End of training
Browse files- README.md +1 -1
- all_results.json +15 -0
- eval_results.json +9 -0
- runs/Nov02_13-48-22_daf84da58340/events.out.tfevents.1730574661.daf84da58340.37.1 +3 -0
- train_results.json +9 -0
- trainer_state.json +1487 -0
README.md
CHANGED
|
@@ -16,7 +16,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 16 |
|
| 17 |
# whisper-base-ft-btb-ca-ec-cv-cy-en
|
| 18 |
|
| 19 |
-
This model is a fine-tuned version of [openai/whisper-base](https://huggingface.co/openai/whisper-base) on
|
| 20 |
It achieves the following results on the evaluation set:
|
| 21 |
- Loss: 0.7135
|
| 22 |
- Wer: 0.5089
|
|
|
|
| 16 |
|
| 17 |
# whisper-base-ft-btb-ca-ec-cv-cy-en
|
| 18 |
|
| 19 |
+
This model is a fine-tuned version of [openai/whisper-base](https://huggingface.co/openai/whisper-base) on the DewiBrynJones/banc-trawsgrifiadau-bangor-clean train main, cymen-arfor/15awr train+dev+test main, wanasash/enwaucymraeg train+dev+test main, DewiBrynJones/commonvoice_18_0_cy_en train main dataset.
|
| 20 |
It achieves the following results on the evaluation set:
|
| 21 |
- Loss: 0.7135
|
| 22 |
- Wer: 0.5089
|
all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.6299918500407498,
|
| 3 |
+
"eval_loss": 0.7135367393493652,
|
| 4 |
+
"eval_runtime": 434.4772,
|
| 5 |
+
"eval_samples": 3901,
|
| 6 |
+
"eval_samples_per_second": 8.979,
|
| 7 |
+
"eval_steps_per_second": 0.562,
|
| 8 |
+
"eval_wer": 0.508886313145948,
|
| 9 |
+
"total_flos": 1.037708050563072e+19,
|
| 10 |
+
"train_loss": 1.5516241161346436,
|
| 11 |
+
"train_runtime": 17280.0888,
|
| 12 |
+
"train_samples": 98152,
|
| 13 |
+
"train_samples_per_second": 9.259,
|
| 14 |
+
"train_steps_per_second": 0.289
|
| 15 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.6299918500407498,
|
| 3 |
+
"eval_loss": 0.7135367393493652,
|
| 4 |
+
"eval_runtime": 434.4772,
|
| 5 |
+
"eval_samples": 3901,
|
| 6 |
+
"eval_samples_per_second": 8.979,
|
| 7 |
+
"eval_steps_per_second": 0.562,
|
| 8 |
+
"eval_wer": 0.508886313145948
|
| 9 |
+
}
|
runs/Nov02_13-48-22_daf84da58340/events.out.tfevents.1730574661.daf84da58340.37.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6313d25ebe3e69feabd91b67e145ffa8d9fd47b76ea1e501019fed680308ed32
|
| 3 |
+
size 406
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.6299918500407498,
|
| 3 |
+
"total_flos": 1.037708050563072e+19,
|
| 4 |
+
"train_loss": 1.5516241161346436,
|
| 5 |
+
"train_runtime": 17280.0888,
|
| 6 |
+
"train_samples": 98152,
|
| 7 |
+
"train_samples_per_second": 9.259,
|
| 8 |
+
"train_steps_per_second": 0.289
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1487 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.6299918500407498,
|
| 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": 0.008149959250203748,
|
| 13 |
+
"grad_norm": 34.218711853027344,
|
| 14 |
+
"learning_rate": 4.6000000000000004e-07,
|
| 15 |
+
"loss": 4.8764,
|
| 16 |
+
"step": 25
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.016299918500407497,
|
| 20 |
+
"grad_norm": 28.213876724243164,
|
| 21 |
+
"learning_rate": 9.600000000000001e-07,
|
| 22 |
+
"loss": 4.6956,
|
| 23 |
+
"step": 50
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.02444987775061125,
|
| 27 |
+
"grad_norm": 25.532794952392578,
|
| 28 |
+
"learning_rate": 1.46e-06,
|
| 29 |
+
