
|
| 5 |
+
"eval_samples": 3901,
|
| 6 |
+
"eval_samples_per_second": 2.331,
|
| 7 |
+
"eval_steps_per_second": 0.146,
|
| 8 |
+
"eval_wer": 0.27318168646769053,
|
| 9 |
+
"total_flos": 5.435589590699213e+20,
|
| 10 |
+
"train_loss": 0.3002769865989685,
|
| 11 |
+
"train_runtime": 59305.2217,
|
| 12 |
+
"train_samples": 56026,
|
| 13 |
+
"train_samples_per_second": 2.698,
|
| 14 |
+
"train_steps_per_second": 0.084
|
| 15 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.8555111364934325,
|
| 3 |
+
"eval_loss": 0.38383349776268005,
|
| 4 |
+
"eval_runtime": 1673.49,
|
| 5 |
+
"eval_samples": 3901,
|
| 6 |
+
"eval_samples_per_second": 2.331,
|
| 7 |
+
"eval_steps_per_second": 0.146,
|
| 8 |
+
"eval_wer": 0.27318168646769053
|
| 9 |
+
}
|
runs/Oct04_07-16-17_a99ffc333f86/events.out.tfevents.1728085226.a99ffc333f86.35.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:904261755ca0d94d881ed13df66b1550c5a8d171c4b4e9b0aa1bc71f4e005027
|
| 3 |
+
size 406
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.8555111364934325,
|
| 3 |
+
"total_flos": 5.435589590699213e+20,
|
| 4 |
+
"train_loss": 0.3002769865989685,
|
| 5 |
+
"train_runtime": 59305.2217,
|
| 6 |
+
"train_samples": 56026,
|
| 7 |
+
"train_samples_per_second": 2.698,
|
| 8 |
+
"train_steps_per_second": 0.084
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1487 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.8555111364934325,
|
| 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": 0.014277555682467162,
|
| 13 |
+
"grad_norm": 7.571424961090088,
|
| 14 |
+
"learning_rate": 5.000000000000001e-07,
|
| 15 |
+
"loss": 1.5088,
|
| 16 |
+
"step": 25
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.028555111364934323,
|
| 20 |
+
"grad_norm": 5.992729187011719,
|
| 21 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 22 |
+
"loss": 1.2038,
|
| 23 |
+
"step": 50
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.04283266704740148,
|
| 27 |
+
"grad_norm": 5.949503421783447,
|
| 28 |
+
"learning_rate": 1.5e-06,
|
| 29 |
+
"loss": 0.8879,
|
| 30 |
+
"step": 75
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.05711022272986865,
|
| 34 |
+
"grad_norm": 4.452832221984863,
|
| 35 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 36 |
+
"loss": 0.7647,
|
| 37 |
+
"step": 100
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.0713877784123358,
|
| 41 |
+
"grad_norm": 4.690545558929443,
|
| 42 |
+
"learning_rate": 2.5e-06,
|
| 43 |
+
"loss": 0.6792,
|
| 44 |
+
"step": 125
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.08566533409480297,
|
| 48 |
+
"grad_norm": 4.969720840454102,
|
| 49 |
+
"learning_rate": 3e-06,
|
| 50 |
+
"loss": 0.6549,
|
| 51 |
+
"step": 150
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.09994288977727013,
|
| 55 |
+
"grad_norm": 5.184281349182129,
|
| 56 |
+
"learning_rate": 3.5e-06,
|
| 57 |
+
"loss": 0.6376,
|
| 58 |
+
"step": 175
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.1142204454597373,
|
| 62 |
+
"grad_norm": 5.00349235534668,
|
| 63 |
+
"learning_rate": 4.000000000000001e-06,
|
| 64 |
+
"loss": 0.5982,
|
| 65 |
+
"step": 200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.12849800114220444,
|
| 69 |
+
"grad_norm": 4.239490032196045,
|
| 70 |
+
"learning_rate": 4.5e-06,
|
| 71 |
+
"loss": 0.6084,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.1427755568246716,
|
| 76 |
+
"grad_norm": 4.2740068435668945,
|
| 77 |
+
"learning_rate": 5e-06,
|
| 78 |
+
"loss": 0.58,
|
| 79 |
+
"step": 250
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.15705311250713877,
|
| 83 |
+
"grad_norm": 4.718848705291748,
|
| 84 |
+
"learning_rate": 5.500000000000001e-06,
|
| 85 |
+
"loss": 0.5759,
|
| 86 |
+
"step": 275
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.17133066818960593,
|
| 90 |
+
"grad_norm": 4.2935638427734375,
|
| 91 |
+
"learning_rate": 6e-06,
|
| 92 |
+
"loss": 0.5625,
|
| 93 |
+
"step": 300
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.1856082238720731,
|
| 97 |
+
"grad_norm": 4.917020797729492,
|
| 98 |
+
"learning_rate": 6.5000000000000004e-06,
|
| 99 |
+
"loss": 0.5621,
|
| 100 |
+
"step": 325
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.19988577955454026,
|
| 104 |
+
"grad_norm": 3.9521942138671875,
|
| 105 |
+
"learning_rate": 7e-06,
|
| 106 |
+
"loss": 0.5644,
|
| 107 |
+
"step": 350
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.21416333523700742,
|
| 111 |
+
"grad_norm": 4.506232738494873,
|
| 112 |
+
"learning_rate": 7.500000000000001e-06,
|
| 113 |
+
"loss": 0.5508,
|
| 114 |
+
"step": 375
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.2284408909194746,
|
| 118 |
+
"grad_norm": 4.1483540534973145,
|
| 119 |
+
"learning_rate": 8.000000000000001e-06,
|
| 120 |
+
"loss": 0.5244,
|
| 121 |
+
"step": 400
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.24271844660194175,
|
| 125 |
+
"grad_norm": 4.077396392822266,
|
| 126 |
+
"learning_rate": 8.5e-06,
|
| 127 |
+
"loss": 0.5051,
|
| 128 |
+
"step": 425
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.2569960022844089,
|
| 132 |
+
"grad_norm": 4.375626087188721,
|
| 133 |
+
"learning_rate": 9e-06,
|
| 134 |
+
"loss": 0.5222,
|
| 135 |
+
"step": 450
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.2712735579668761,
|
| 139 |
+
"grad_norm": 3.5698530673980713,
|
| 140 |
+
"learning_rate": 9.5e-06,
|
| 141 |
+
"loss": 0.5038,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.2855511136493432,
|
| 146 |
+
"grad_norm": 4.99509859085083,
|
| 147 |
+
"learning_rate": 1e-05,
|
| 148 |
+
"loss": 0.5196,
|
| 149 |
+
"step": 500
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.2998286693318104,
|
| 153 |
+
"grad_norm": 3.666332721710205,
|
| 154 |
+
"learning_rate": 9.944444444444445e-06,
|
| 155 |
+
"loss": 0.5066,
|
| 156 |
+
"step": 525
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.31410622501427754,
|
| 160 |
+
"grad_norm": 3.9203736782073975,
|
| 161 |
+
"learning_rate": 9.88888888888889e-06,
|
| 162 |
+
"loss": 0.4822,
|
| 163 |
+