"loss": 4.4548,
|
| 30 |
+
"step": 75
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.032599837000814993,
|
| 34 |
+
"grad_norm": 20.307037353515625,
|
| 35 |
+
"learning_rate": 1.9600000000000003e-06,
|
| 36 |
+
"loss": 4.0535,
|
| 37 |
+
"step": 100
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.040749796251018745,
|
| 41 |
+
"grad_norm": 20.705177307128906,
|
| 42 |
+
"learning_rate": 2.46e-06,
|
| 43 |
+
"loss": 3.7212,
|
| 44 |
+
"step": 125
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.0488997555012225,
|
| 48 |
+
"grad_norm": 20.950613021850586,
|
| 49 |
+
"learning_rate": 2.96e-06,
|
| 50 |
+
"loss": 3.3639,
|
| 51 |
+
"step": 150
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.05704971475142624,
|
| 55 |
+
"grad_norm": 22.719501495361328,
|
| 56 |
+
"learning_rate": 3.46e-06,
|
| 57 |
+
"loss": 3.2023,
|
| 58 |
+
"step": 175
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.06519967400162999,
|
| 62 |
+
"grad_norm": 21.53318977355957,
|
| 63 |
+
"learning_rate": 3.96e-06,
|
| 64 |
+
"loss": 3.1417,
|
| 65 |
+
"step": 200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.07334963325183375,
|
| 69 |
+
"grad_norm": 18.566478729248047,
|
| 70 |
+
"learning_rate": 4.4600000000000005e-06,
|
| 71 |
+
"loss": 2.915,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.08149959250203749,
|
| 76 |
+
"grad_norm": 21.804105758666992,
|
| 77 |
+
"learning_rate": 4.960000000000001e-06,
|
| 78 |
+
"loss": 2.8108,
|
| 79 |
+
"step": 250
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.08964955175224124,
|
| 83 |
+
"grad_norm": 20.4385929107666,
|
| 84 |
+
"learning_rate": 5.460000000000001e-06,
|
| 85 |
+
"loss": 2.7508,
|
| 86 |
+
"step": 275
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.097799511002445,
|
| 90 |
+
"grad_norm": 20.23823356628418,
|
| 91 |
+
"learning_rate": 5.9600000000000005e-06,
|
| 92 |
+
"loss": 2.7783,
|
| 93 |
+
"step": 300
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.10594947025264874,
|
| 97 |
+
"grad_norm": 18.606101989746094,
|
| 98 |
+
"learning_rate": 6.460000000000001e-06,
|
| 99 |
+
"loss": 2.6393,
|
| 100 |
+
"step": 325
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.11409942950285248,
|
| 104 |
+
"grad_norm": 18.78560447692871,
|
| 105 |
+
"learning_rate": 6.96e-06,
|
| 106 |
+
"loss": 2.4998,
|
| 107 |
+
"step": 350
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.12224938875305623,
|
| 111 |
+
"grad_norm": 18.703861236572266,
|
| 112 |
+
"learning_rate": 7.4600000000000006e-06,
|
| 113 |
+
"loss": 2.5621,
|
| 114 |
+
"step": 375
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.13039934800325997,
|
| 118 |
+
"grad_norm": 17.321996688842773,
|
| 119 |
+
"learning_rate": 7.960000000000002e-06,
|
| 120 |
+
"loss": 2.3896,
|
| 121 |
+
"step": 400
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.13854930725346373,
|
| 125 |
+
"grad_norm": 17.575088500976562,
|
| 126 |
+
"learning_rate": 8.46e-06,
|
| 127 |
+
"loss": 2.4155,
|
| 128 |
+
"step": 425
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.1466992665036675,
|
| 132 |
+
"grad_norm": 19.757349014282227,
|
| 133 |
+
"learning_rate": 8.96e-06,
|
| 134 |
+
"loss": 2.3126,
|
| 135 |
+
"step": 450
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.15484922575387122,
|
| 139 |
+
"grad_norm": 16.2225341796875,
|
| 140 |
+
"learning_rate": 9.460000000000001e-06,
|
| 141 |
+
"loss": 2.2792,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.16299918500407498,
|
| 146 |
+
"grad_norm": 20.059782028198242,
|
| 147 |
+
"learning_rate": 9.960000000000001e-06,
|
| 148 |
+
"loss": 2.2343,
|
| 149 |
+
"step": 500
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.17114914425427874,
|
| 153 |
+
"grad_norm": 17.442007064819336,
|
| 154 |
+
"learning_rate": 9.94888888888889e-06,
|
| 155 |
+
"loss": 2.2066,
|
| 156 |
+
"step": 525
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.17929910350448247,
|
| 160 |
+
"grad_norm": 18.0509033203125,
|
| 161 |
+
"learning_rate": 9.893333333333334e-06,
|
| 162 |
+
"loss": 2.1635,
|
| 163 |
+
"step": 550
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.18744906275468623,
|
| 167 |
+
"grad_norm": 16.07720375061035,
|
| 168 |
+
"learning_rate": 9.837777777777778e-06,
|
| 169 |
+
"loss": 2.086,
|
| 170 |
+
"step": 575
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.19559902200489,
|
| 174 |
+
"grad_norm": 19.739547729492188,
|
| 175 |
+
"learning_rate": 9.782222222222222e-06,
|
| 176 |
+
"loss": 2.0068,
|
| 177 |
+
"step": 600
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.20374898125509372,
|
| 181 |
+
"grad_norm": 17.144306182861328,
|
| 182 |
+
"learning_rate": 9.726666666666668e-06,
|
| 183 |
+
"loss": 1.9867,
|
| 184 |
+
"step": 625
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.21189894050529748,
|
| 188 |
+
"grad_norm": 17.193174362182617,
|
| 189 |
+
"learning_rate": 9.671111111111112e-06,
|
| 190 |
+
"loss": 1.9281,
|
| 191 |
+
"step": 650
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.2200488997555012,
|
| 195 |
+
"grad_norm": 16.084430694580078,
|
| 196 |
+
"learning_rate": 9.615555555555558e-06,
|
| 197 |
+
"loss": 1.9546,
|
| 198 |
+
"step": 675
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.22819885900570497,
|
| 202 |
+
"grad_norm": 15.92422103881836,
|
| 203 |
+
"learning_rate": 9.56e-06,
|
| 204 |
+
"loss": 1.9512,
|
| 205 |
+
"step": 700
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.23634881825590873,
|
| 209 |
+
"grad_norm": 17.584861755371094,
|
| 210 |
+
"learning_rate": 9.504444444444446e-06,
|
| 211 |
+
"loss": 1.9593,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.24449877750611246,
|
| 216 |
+
"grad_norm": 16.338075637817383,
|
| 217 |
+
"learning_rate": 9.44888888888889e-06,
|
| 218 |
+
"loss": 1.8831,
|
| 219 |
+
"step": 750
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.2526487367563162,
|
| 223 |
+
"grad_norm": 16.870452880859375,
|
| 224 |
+
"learning_rate": 9.393333333333334e-06,
|
| 225 |
+
"loss": 1.888,
|
| 226 |
+
"step": 775
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.26079869600651995,
|
| 230 |
+
"grad_norm": 14.461226463317871,
|
| 231 |
+
"learning_rate": 9.33777777777778e-06,
|
| 232 |
+
"loss": 1.8341,
|
| 233 |
+
"step": 800
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.26894865525672373,
|
| 237 |
+
"grad_norm": 18.07695770263672,
|
| 238 |
+
"learning_rate": 9.282222222222222e-06,
|
| 239 |
+
"loss": 1.854,
|
| 240 |
+
"step": 825
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.27709861450692747,
|
| 244 |
+
"grad_norm": 14.117804527282715,
|
| 245 |
+
"learning_rate": 9.226666666666668e-06,
|
| 246 |