"step": 550
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.32838378069674473,
|
| 167 |
+
"grad_norm": 3.5677530765533447,
|
| 168 |
+
"learning_rate": 9.833333333333333e-06,
|
| 169 |
+
"loss": 0.519,
|
| 170 |
+
"step": 575
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.34266133637921187,
|
| 174 |
+
"grad_norm": 3.3873414993286133,
|
| 175 |
+
"learning_rate": 9.777777777777779e-06,
|
| 176 |
+
"loss": 0.5205,
|
| 177 |
+
"step": 600
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.35693889206167906,
|
| 181 |
+
"grad_norm": 3.9527816772460938,
|
| 182 |
+
"learning_rate": 9.722222222222223e-06,
|
| 183 |
+
"loss": 0.4769,
|
| 184 |
+
"step": 625
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.3712164477441462,
|
| 188 |
+
"grad_norm": 3.3437490463256836,
|
| 189 |
+
"learning_rate": 9.666666666666667e-06,
|
| 190 |
+
"loss": 0.4629,
|
| 191 |
+
"step": 650
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.3854940034266134,
|
| 195 |
+
"grad_norm": 3.7754790782928467,
|
| 196 |
+
"learning_rate": 9.611111111111112e-06,
|
| 197 |
+
"loss": 0.4812,
|
| 198 |
+
"step": 675
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.3997715591090805,
|
| 202 |
+
"grad_norm": 3.744267225265503,
|
| 203 |
+
"learning_rate": 9.555555555555556e-06,
|
| 204 |
+
"loss": 0.467,
|
| 205 |
+
"step": 700
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.4140491147915477,
|
| 209 |
+
"grad_norm": 3.5076072216033936,
|
| 210 |
+
"learning_rate": 9.5e-06,
|
| 211 |
+
"loss": 0.4454,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.42832667047401485,
|
| 216 |
+
"grad_norm": 3.556335687637329,
|
| 217 |
+
"learning_rate": 9.444444444444445e-06,
|
| 218 |
+
"loss": 0.4447,
|
| 219 |
+
"step": 750
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.442604226156482,
|
| 223 |
+
"grad_norm": 4.256951332092285,
|
| 224 |
+
"learning_rate": 9.38888888888889e-06,
|
| 225 |
+
"loss": 0.4809,
|
| 226 |
+
"step": 775
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.4568817818389492,
|
| 230 |
+
"grad_norm": 3.533447742462158,
|
| 231 |
+
"learning_rate": 9.333333333333334e-06,
|
| 232 |
+
"loss": 0.4425,
|
| 233 |
+
"step": 800
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.4711593375214163,
|
| 237 |
+
"grad_norm": 4.324098587036133,
|
| 238 |
+
"learning_rate": 9.277777777777778e-06,
|
| 239 |
+
"loss": 0.424,
|
| 240 |
+
"step": 825
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.4854368932038835,
|
| 244 |
+
"grad_norm": 2.913189649581909,
|
| 245 |
+
"learning_rate": 9.222222222222224e-06,
|
| 246 |
+
"loss": 0.4314,
|
| 247 |
+
"step": 850
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.49971444888635064,
|
| 251 |
+
"grad_norm": 3.432490825653076,
|
| 252 |
+
"learning_rate": 9.166666666666666e-06,
|
| 253 |
+
"loss": 0.4355,
|
| 254 |
+
"step": 875
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.5139920045688178,
|
| 258 |
+
"grad_norm": 3.645869255065918,
|
| 259 |
+
"learning_rate": 9.111111111111112e-06,
|
| 260 |
+
"loss": 0.4395,
|
| 261 |
+
"step": 900
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.528269560251285,
|
| 265 |
+
"grad_norm": 3.2094240188598633,
|
| 266 |
+
"learning_rate": 9.055555555555556e-06,
|
| 267 |
+
"loss": 0.4144,
|
| 268 |
+
"step": 925
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.5425471159337522,
|
| 272 |
+
"grad_norm": 3.4623546600341797,
|
| 273 |
+
"learning_rate": 9e-06,
|
| 274 |
+
"loss": 0.4277,
|
| 275 |
+
"step": 950
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.5568246716162193,
|
| 279 |
+
"grad_norm": 3.640333414077759,
|
| 280 |
+
"learning_rate": 8.944444444444446e-06,
|
| 281 |
+
"loss": 0.4246,
|
| 282 |
+
"step": 975
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.5711022272986864,
|
| 286 |
+
"grad_norm": 3.0283167362213135,
|
| 287 |
+
"learning_rate": 8.888888888888888e-06,
|
| 288 |
+
"loss": 0.4047,
|
| 289 |
+
"step": 1000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.5711022272986864,
|
| 293 |
+
"eval_loss": 0.4848648011684418,
|
| 294 |
+
"eval_runtime": 1825.4203,
|
| 295 |
+
"eval_samples_per_second": 2.137,
|
| 296 |
+
"eval_steps_per_second": 0.134,
|
| 297 |
+
"eval_wer": 0.35052641746353713,
|
| 298 |
+
"step": 1000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.5853797829811537,
|
| 302 |
+
"grad_norm": 3.7762739658355713,
|
| 303 |
+
"learning_rate": 8.833333333333334e-06,
|
| 304 |
+
"loss": 0.4218,
|
| 305 |
+
"step": 1025
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.5996573386636208,
|
| 309 |
+
"grad_norm": 3.495347023010254,
|
| 310 |
+
"learning_rate": 8.777777777777778e-06,
|
| 311 |
+
"loss": 0.3968,
|
| 312 |
+
"step": 1050
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.613934894346088,
|
| 316 |
+
"grad_norm": 3.5088939666748047,
|
| 317 |
+
"learning_rate": 8.722222222222224e-06,
|
| 318 |
+
"loss": 0.4108,
|
| 319 |
+
"step": 1075
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.6282124500285551,
|
| 323 |
+
"grad_norm": 3.555328845977783,
|
| 324 |
+
"learning_rate": 8.666666666666668e-06,
|
| 325 |
+
"loss": 0.4063,
|
| 326 |
+
"step": 1100
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.6424900057110223,
|
| 330 |
+
"grad_norm": 2.9576587677001953,
|
| 331 |
+
"learning_rate": 8.611111111111112e-06,
|
| 332 |
+
"loss": 0.4116,
|
| 333 |
+
"step": 1125
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.6567675613934895,
|
| 337 |
+
"grad_norm": 3.280855178833008,
|
| 338 |
+
"learning_rate": 8.555555555555556e-06,
|
| 339 |
+
"loss": 0.4083,
|
| 340 |
+
"step": 1150
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.6710451170759566,
|
| 344 |
+
"grad_norm": 3.903722047805786,
|
| 345 |
+
"learning_rate": 8.5e-06,
|
| 346 |
+
"loss": 0.411,
|
| 347 |
+
"step": 1175
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 0.6853226727584237,
|
| 351 |
+
"grad_norm": 3.519038438796997,
|
| 352 |
+
"learning_rate": 8.444444444444446e-06,
|
| 353 |
+
"loss": 0.3964,
|
| 354 |
+
"step": 1200