+
"loss": 1.8034,
|
| 247 |
+
"step": 850
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.2852485737571312,
|
| 251 |
+
"grad_norm": 17.753490447998047,
|
| 252 |
+
"learning_rate": 9.171111111111112e-06,
|
| 253 |
+
"loss": 1.7948,
|
| 254 |
+
"step": 875
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.293398533007335,
|
| 258 |
+
"grad_norm": 15.982026100158691,
|
| 259 |
+
"learning_rate": 9.115555555555556e-06,
|
| 260 |
+
"loss": 1.7752,
|
| 261 |
+
"step": 900
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.3015484922575387,
|
| 265 |
+
"grad_norm": 15.503934860229492,
|
| 266 |
+
"learning_rate": 9.060000000000001e-06,
|
| 267 |
+
"loss": 1.7726,
|
| 268 |
+
"step": 925
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.30969845150774244,
|
| 272 |
+
"grad_norm": 18.943740844726562,
|
| 273 |
+
"learning_rate": 9.004444444444445e-06,
|
| 274 |
+
"loss": 1.773,
|
| 275 |
+
"step": 950
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.31784841075794623,
|
| 279 |
+
"grad_norm": 14.561786651611328,
|
| 280 |
+
"learning_rate": 8.94888888888889e-06,
|
| 281 |
+
"loss": 1.7569,
|
| 282 |
+
"step": 975
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.32599837000814996,
|
| 286 |
+
"grad_norm": 16.375633239746094,
|
| 287 |
+
"learning_rate": 8.893333333333333e-06,
|
| 288 |
+
"loss": 1.7363,
|
| 289 |
+
"step": 1000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.32599837000814996,
|
| 293 |
+
"eval_loss": 0.9845153093338013,
|
| 294 |
+
"eval_runtime": 457.2834,
|
| 295 |
+
"eval_samples_per_second": 8.531,
|
| 296 |
+
"eval_steps_per_second": 0.534,
|
| 297 |
+
"eval_wer": 0.6733314015261277,
|
| 298 |
+
"step": 1000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.3341483292583537,
|
| 302 |
+
"grad_norm": 14.428193092346191,
|
| 303 |
+
"learning_rate": 8.83777777777778e-06,
|
| 304 |
+
"loss": 1.731,
|
| 305 |
+
"step": 1025
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.3422982885085575,
|
| 309 |
+
"grad_norm": 15.243473052978516,
|
| 310 |
+
"learning_rate": 8.782222222222223e-06,
|
| 311 |
+
"loss": 1.7255,
|
| 312 |
+
"step": 1050
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.3504482477587612,
|
| 316 |
+
"grad_norm": 17.20938491821289,
|
| 317 |
+
"learning_rate": 8.726666666666667e-06,
|
| 318 |
+
"loss": 1.726,
|
| 319 |
+
"step": 1075
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.35859820700896494,
|
| 323 |
+
"grad_norm": 14.864399909973145,
|
| 324 |
+
"learning_rate": 8.671111111111113e-06,
|
| 325 |
+
"loss": 1.6554,
|
| 326 |
+
"step": 1100
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.36674816625916873,
|
| 330 |
+
"grad_norm": 15.314199447631836,
|
| 331 |
+
"learning_rate": 8.615555555555555e-06,
|
| 332 |
+
"loss": 1.7178,
|
| 333 |
+
"step": 1125
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.37489812550937246,
|
| 337 |
+
"grad_norm": 17.271549224853516,
|
| 338 |
+
"learning_rate": 8.560000000000001e-06,
|
| 339 |
+
"loss": 1.7197,
|
| 340 |
+
"step": 1150
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.3830480847595762,
|
| 344 |
+
"grad_norm": 16.468917846679688,
|
| 345 |
+
"learning_rate": 8.504444444444445e-06,
|
| 346 |
+
"loss": 1.677,
|
| 347 |
+
"step": 1175
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 0.39119804400978,
|
| 351 |