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.6996002284408909,
|
| 358 |
+
"grad_norm": 3.3553972244262695,
|
| 359 |
+
"learning_rate": 8.38888888888889e-06,
|
| 360 |
+
"loss": 0.4049,
|
| 361 |
+
"step": 1225
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 0.7138777841233581,
|
| 365 |
+
"grad_norm": 3.3820197582244873,
|
| 366 |
+
"learning_rate": 8.333333333333334e-06,
|
| 367 |
+
"loss": 0.4159,
|
| 368 |
+
"step": 1250
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 0.7281553398058253,
|
| 372 |
+
"grad_norm": 2.782127857208252,
|
| 373 |
+
"learning_rate": 8.277777777777778e-06,
|
| 374 |
+
"loss": 0.3859,
|
| 375 |
+
"step": 1275
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.7424328954882924,
|
| 379 |
+
"grad_norm": 3.5839345455169678,
|
| 380 |
+
"learning_rate": 8.222222222222222e-06,
|
| 381 |
+
"loss": 0.392,
|
| 382 |
+
"step": 1300
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 0.7567104511707595,
|
| 386 |
+
"grad_norm": 3.0308761596679688,
|
| 387 |
+
"learning_rate": 8.166666666666668e-06,
|
| 388 |
+
"loss": 0.3899,
|
| 389 |
+
"step": 1325
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 0.7709880068532268,
|
| 393 |
+
"grad_norm": 3.136904001235962,
|
| 394 |
+
"learning_rate": 8.111111111111112e-06,
|
| 395 |
+
"loss": 0.3907,
|
| 396 |
+
"step": 1350
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 0.7852655625356939,
|
| 400 |
+
"grad_norm": 3.3192756175994873,
|
| 401 |
+
"learning_rate": 8.055555555555557e-06,
|
| 402 |
+
"loss": 0.3941,
|
| 403 |
+
"step": 1375
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.799543118218161,
|
| 407 |
+
"grad_norm": 4.766107082366943,
|
| 408 |
+
"learning_rate": 8.000000000000001e-06,
|
| 409 |
+
"loss": 0.3887,
|
| 410 |
+
"step": 1400
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.8138206739006282,
|
| 414 |
+
"grad_norm": 4.241744041442871,
|
| 415 |
+
"learning_rate": 7.944444444444445e-06,
|
| 416 |
+
"loss": 0.4033,
|
| 417 |
+
"step": 1425
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.8280982295830954,
|
| 421 |
+
"grad_norm": 3.1559460163116455,
|
| 422 |
+
"learning_rate": 7.88888888888889e-06,
|
| 423 |
+
"loss": 0.3567,
|
| 424 |
+
"step": 1450
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 0.8423757852655626,
|
| 428 |
+
"grad_norm": 3.142645835876465,
|
| 429 |
+
"learning_rate": 7.833333333333333e-06,
|
| 430 |
+
"loss": 0.3731,
|
| 431 |
+
"step": 1475
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.8566533409480297,
|
| 435 |
+
"grad_norm": 3.1183199882507324,
|
| 436 |
+
"learning_rate": 7.77777777777778e-06,
|
| 437 |
+
"loss": 0.3668,
|
| 438 |
+
"step": 1500
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 0.8709308966304968,
|
| 442 |
+
"grad_norm": 2.7859325408935547,
|
| 443 |
+
"learning_rate": 7.722222222222223e-06,
|
| 444 |
+
"loss": 0.3965,
|
| 445 |
+
"step": 1525
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 0.885208452312964,
|
| 449 |
+
"grad_norm": 3.191088914871216,
|
| 450 |
+
"learning_rate": 7.666666666666667e-06,
|
| 451 |
+
"loss": 0.3574,
|
| 452 |
+
"step": 1550
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 0.8994860079954312,
|
| 456 |
+
"grad_norm": 3.0640053749084473,
|
| 457 |
+
"learning_rate": 7.611111111111111e-06,
|
| 458 |
+
"loss": 0.3811,
|
| 459 |
+
"step": 1575
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.9137635636778983,
|
| 463 |
+
"grad_norm": 3.0769450664520264,
|
| 464 |
+
"learning_rate": 7.555555555555556e-06,
|
| 465 |
+
"loss": 0.3788,
|
| 466 |
+
"step": 1600
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 0.9280411193603655,
|
| 470 |
+
"grad_norm": 3.1407933235168457,
|
| 471 |
+
"learning_rate": 7.500000000000001e-06,
|
| 472 |
+
"loss": 0.3698,
|
| 473 |
+
"step": 1625
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 0.9423186750428326,
|
| 477 |
+
"grad_norm": 3.410187244415283,
|
| 478 |
+
"learning_rate": 7.444444444444445e-06,
|
| 479 |
+
"loss": 0.3907,
|
| 480 |
+
"step": 1650
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 0.9565962307252999,
|
| 484 |
+
"grad_norm": 3.3382880687713623,
|
| 485 |
+
"learning_rate": 7.38888888888889e-06,
|
| 486 |
+
"loss": 0.3368,
|
| 487 |
+
"step": 1675
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 0.970873786407767,
|
| 491 |
+
"grad_norm": 3.194368600845337,
|
| 492 |
+
"learning_rate": 7.333333333333333e-06,
|
| 493 |
+
"loss": 0.369,
|
| 494 |
+
"step": 1700
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.9851513420902341,
|
| 498 |
+
"grad_norm": 3.089852809906006,
|
| 499 |
+
"learning_rate": 7.277777777777778e-06,
|
| 500 |
+
"loss": 0.3765,
|
| 501 |
+
"step": 1725
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.9994288977727013,
|
| 505 |
+
"grad_norm": 3.0002810955047607,
|
| 506 |
+
"learning_rate": 7.222222222222223e-06,
|
| 507 |
+
"loss": 0.3705,
|
| 508 |
+
"step": 1750
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 1.0137064534551685,
|
| 512 |
+
"grad_norm": 2.3977696895599365,
|
| 513 |
+
"learning_rate": 7.166666666666667e-06,
|
| 514 |
+
"loss": 0.2584,
|
| 515 |
+
"step": 1775
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 1.0279840091376355,
|
| 519 |
+
"grad_norm": 2.3220465183258057,
|
| 520 |
+
"learning_rate": 7.111111111111112e-06,
|
| 521 |
+
"loss": 0.2538,
|
| 522 |
+
"step": 1800
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 1.0422615648201028,
|
| 526 |
+
"grad_norm": 2.819687843322754,
|
| 527 |
+
"learning_rate": 7.055555555555557e-06,
|
| 528 |
+
"loss": 0.2571,
|
| 529 |
+
"step": 1825
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 1.05653912050257,
|
| 533 |
+
"grad_norm": 2.514644145965576,
|
| 534 |
+
"learning_rate": 7e-06,
|
| 535 |
+
"loss": 0.2806,
|
| 536 |
+
"step": 1850
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 1.070816676185037,
|
| 540 |
+
"grad_norm": 2.1887128353118896,
|
| 541 |
+
"learning_rate": 6.944444444444445e-06,
|
| 542 |
+
"loss": 0.2626,
|
| 543 |
+
"step": 1875
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 1.0850942318675043,
|
| 547 |
+
"grad_norm": 2.592247486114502,
|
| 548 |
+
"learning_rate": 6.88888888888889e-06,
|
| 549 |
+
"loss": 0.2509,
|
| 550 |
+
"step": 1900
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 1.0993717875499716,
|
| 554 |
+
"grad_norm": 2.371534824371338,
|
| 555 |
+
"learning_rate": 6.833333333333334e-06,
|
| 556 |
+
"loss": 0.2605,
|
| 557 |
+
"step": 1925
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.1136493432324386,
|
| 561 |
+
"grad_norm": 3.1825778484344482,
|
| 562 |
+
"learning_rate": 6.777777777777779e-06,
|
| 563 |
+
"loss": 0.2495,
|
| 564 |
+
"step": 1950
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.1279268989149058,
|
| 568 |
+
"grad_norm": 2.901749849319458,
|
| 569 |
+
"learning_rate": 6.7222222222222235e-06,
|
| 570 |
+
"loss": 0.261,
|
| 571 |
+
"step": 1975
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.1422044545973729,
|
| 575 |
+
"grad_norm": 2.658766984939575,
|
| 576 |
+
"learning_rate": 6.666666666666667e-06,
|
| 577 |
+
"loss": 0.2476,
|
| 578 |
+
"step": 2000
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.1422044545973729,
|
| 582 |
+
"eval_loss": 0.41870468854904175,
|
| 583 |
+
"eval_runtime": 1722.2575,
|
| 584 |
+
"eval_samples_per_second": 2.265,
|
| 585 |
+
"eval_steps_per_second": 0.142,
|
| 586 |
+
"eval_wer": 0.3136771950159374,
|
| 587 |
+
"step": 2000
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 1.15648201027984,
|
| 591 |
+
"grad_norm": 2.711312770843506,
|
| 592 |
+
"learning_rate": 6.6111111111111115e-06,
|
| 593 |
+
"loss": 0.2414,
|
| 594 |
+
"step": 2025
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 1.1707595659623073,
|
| 598 |
+
"grad_norm": 2.9044759273529053,
|
| 599 |
+
"learning_rate": 6.555555555555556e-06,
|
| 600 |
+
"loss": 0.2502,
|
| 601 |
+
"step": 2050
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 1.1850371216447744,
|
| 605 |
+
"grad_norm": 2.549725294113159,
|
| 606 |
+
"learning_rate": 6.5000000000000004e-06,
|
| 607 |
+
"loss": 0.2511,
|
| 608 |
+
"step": 2075
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 1.1993146773272416,
|
| 612 |
+
"grad_norm": 2.95792555809021,
|
| 613 |
+
"learning_rate": 6.444444444444445e-06,
|
| 614 |
+
"loss": 0.2427,
|
| 615 |
+
"step": 2100
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 1.2135922330097086,
|
| 619 |
+
"grad_norm": 2.686870574951172,
|
| 620 |
+
"learning_rate": 6.3888888888888885e-06,
|
| 621 |
+
"loss": 0.2637,
|
| 622 |
+
"step": 2125
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 1.227869788692176,
|
| 626 |
+
"grad_norm": 3.7834455966949463,
|
| 627 |
+
"learning_rate": 6.333333333333333e-06,
|
| 628 |
+
"loss": 0.2554,
|
| 629 |
+
"step": 2150
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 1.2421473443746431,
|
| 633 |
+
"grad_norm": 3.0891430377960205,
|
| 634 |
+
"learning_rate": 6.277777777777778e-06,
|
| 635 |
+
"loss": 0.2467,
|
| 636 |
+
"step": 2175
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 1.2564249000571102,
|
| 640 |
+
"grad_norm": 2.771472930908203,
|
| 641 |
+
"learning_rate": 6.222222222222223e-06,
|
| 642 |
+
"loss": 0.2467,
|
| 643 |
+
"step": 2200
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 1.2707024557395774,
|
| 647 |
+
"grad_norm": 2.6807925701141357,
|
| 648 |
+
"learning_rate": 6.166666666666667e-06,
|
| 649 |
+
"loss": 0.2682,
|
| 650 |
+
"step": 2225
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 1.2849800114220447,
|
| 654 |
+
"grad_norm": 2.2320196628570557,
|
| 655 |
+
"learning_rate": 6.111111111111112e-06,
|
| 656 |
+
"loss": 0.2408,
|
| 657 |
+
"step": 2250
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.2992575671045117,
|
| 661 |
+
"grad_norm": 3.066009759902954,
|
| 662 |
+
"learning_rate": 6.055555555555555e-06,
|
| 663 |
+
"loss": 0.2363,
|
| 664 |
+
"step": 2275
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 1.313535122786979,
|
| 668 |
+
"grad_norm": 2.6043167114257812,
|
| 669 |
+
"learning_rate": 6e-06,
|
| 670 |
+
"loss": 0.2483,
|
| 671 |
+
"step": 2300
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 1.327812678469446,
|
| 675 |
+
"grad_norm": 2.6250624656677246,
|
| 676 |
+
"learning_rate": 5.944444444444445e-06,
|
| 677 |
+
"loss": 0.2563,
|
| 678 |
+
"step": 2325
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 1.3420902341519132,
|
| 682 |
+
"grad_norm": 2.508998394012451,
|
| 683 |
+
"learning_rate": 5.88888888888889e-06,
|
| 684 |
+
"loss": 0.2581,
|
| 685 |
+
"step": 2350
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 1.3563677898343802,
|
| 689 |
+
"grad_norm": 2.872715473175049,
|
| 690 |
+
"learning_rate": 5.833333333333334e-06,
|
| 691 |
+
"loss": 0.2371,
|
| 692 |
+
"step": 2375
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 1.3706453455168475,
|
| 696 |
+
"grad_norm": 3.1910557746887207,
|
| 697 |
+
"learning_rate": 5.777777777777778e-06,
|
| 698 |
+
"loss": 0.2515,
|
| 699 |
+
"step": 2400
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 1.3849229011993147,
|
| 703 |
+
"grad_norm": 2.7466485500335693,
|
| 704 |
+
"learning_rate": 5.722222222222222e-06,
|
| 705 |
+
"loss": 0.2578,
|
| 706 |
+
"step": 2425
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 1.3992004568817817,
|
| 710 |
+
"grad_norm": 2.388066530227661,
|
| 711 |
+
"learning_rate": 5.666666666666667e-06,
|
| 712 |
+
"loss": 0.2541,
|
| 713 |
+
"step": 2450
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 1.413478012564249,
|
| 717 |
+
"grad_norm": 2.688497304916382,
|
| 718 |
+
"learning_rate": 5.611111111111112e-06,
|
| 719 |
+
"loss": 0.2514,
|
| 720 |
+
"step": 2475
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 1.4277555682467162,
|
| 724 |
+
"grad_norm": 2.710899591445923,
|
| 725 |
+
"learning_rate": 5.555555555555557e-06,
|
| 726 |
+
"loss": 0.2765,
|
| 727 |
+
"step": 2500
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 1.4420331239291833,
|
| 731 |
+
"grad_norm": 2.296635389328003,
|
| 732 |
+