+
"grad_norm": 17.015993118286133,
|
| 352 |
+
"learning_rate": 8.448888888888889e-06,
|
| 353 |
+
"loss": 1.7056,
|
| 354 |
+
"step": 1200
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.3993480032599837,
|
| 358 |
+
"grad_norm": 16.396427154541016,
|
| 359 |
+
"learning_rate": 8.393333333333335e-06,
|
| 360 |
+
"loss": 1.6895,
|
| 361 |
+
"step": 1225
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 0.40749796251018744,
|
| 365 |
+
"grad_norm": 15.659330368041992,
|
| 366 |
+
"learning_rate": 8.337777777777777e-06,
|
| 367 |
+
"loss": 1.6117,
|
| 368 |
+
"step": 1250
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 0.4156479217603912,
|
| 372 |
+
"grad_norm": 18.017431259155273,
|
| 373 |
+
"learning_rate": 8.282222222222223e-06,
|
| 374 |
+
"loss": 1.6787,
|
| 375 |
+
"step": 1275
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.42379788101059496,
|
| 379 |
+
"grad_norm": 16.356630325317383,
|
| 380 |
+
"learning_rate": 8.226666666666667e-06,
|
| 381 |
+
"loss": 1.6565,
|
| 382 |
+
"step": 1300
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 0.4319478402607987,
|
| 386 |
+
"grad_norm": 14.949164390563965,
|
| 387 |
+
"learning_rate": 8.171111111111113e-06,
|
| 388 |
+
"loss": 1.5925,
|
| 389 |
+
"step": 1325
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 0.4400977995110024,
|
| 393 |
+
"grad_norm": 14.068632125854492,
|
| 394 |
+
"learning_rate": 8.115555555555557e-06,
|
| 395 |
+
"loss": 1.5924,
|
| 396 |
+
"step": 1350
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 0.4482477587612062,
|
| 400 |
+
"grad_norm": 17.71388816833496,
|
| 401 |
+
"learning_rate": 8.06e-06,
|
| 402 |
+
"loss": 1.6527,
|
| 403 |
+
"step": 1375
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.45639771801140994,
|
| 407 |
+
"grad_norm": 15.577716827392578,
|
| 408 |
+
"learning_rate": 8.004444444444445e-06,
|
| 409 |
+
"loss": 1.6216,
|
| 410 |
+
"step": 1400
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.46454767726161367,
|
| 414 |
+
"grad_norm": 14.626981735229492,
|
| 415 |
+
"learning_rate": 7.948888888888889e-06,
|
| 416 |
+
"loss": 1.5345,
|
| 417 |
+
"step": 1425
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.47269763651181745,
|
| 421 |
+
"grad_norm": 16.892454147338867,
|
| 422 |
+
"learning_rate": 7.893333333333335e-06,
|
| 423 |
+
"loss": 1.5763,
|
| 424 |
+
"step": 1450
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 0.4808475957620212,
|
| 428 |
+
"grad_norm": 18.04649543762207,
|
| 429 |
+
"learning_rate": 7.837777777777779e-06,
|
| 430 |
+
"loss": 1.6572,
|
| 431 |
+
"step": 1475
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.4889975550122249,
|
| 435 |
+
"grad_norm": 15.368581771850586,
|
| 436 |
+
"learning_rate": 7.782222222222223e-06,
|
| 437 |
+
"loss": 1.5422,
|
| 438 |
+
"step": 1500
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 0.4971475142624287,
|
| 442 |
+
"grad_norm": 14.239161491394043,
|
| 443 |
+
"learning_rate": 7.726666666666667e-06,
|
| 444 |
+
"loss": 1.5459,
|
| 445 |
+
"step": 1525
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 0.5052974735126324,
|
| 449 |
+
"grad_norm": 16.011531829833984,
|
| 450 |
+
"learning_rate": 7.67111111111111e-06,
|
| 451 |
+
"loss": 1.5944,
|
| 452 |
+
"step": 1550
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 0.5134474327628362,