"learning_rate": 5.500000000000001e-06,
|
| 733 |
+
"loss": 0.2487,
|
| 734 |
+
"step": 2525
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 1.4563106796116505,
|
| 738 |
+
"grad_norm": 2.7988133430480957,
|
| 739 |
+
"learning_rate": 5.444444444444445e-06,
|
| 740 |
+
"loss": 0.2499,
|
| 741 |
+
"step": 2550
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 1.4705882352941178,
|
| 745 |
+
"grad_norm": 3.1988582611083984,
|
| 746 |
+
"learning_rate": 5.388888888888889e-06,
|
| 747 |
+
"loss": 0.2456,
|
| 748 |
+
"step": 2575
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 1.4848657909765848,
|
| 752 |
+
"grad_norm": 2.657517910003662,
|
| 753 |
+
"learning_rate": 5.333333333333334e-06,
|
| 754 |
+
"loss": 0.2613,
|
| 755 |
+
"step": 2600
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 1.499143346659052,
|
| 759 |
+
"grad_norm": 2.5517725944519043,
|
| 760 |
+
"learning_rate": 5.2777777777777785e-06,
|
| 761 |
+
"loss": 0.2528,
|
| 762 |
+
"step": 2625
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 1.5134209023415193,
|
| 766 |
+
"grad_norm": 2.7166850566864014,
|
| 767 |
+
"learning_rate": 5.2222222222222226e-06,
|
| 768 |
+
"loss": 0.2476,
|
| 769 |
+
"step": 2650
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 1.5276984580239863,
|
| 773 |
+
"grad_norm": 2.7338292598724365,
|
| 774 |
+
"learning_rate": 5.1666666666666675e-06,
|
| 775 |
+
"loss": 0.2489,
|
| 776 |
+
"step": 2675
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 1.5419760137064533,
|
| 780 |
+
"grad_norm": 2.1498470306396484,
|
| 781 |
+
"learning_rate": 5.1111111111111115e-06,
|
| 782 |
+
"loss": 0.2388,
|
| 783 |
+
"step": 2700
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 1.5562535693889206,
|
| 787 |
+
"grad_norm": 2.595247745513916,
|
| 788 |
+
"learning_rate": 5.0555555555555555e-06,
|
| 789 |
+
"loss": 0.2566,
|
| 790 |
+
"step": 2725
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 1.5705311250713878,
|
| 794 |
+
"grad_norm": 2.652132987976074,
|
| 795 |
+
"learning_rate": 5e-06,
|
| 796 |
+
"loss": 0.239,
|
| 797 |
+
"step": 2750
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 1.5848086807538548,
|
| 801 |
+
"grad_norm": 2.436605930328369,
|
| 802 |
+
"learning_rate": 4.944444444444445e-06,
|
| 803 |
+
"loss": 0.2419,
|
| 804 |
+
"step": 2775
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 1.599086236436322,
|
| 808 |
+
"grad_norm": 2.618035316467285,
|
| 809 |
+
"learning_rate": 4.888888888888889e-06,
|
| 810 |
+
"loss": 0.2295,
|
| 811 |
+
"step": 2800
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 1.6133637921187893,
|
| 815 |
+
"grad_norm": 2.2901298999786377,
|
| 816 |
+
"learning_rate": 4.833333333333333e-06,
|
| 817 |
+
"loss": 0.2446,
|
| 818 |
+
"step": 2825
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 1.6276413478012564,
|
| 822 |
+
"grad_norm": 2.899315595626831,
|
| 823 |
+
"learning_rate": 4.777777777777778e-06,
|
| 824 |
+
"loss": 0.2628,
|
| 825 |
+
"step": 2850
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 1.6419189034837236,
|
| 829 |
+
"grad_norm": 2.616224527359009,
|
| 830 |
+
"learning_rate": 4.722222222222222e-06,
|
| 831 |
+
"loss": 0.2273,
|
| 832 |
+
"step": 2875
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 1.6561964591661908,
|
| 836 |
+
"grad_norm": 2.43113112449646,
|
| 837 |
+
"learning_rate": 4.666666666666667e-06,
|
| 838 |
+
"loss": 0.2362,
|
| 839 |
+
"step": 2900
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 1.6704740148486579,
|
| 843 |
+
"grad_norm": 2.5203065872192383,
|
| 844 |
+
"learning_rate": 4.611111111111112e-06,
|
| 845 |
+
"loss": 0.2428,
|
| 846 |
+
"step": 2925
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 1.6847515705311251,
|
| 850 |
+
"grad_norm": 2.3064985275268555,
|
| 851 |
+
"learning_rate": 4.555555555555556e-06,
|
| 852 |
+
"loss": 0.2441,
|
| 853 |
+
"step": 2950
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 1.6990291262135924,
|
| 857 |
+
"grad_norm": 2.201695680618286,
|
| 858 |
+
"learning_rate": 4.5e-06,
|
| 859 |
+
"loss": 0.2324,
|
| 860 |
+
"step": 2975
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 1.7133066818960594,
|
| 864 |
+
"grad_norm": 2.442471981048584,
|
| 865 |
+
"learning_rate": 4.444444444444444e-06,
|
| 866 |
+
"loss": 0.2527,
|
| 867 |
+
"step": 3000
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 1.7133066818960594,
|
| 871 |
+
"eval_loss": 0.3882293701171875,
|
| 872 |
+
"eval_runtime": 1749.1422,
|
| 873 |
+
"eval_samples_per_second": 2.23,
|
| 874 |
+
"eval_steps_per_second": 0.139,
|
| 875 |
+
"eval_wer": 0.2901091471071187,
|
| 876 |
+
"step": 3000
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 1.7275842375785264,
|
| 880 |
+
"grad_norm": 2.77786922454834,
|
| 881 |
+
"learning_rate": 4.388888888888889e-06,
|
| 882 |
+
"loss": 0.2492,
|
| 883 |
+
"step": 3025
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 1.7418617932609937,
|
| 887 |
+
"grad_norm": 2.5009052753448486,
|
| 888 |
+
"learning_rate": 4.333333333333334e-06,
|
| 889 |
+
"loss": 0.2341,
|
| 890 |
+
"step": 3050
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 1.756139348943461,
|
| 894 |
+
"grad_norm": 2.780186176300049,
|
| 895 |
+
"learning_rate": 4.277777777777778e-06,
|
| 896 |
+
"loss": 0.2407,
|
| 897 |
+
"step": 3075
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 1.770416904625928,
|
| 901 |
+
"grad_norm": 1.9574618339538574,
|
| 902 |
+
"learning_rate": 4.222222222222223e-06,
|
| 903 |
+
"loss": 0.2437,
|
| 904 |
+
"step": 3100
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 1.7846944603083952,
|
| 908 |
+
"grad_norm": 2.151125907897949,
|
| 909 |
+
"learning_rate": 4.166666666666667e-06,
|
| 910 |
+
"loss": 0.2341,
|
| 911 |
+
"step": 3125
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 1.7989720159908624,
|
| 915 |
+
"grad_norm": 2.170015811920166,
|
| 916 |
+
"learning_rate": 4.111111111111111e-06,
|
| 917 |
+
"loss": 0.2373,
|
| 918 |
+
"step": 3150
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 1.8132495716733295,
|
| 922 |
+
"grad_norm": 3.0467231273651123,
|
| 923 |
+
"learning_rate": 4.055555555555556e-06,
|
| 924 |
+
"loss": 0.2317,
|
| 925 |
+
"step": 3175
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 1.8275271273557967,
|
| 929 |
+
"grad_norm": 3.0150015354156494,
|
| 930 |
+
"learning_rate": 4.000000000000001e-06,
|
| 931 |
+
"loss": 0.228,
|
| 932 |
+
"step": 3200
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 1.841804683038264,
|
| 936 |
+
"grad_norm": 3.275949001312256,
|
| 937 |
+
"learning_rate": 3.944444444444445e-06,
|
| 938 |
+
"loss": 0.2438,
|
| 939 |
+
"step": 3225
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 1.856082238720731,
|
| 943 |
+
"grad_norm": 3.0381839275360107,
|
| 944 |
+
"learning_rate": 3.88888888888889e-06,
|
| 945 |
+
"loss": 0.2478,
|
| 946 |
+
"step": 3250
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 1.8703597944031982,
|
| 950 |
+
"grad_norm": 2.770716428756714,
|
| 951 |
+
"learning_rate": 3.833333333333334e-06,
|
| 952 |
+
"loss": 0.2312,
|
| 953 |
+
"step": 3275
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 1.8846373500856655,
|
| 957 |
+
"grad_norm": 2.6976678371429443,
|
| 958 |
+
"learning_rate": 3.777777777777778e-06,
|
| 959 |
+
"loss": 0.2284,
|
| 960 |
+
"step": 3300
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 1.8989149057681325,
|
| 964 |
+
"grad_norm": 2.8799102306365967,
|
| 965 |
+
"learning_rate": 3.7222222222222225e-06,
|
| 966 |
+
"loss": 0.2484,
|
| 967 |
+
"step": 3325
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 1.9131924614505995,
|
| 971 |
+
"grad_norm": 2.574629545211792,
|
| 972 |
+
"learning_rate": 3.6666666666666666e-06,
|
| 973 |
+
"loss": 0.2295,
|
| 974 |
+
"step": 3350
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 1.927470017133067,
|
| 978 |
+
"grad_norm": 2.4746835231781006,
|
| 979 |
+
"learning_rate": 3.6111111111111115e-06,
|
| 980 |
+
"loss": 0.2335,
|
| 981 |
+
"step": 3375
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 1.941747572815534,
|
| 985 |
+
"grad_norm": 3.084383964538574,
|
| 986 |
+
"learning_rate": 3.555555555555556e-06,
|
| 987 |
+
"loss": 0.212,
|
| 988 |
+
"step": 3400
|
| 989 |
+
},
|
| 990 |
+
{
|
| 991 |
+
"epoch": 1.956025128498001,
|
| 992 |
+
"grad_norm": 2.4441068172454834,
|
| 993 |
+
"learning_rate": 3.5e-06,
|
| 994 |
+
"loss": 0.221,
|
| 995 |
+
"step": 3425
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 1.9703026841804683,
|
| 999 |
+
"grad_norm": 3.031568765640259,
|
| 1000 |
+
"learning_rate": 3.444444444444445e-06,
|
| 1001 |
+
"loss": 0.2341,
|
| 1002 |
+
"step": 3450
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 1.9845802398629355,
|
| 1006 |
+
"grad_norm": 2.3584327697753906,
|
| 1007 |
+
"learning_rate": 3.3888888888888893e-06,
|
| 1008 |
+
"loss": 0.2431,
|
| 1009 |
+
"step": 3475
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 1.9988577955454025,
|
| 1013 |
+
"grad_norm": 2.1590421199798584,
|
| 1014 |
+
"learning_rate": 3.3333333333333333e-06,
|
| 1015 |
+
"loss": 0.2357,
|
| 1016 |
+
"step": 3500
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 2.0131353512278696,
|
| 1020 |
+
"grad_norm": 2.2845587730407715,
|
| 1021 |
+
"learning_rate": 3.277777777777778e-06,
|
| 1022 |
+
"loss": 0.1576,
|
| 1023 |
+
"step": 3525
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 2.027412906910337,
|
| 1027 |
+
"grad_norm": 2.033133029937744,
|
| 1028 |
+
"learning_rate": 3.2222222222222227e-06,
|
| 1029 |
+
"loss": 0.1422,
|
| 1030 |
+
"step": 3550
|
| 1031 |
+
},
|
| 1032 |
+
{
|
| 1033 |
+
"epoch": 2.041690462592804,
|
| 1034 |
+
"grad_norm": 2.2549259662628174,
|
| 1035 |
+
"learning_rate": 3.1666666666666667e-06,
|
| 1036 |
+
"loss": 0.1473,
|
| 1037 |
+
"step": 3575
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 2.055968018275271,
|
| 1041 |
+
"grad_norm": 1.5837754011154175,
|
| 1042 |
+
"learning_rate": 3.1111111111111116e-06,
|
| 1043 |
+
"loss": 0.143,
|
| 1044 |
+
"step": 3600
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 2.0702455739577386,
|
| 1048 |
+
"grad_norm": 1.9988360404968262,
|
| 1049 |
+
"learning_rate": 3.055555555555556e-06,
|
| 1050 |
+
"loss": 0.1416,
|
| 1051 |
+
"step": 3625
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 2.0845231296402056,
|
| 1055 |
+
"grad_norm": 2.148613929748535,
|
| 1056 |
+
"learning_rate": 3e-06,
|
| 1057 |
+
"loss": 0.1338,
|
| 1058 |
+
"step": 3650
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 2.0988006853226726,
|
| 1062 |
+
"grad_norm": 1.8176393508911133,
|
| 1063 |
+
"learning_rate": 2.944444444444445e-06,
|
| 1064 |
+
"loss": 0.1514,
|
| 1065 |
+
"step": 3675
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 2.11307824100514,
|
| 1069 |
+
"grad_norm": 2.60271954536438,
|
| 1070 |
+
"learning_rate": 2.888888888888889e-06,
|
| 1071 |
+
"loss": 0.1533,
|
| 1072 |
+
"step": 3700
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"epoch": 2.127355796687607,
|
| 1076 |
+
"grad_norm": 2.120281457901001,
|
| 1077 |
+
"learning_rate": 2.8333333333333335e-06,
|
| 1078 |
+
"loss": 0.1404,
|
| 1079 |
+
"step": 3725
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 2.141633352370074,
|
| 1083 |
+
"grad_norm": 2.3522286415100098,
|
| 1084 |
+
"learning_rate": 2.7777777777777783e-06,
|
| 1085 |
+
"loss": 0.1511,
|
| 1086 |
+
"step": 3750
|
| 1087 |
+
},
|
| 1088 |
+
{
|
| 1089 |
+
"epoch": 2.1559109080525416,
|
| 1090 |
+
"grad_norm": 1.8738924264907837,
|
| 1091 |
+
"learning_rate": 2.7222222222222224e-06,
|
| 1092 |
+
"loss": 0.1417,
|
| 1093 |
+
"step": 3775
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"epoch": 2.1701884637350086,
|
| 1097 |
+
"grad_norm": 2.255291223526001,
|
| 1098 |
+
"learning_rate": 2.666666666666667e-06,
|
| 1099 |
+
"loss": 0.1437,
|
| 1100 |
+
"step": 3800
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 2.1844660194174756,
|
| 1104 |
+
"grad_norm": 1.7046154737472534,
|
| 1105 |
+
"learning_rate": 2.6111111111111113e-06,
|
| 1106 |
+
"loss": 0.1446,
|
| 1107 |