|
| 456 |
+
"grad_norm": 15.135168075561523,
|
| 457 |
+
"learning_rate": 7.6155555555555564e-06,
|
| 458 |
+
"loss": 1.5705,
|
| 459 |
+
"step": 1575
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.5215973920130399,
|
| 463 |
+
"grad_norm": 17.60437774658203,
|
| 464 |
+
"learning_rate": 7.5600000000000005e-06,
|
| 465 |
+
"loss": 1.4844,
|
| 466 |
+
"step": 1600
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 0.5297473512632437,
|
| 470 |
+
"grad_norm": 16.583059310913086,
|
| 471 |
+
"learning_rate": 7.504444444444445e-06,
|
| 472 |
+
"loss": 1.5222,
|
| 473 |
+
"step": 1625
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 0.5378973105134475,
|
| 477 |
+
"grad_norm": 16.068408966064453,
|
| 478 |
+
"learning_rate": 7.44888888888889e-06,
|
| 479 |
+
"loss": 1.5425,
|
| 480 |
+
"step": 1650
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 0.5460472697636511,
|
| 484 |
+
"grad_norm": 14.316849708557129,
|
| 485 |
+
"learning_rate": 7.393333333333333e-06,
|
| 486 |
+
"loss": 1.4927,
|
| 487 |
+
"step": 1675
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 0.5541972290138549,
|
| 491 |
+
"grad_norm": 14.973306655883789,
|
| 492 |
+
"learning_rate": 7.337777777777778e-06,
|
| 493 |
+
"loss": 1.4961,
|
| 494 |
+
"step": 1700
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.5623471882640587,
|
| 498 |
+
"grad_norm": 16.13720703125,
|
| 499 |
+
"learning_rate": 7.282222222222222e-06,
|
| 500 |
+
"loss": 1.5945,
|
| 501 |
+
"step": 1725
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.5704971475142624,
|
| 505 |
+
"grad_norm": 15.234308242797852,
|
| 506 |
+
"learning_rate": 7.226666666666667e-06,
|
| 507 |
+
"loss": 1.5252,
|
| 508 |
+
"step": 1750
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 0.5786471067644662,
|
| 512 |
+
"grad_norm": 15.829686164855957,
|
| 513 |
+
"learning_rate": 7.171111111111112e-06,
|
| 514 |
+
"loss": 1.5186,
|
| 515 |
+
"step": 1775
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 0.58679706601467,
|
| 519 |
+
"grad_norm": 15.698533058166504,
|
| 520 |
+
"learning_rate": 7.115555555555557e-06,
|
| 521 |
+
"loss": 1.5347,
|
| 522 |
+
"step": 1800
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.5949470252648736,
|
| 526 |
+
"grad_norm": 16.133045196533203,
|
| 527 |
+
"learning_rate": 7.06e-06,
|
| 528 |
+
"loss": 1.538,
|
| 529 |
+
"step": 1825
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 0.6030969845150774,
|
| 533 |
+
"grad_norm": 15.596349716186523,
|
| 534 |
+
"learning_rate": 7.006666666666667e-06,
|
| 535 |
+
"loss": 1.4263,
|
| 536 |
+
"step": 1850
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 0.6112469437652812,
|
| 540 |
+
"grad_norm": 14.529706001281738,
|
| 541 |
+
"learning_rate": 6.951111111111112e-06,
|
| 542 |
+
"loss": 1.4455,
|
| 543 |
+
"step": 1875
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 0.6193969030154849,
|
| 547 |
+
"grad_norm": 16.734785079956055,
|
| 548 |
+
"learning_rate": 6.8955555555555565e-06,
|
| 549 |
+
"loss": 1.4879,
|
| 550 |
+
"step": 1900
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 0.6275468622656887,
|
| 554 |
+
"grad_norm": 14.141518592834473,
|
| 555 |
+
"learning_rate": 6.8400000000000014e-06,
|
| 556 |
+
"loss": 1.4886,
|
| 557 |
+
"step": 1925
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 0.6356968215158925,