+
"step": 3825
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 2.198743575099943,
|
| 1111 |
+
"grad_norm": 2.0543861389160156,
|
| 1112 |
+
"learning_rate": 2.5555555555555557e-06,
|
| 1113 |
+
"loss": 0.1504,
|
| 1114 |
+
"step": 3850
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 2.21302113078241,
|
| 1118 |
+
"grad_norm": 2.139716863632202,
|
| 1119 |
+
"learning_rate": 2.5e-06,
|
| 1120 |
+
"loss": 0.1345,
|
| 1121 |
+
"step": 3875
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 2.227298686464877,
|
| 1125 |
+
"grad_norm": 1.7999951839447021,
|
| 1126 |
+
"learning_rate": 2.4444444444444447e-06,
|
| 1127 |
+
"loss": 0.1389,
|
| 1128 |
+
"step": 3900
|
| 1129 |
+
},
|
| 1130 |
+
{
|
| 1131 |
+
"epoch": 2.241576242147344,
|
| 1132 |
+
"grad_norm": 1.7282090187072754,
|
| 1133 |
+
"learning_rate": 2.388888888888889e-06,
|
| 1134 |
+
"loss": 0.1324,
|
| 1135 |
+
"step": 3925
|
| 1136 |
+
},
|
| 1137 |
+
{
|
| 1138 |
+
"epoch": 2.2558537978298117,
|
| 1139 |
+
"grad_norm": 2.6271605491638184,
|
| 1140 |
+
"learning_rate": 2.3333333333333336e-06,
|
| 1141 |
+
"loss": 0.1551,
|
| 1142 |
+
"step": 3950
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 2.2701313535122787,
|
| 1146 |
+
"grad_norm": 2.170382022857666,
|
| 1147 |
+
"learning_rate": 2.277777777777778e-06,
|
| 1148 |
+
"loss": 0.144,
|
| 1149 |
+
"step": 3975
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 2.2844089091947457,
|
| 1153 |
+
"grad_norm": 1.796635627746582,
|
| 1154 |
+
"learning_rate": 2.222222222222222e-06,
|
| 1155 |
+
"loss": 0.1568,
|
| 1156 |
+
"step": 4000
|
| 1157 |
+
},
|
| 1158 |
+
{
|
| 1159 |
+
"epoch": 2.2844089091947457,
|
| 1160 |
+
"eval_loss": 0.3901652991771698,
|
| 1161 |
+
"eval_runtime": 1765.3609,
|
| 1162 |
+
"eval_samples_per_second": 2.21,
|
| 1163 |
+
"eval_steps_per_second": 0.138,
|
| 1164 |
+
"eval_wer": 0.28160919540229884,
|
| 1165 |
+
"step": 4000
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 2.298686464877213,
|
| 1169 |
+
"grad_norm": 2.0357980728149414,
|
| 1170 |
+
"learning_rate": 2.166666666666667e-06,
|
| 1171 |
+
"loss": 0.161,
|
| 1172 |
+
"step": 4025
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 2.31296402055968,
|
| 1176 |
+
"grad_norm": 2.027215003967285,
|
| 1177 |
+
"learning_rate": 2.1111111111111114e-06,
|
| 1178 |
+
"loss": 0.1353,
|
| 1179 |
+
"step": 4050
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 2.3272415762421472,
|
| 1183 |
+
"grad_norm": 2.8169405460357666,
|
| 1184 |
+
"learning_rate": 2.0555555555555555e-06,
|
| 1185 |
+
"loss": 0.1449,
|
| 1186 |
+
"step": 4075
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 2.3415191319246147,
|
| 1190 |
+
"grad_norm": 1.9528751373291016,
|
| 1191 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 1192 |
+
"loss": 0.1376,
|
| 1193 |
+
"step": 4100
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 2.3557966876070817,
|
| 1197 |
+
"grad_norm": 2.5781335830688477,
|
| 1198 |
+
"learning_rate": 1.944444444444445e-06,
|
| 1199 |
+
"loss": 0.1383,
|
| 1200 |
+
"step": 4125
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 2.3700742432895487,
|
| 1204 |
+
"grad_norm": 2.083077907562256,
|
| 1205 |
+
"learning_rate": 1.888888888888889e-06,
|
| 1206 |
+
"loss": 0.1362,
|
| 1207 |
+
"step": 4150
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 2.384351798972016,
|
| 1211 |
+
"grad_norm": 2.431272029876709,
|
| 1212 |
+
"learning_rate": 1.8333333333333333e-06,
|
| 1213 |
+
"loss": 0.1329,
|
| 1214 |
+
"step": 4175
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 2.3986293546544832,
|
| 1218 |
+
"grad_norm": 2.157139539718628,
|
| 1219 |
+
"learning_rate": 1.777777777777778e-06,
|
| 1220 |
+
"loss": 0.1377,
|
| 1221 |
+
"step": 4200
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 2.4129069103369503,
|
| 1225 |
+
"grad_norm": 2.5328071117401123,
|
| 1226 |
+
"learning_rate": 1.7222222222222224e-06,
|
| 1227 |
+
"loss": 0.1361,
|
| 1228 |
+
"step": 4225
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 2.4271844660194173,
|
| 1232 |
+
"grad_norm": 2.433239459991455,
|
| 1233 |
+
"learning_rate": 1.6666666666666667e-06,
|
| 1234 |
+
"loss": 0.157,
|
| 1235 |
+
"step": 4250
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 2.4414620217018848,
|
| 1239 |
+
"grad_norm": 2.5167510509490967,
|
| 1240 |
+
"learning_rate": 1.6111111111111113e-06,
|
| 1241 |
+
"loss": 0.132,
|
| 1242 |
+
"step": 4275
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 2.455739577384352,
|
| 1246 |
+
"grad_norm": 1.9507442712783813,
|
| 1247 |
+
"learning_rate": 1.5555555555555558e-06,
|
| 1248 |
+
"loss": 0.1625,
|
| 1249 |
+
"step": 4300
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 2.470017133066819,
|
| 1253 |
+
"grad_norm": 2.2467007637023926,
|
| 1254 |
+
"learning_rate": 1.5e-06,
|
| 1255 |
+
"loss": 0.1333,
|
| 1256 |
+
"step": 4325
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 2.4842946887492863,
|
| 1260 |
+
"grad_norm": 2.4816768169403076,
|
| 1261 |
+
"learning_rate": 1.4444444444444445e-06,
|
| 1262 |
+
"loss": 0.1499,
|
| 1263 |
+
"step": 4350
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 2.4985722444317533,
|
| 1267 |
+
"grad_norm": 2.0616416931152344,
|
| 1268 |
+
"learning_rate": 1.3888888888888892e-06,
|
| 1269 |
+
"loss": 0.1508,
|
| 1270 |
+
"step": 4375
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 2.5128498001142203,
|
| 1274 |
+
"grad_norm": 2.089355230331421,
|
| 1275 |
+
"learning_rate": 1.3333333333333334e-06,
|
| 1276 |
+
"loss": 0.1344,
|
| 1277 |
+
"step": 4400
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 2.5271273557966873,
|
| 1281 |
+
"grad_norm": 2.2235498428344727,
|
| 1282 |
+
"learning_rate": 1.28e-06,
|
| 1283 |
+
"loss": 0.1717,
|
| 1284 |
+
"step": 4425
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 2.541404911479155,
|
| 1288 |
+
"grad_norm": 1.9268138408660889,
|
| 1289 |
+
"learning_rate": 1.2244444444444445e-06,
|
| 1290 |
+
"loss": 0.143,
|
| 1291 |
+
"step": 4450