|
| 561 |
+
"grad_norm": 13.811936378479004,
|
| 562 |
+
"learning_rate": 6.784444444444445e-06,
|
| 563 |
+
"loss": 1.4527,
|
| 564 |
+
"step": 1950
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 0.6438467807660961,
|
| 568 |
+
"grad_norm": 13.9077787399292,
|
| 569 |
+
"learning_rate": 6.7288888888888895e-06,
|
| 570 |
+
"loss": 1.4277,
|
| 571 |
+
"step": 1975
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 0.6519967400162999,
|
| 575 |
+
"grad_norm": 15.057381629943848,
|
| 576 |
+
"learning_rate": 6.6733333333333335e-06,
|
| 577 |
+
"loss": 1.4155,
|
| 578 |
+
"step": 2000
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 0.6519967400162999,
|
| 582 |
+
"eval_loss": 0.8236681818962097,
|
| 583 |
+
"eval_runtime": 437.2458,
|
| 584 |
+
"eval_samples_per_second": 8.922,
|
| 585 |
+
"eval_steps_per_second": 0.558,
|
| 586 |
+
"eval_wer": 0.571694194919347,
|
| 587 |
+
"step": 2000
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 0.6601466992665037,
|
| 591 |
+
"grad_norm": 15.232809066772461,
|
| 592 |
+
"learning_rate": 6.617777777777778e-06,
|
| 593 |
+
"loss": 1.4386,
|
| 594 |
+
"step": 2025
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 0.6682966585167074,
|
| 598 |
+
"grad_norm": 13.116307258605957,
|
| 599 |
+
"learning_rate": 6.562222222222223e-06,
|
| 600 |
+
"loss": 1.4113,
|
| 601 |
+
"step": 2050
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 0.6764466177669112,
|
| 605 |
+
"grad_norm": 17.645702362060547,
|
| 606 |
+
"learning_rate": 6.5066666666666665e-06,
|
| 607 |
+
"loss": 1.5237,
|
| 608 |
+
"step": 2075
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 0.684596577017115,
|
| 612 |
+
"grad_norm": 15.816476821899414,
|
| 613 |
+
"learning_rate": 6.451111111111111e-06,
|
| 614 |
+
"loss": 1.4043,
|
| 615 |
+
"step": 2100
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 0.6927465362673186,
|
| 619 |
+
"grad_norm": 17.2808780670166,
|
| 620 |
+
"learning_rate": 6.395555555555556e-06,
|
| 621 |
+
"loss": 1.4642,
|
| 622 |
+
"step": 2125
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 0.7008964955175224,
|
| 626 |
+
"grad_norm": 16.934045791625977,
|
| 627 |
+
"learning_rate": 6.34e-06,
|
| 628 |
+
"loss": 1.5066,
|
| 629 |
+
"step": 2150
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 0.7090464547677262,
|
| 633 |
+
"grad_norm": 16.353330612182617,
|
| 634 |
+
"learning_rate": 6.284444444444445e-06,
|
| 635 |
+
"loss": 1.4638,
|
| 636 |
+
"step": 2175
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 0.7171964140179299,
|
| 640 |
+
"grad_norm": 13.666918754577637,
|
| 641 |
+
"learning_rate": 6.22888888888889e-06,
|
| 642 |
+
"loss": 1.4241,
|
| 643 |
+
"step": 2200
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 0.7253463732681337,
|
| 647 |
+
"grad_norm": 16.40180206298828,
|
| 648 |
+
"learning_rate": 6.173333333333333e-06,
|
| 649 |
+
"loss": 1.4109,
|
| 650 |
+
"step": 2225
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 0.7334963325183375,
|
| 654 |
+
"grad_norm": 15.645243644714355,
|
| 655 |
+
"learning_rate": 6.117777777777778e-06,
|
| 656 |
+
"loss": 1.4354,
|
| 657 |
+
"step": 2250
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.7416462917685411,
|
| 661 |
+
"grad_norm": 14.662041664123535,
|
| 662 |
+