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 2.555682467161622,
|
| 1295 |
+
"grad_norm": 1.8911551237106323,
|
| 1296 |
+
"learning_rate": 1.168888888888889e-06,
|
| 1297 |
+
"loss": 0.1439,
|
| 1298 |
+
"step": 4475
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 2.5699600228440893,
|
| 1302 |
+
"grad_norm": 2.5078868865966797,
|
| 1303 |
+
"learning_rate": 1.1133333333333334e-06,
|
| 1304 |
+
"loss": 0.1341,
|
| 1305 |
+
"step": 4500
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 2.5842375785265563,
|
| 1309 |
+
"grad_norm": 2.1232492923736572,
|
| 1310 |
+
"learning_rate": 1.0577777777777779e-06,
|
| 1311 |
+
"loss": 0.1415,
|
| 1312 |
+
"step": 4525
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 2.5985151342090234,
|
| 1316 |
+
"grad_norm": 1.9214311838150024,
|
| 1317 |
+
"learning_rate": 1.0022222222222223e-06,
|
| 1318 |
+
"loss": 0.1301,
|
| 1319 |
+
"step": 4550
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 2.6127926898914904,
|
| 1323 |
+
"grad_norm": 2.4226858615875244,
|
| 1324 |
+
"learning_rate": 9.466666666666667e-07,
|
| 1325 |
+
"loss": 0.1438,
|
| 1326 |
+
"step": 4575
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 2.627070245573958,
|
| 1330 |
+
"grad_norm": 2.324777126312256,
|
| 1331 |
+
"learning_rate": 8.911111111111112e-07,
|
| 1332 |
+
"loss": 0.1306,
|
| 1333 |
+
"step": 4600
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 2.641347801256425,
|
| 1337 |
+
"grad_norm": 2.427114486694336,
|
| 1338 |
+
"learning_rate": 8.355555555555556e-07,
|
| 1339 |
+
"loss": 0.1359,
|
| 1340 |
+
"step": 4625
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 2.655625356938892,
|
| 1344 |
+
"grad_norm": 1.989882469177246,
|
| 1345 |
+
"learning_rate": 7.8e-07,
|
| 1346 |
+
"loss": 0.1386,
|
| 1347 |
+
"step": 4650
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 2.6699029126213594,
|
| 1351 |
+
"grad_norm": 2.6079118251800537,
|
| 1352 |
+
"learning_rate": 7.244444444444446e-07,
|
| 1353 |
+
"loss": 0.135,
|
| 1354 |
+
"step": 4675
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 2.6841804683038264,
|
| 1358 |
+
"grad_norm": 2.3429243564605713,
|
| 1359 |
+
"learning_rate": 6.68888888888889e-07,
|
| 1360 |
+
"loss": 0.1356,
|
| 1361 |
+
"step": 4700
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 2.6984580239862934,
|
| 1365 |
+
"grad_norm": 2.3358540534973145,
|
| 1366 |
+
"learning_rate": 6.133333333333333e-07,
|
| 1367 |
+
"loss": 0.1304,
|
| 1368 |
+
"step": 4725
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 2.7127355796687604,
|
| 1372 |
+
"grad_norm": 1.917809247970581,
|
| 1373 |
+
"learning_rate": 5.577777777777779e-07,
|
| 1374 |
+
"loss": 0.1395,
|
| 1375 |
+
"step": 4750
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 2.727013135351228,
|
| 1379 |
+
"grad_norm": 2.0677952766418457,
|
| 1380 |
+
"learning_rate": 5.022222222222222e-07,
|
| 1381 |
+
"loss": 0.1309,
|
| 1382 |
+
"step": 4775
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 2.741290691033695,
|
| 1386 |
+
"grad_norm": 2.135127305984497,
|
| 1387 |
+
"learning_rate": 4.466666666666667e-07,
|
| 1388 |
+
"loss": 0.1424,
|
| 1389 |
+
"step": 4800
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 2.7555682467161624,
|
| 1393 |
+
"grad_norm": 2.3306682109832764,
|
| 1394 |
+
"learning_rate": 3.9111111111111115e-07,
|
| 1395 |
+
"loss": 0.1318,
|
| 1396 |
+
"step": 4825
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 2.7698458023986294,
|
| 1400 |
+
"grad_norm": 2.0700454711914062,
|
| 1401 |
+
"learning_rate": 3.3555555555555556e-07,
|
| 1402 |
+
"loss": 0.1566,
|
| 1403 |
+
"step": 4850
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 2.7841233580810965,
|
| 1407 |
+
"grad_norm": 1.8561683893203735,
|
| 1408 |
+
"learning_rate": 2.8e-07,
|
| 1409 |
+
"loss": 0.1453,
|
| 1410 |
+
"step": 4875
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 2.7984009137635635,
|
| 1414 |
+
"grad_norm": 2.2682347297668457,
|
| 1415 |
+
"learning_rate": 2.2444444444444445e-07,
|
| 1416 |
+
"loss": 0.1415,
|
| 1417 |
+
"step": 4900
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 2.812678469446031,
|
| 1421 |
+
"grad_norm": 2.2898778915405273,
|
| 1422 |
+
"learning_rate": 1.6888888888888888e-07,
|
| 1423 |
+
"loss": 0.1427,
|
| 1424 |
+
"step": 4925
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 2.826956025128498,
|
| 1428 |
+
"grad_norm": 2.328401803970337,
|
| 1429 |
+
"learning_rate": 1.1333333333333336e-07,
|
| 1430 |
+
"loss": 0.1357,
|
| 1431 |
+
"step": 4950
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 2.841233580810965,
|
| 1435 |
+
"grad_norm": 2.2169013023376465,
|
| 1436 |
+
"learning_rate": 5.777777777777778e-08,
|
| 1437 |
+
"loss": 0.1343,
|
| 1438 |
+
"step": 4975
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 2.8555111364934325,
|
| 1442 |
+
"grad_norm": 2.42340350151062,
|
| 1443 |
+
"learning_rate": 2.2222222222222225e-09,
|
| 1444 |
+
"loss": 0.1313,
|
| 1445 |
+
"step": 5000
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 2.8555111364934325,
|
| 1449 |
+
"eval_loss": 0.38383349776268005,
|
| 1450 |
+
"eval_runtime": 1820.062,
|
| 1451 |
+
"eval_samples_per_second": 2.143,
|
| 1452 |
+
"eval_steps_per_second": 0.134,
|
| 1453 |
+
"eval_wer": 0.27318168646769053,
|
| 1454 |
+
"step": 5000
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 2.8555111364934325,
|
| 1458 |
+
"step": 5000,
|
| 1459 |
+
"total_flos": 5.435589590699213e+20,
|
| 1460 |
+
"train_loss": 0.3002769865989685,
|
| 1461 |
+
"train_runtime": 59305.2217,
|
| 1462 |
+
"train_samples_per_second": 2.698,
|
| 1463 |
+
"train_steps_per_second": 0.084
|
| 1464 |
+
}
|
| 1465 |
+
],
|
| 1466 |
+
"logging_steps": 25,
|
| 1467 |
+
"max_steps": 5000,
|
| 1468 |
+
"num_input_tokens_seen": 0,
|
| 1469 |
+
"num_train_epochs": 3,
|
| 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": 5.435589590699213e+20,
|
| 1484 |
+
"train_batch_size": 16,
|
| 1485 |
+
"trial_name": null,
|
| 1486 |
+
"trial_params": null
|
| 1487 |
+
}
|