"learning_rate": 6.062222222222223e-06,
|
| 663 |
+
"loss": 1.4034,
|
| 664 |
+
"step": 2275
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 0.7497962510187449,
|
| 668 |
+
"grad_norm": 17.322463989257812,
|
| 669 |
+
"learning_rate": 6.006666666666667e-06,
|
| 670 |
+
"loss": 1.4348,
|
| 671 |
+
"step": 2300
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 0.7579462102689487,
|
| 675 |
+
"grad_norm": 15.498252868652344,
|
| 676 |
+
"learning_rate": 5.951111111111112e-06,
|
| 677 |
+
"loss": 1.4353,
|
| 678 |
+
"step": 2325
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 0.7660961695191524,
|
| 682 |
+
"grad_norm": 15.99593448638916,
|
| 683 |
+
"learning_rate": 5.895555555555557e-06,
|
| 684 |
+
"loss": 1.3877,
|
| 685 |
+
"step": 2350
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 0.7742461287693562,
|
| 689 |
+
"grad_norm": 15.517157554626465,
|
| 690 |
+
"learning_rate": 5.84e-06,
|
| 691 |
+
"loss": 1.4029,
|
| 692 |
+
"step": 2375
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 0.78239608801956,
|
| 696 |
+
"grad_norm": 14.675141334533691,
|
| 697 |
+
"learning_rate": 5.784444444444445e-06,
|
| 698 |
+
"loss": 1.4366,
|
| 699 |
+
"step": 2400
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.7905460472697636,
|
| 703 |
+
"grad_norm": 16.114215850830078,
|
| 704 |
+
"learning_rate": 5.72888888888889e-06,
|
| 705 |
+
"loss": 1.4032,
|
| 706 |
+
"step": 2425
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 0.7986960065199674,
|
| 710 |
+
"grad_norm": 15.094751358032227,
|
| 711 |
+
"learning_rate": 5.673333333333334e-06,
|
| 712 |
+
"loss": 1.4116,
|
| 713 |
+
"step": 2450
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 0.8068459657701712,
|
| 717 |
+
"grad_norm": 15.857089042663574,
|
| 718 |
+
"learning_rate": 5.617777777777779e-06,
|
| 719 |
+
"loss": 1.3738,
|
| 720 |
+
"step": 2475
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 0.8149959250203749,
|
| 724 |
+
"grad_norm": 16.033899307250977,
|
| 725 |
+
"learning_rate": 5.562222222222222e-06,
|
| 726 |
+
"loss": 1.4226,
|
| 727 |
+
"step": 2500
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 0.8231458842705787,
|
| 731 |
+
"grad_norm": 13.293587684631348,
|
| 732 |
+
"learning_rate": 5.506666666666667e-06,
|
| 733 |
+
"loss": 1.3407,
|
| 734 |
+
"step": 2525
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 0.8312958435207825,
|
| 738 |
+
"grad_norm": 14.656893730163574,
|
| 739 |
+
"learning_rate": 5.451111111111112e-06,
|
| 740 |
+
"loss": 1.3327,
|
| 741 |
+
"step": 2550
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 0.8394458027709861,
|
| 745 |
+
"grad_norm": 15.4556245803833,
|
| 746 |
+
"learning_rate": 5.3955555555555565e-06,
|
| 747 |
+
"loss": 1.3217,
|
| 748 |
+
"step": 2575
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 0.8475957620211899,
|
| 752 |
+
"grad_norm": 13.271966934204102,
|
| 753 |
+
"learning_rate": 5.3400000000000005e-06,
|
| 754 |
+
"loss": 1.3906,
|
| 755 |
+
"step": 2600
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 0.8557457212713936,
|
| 759 |
+
"grad_norm": 16.06142234802246,
|
| 760 |
+
"learning_rate": 5.2844444444444454e-06,
|
| 761 |
+
"loss": 1.3629,
|
| 762 |
+
"step": 2625
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 0.8638956805215974,
|
| 766 |
+
"grad_norm": 16.935834884643555,
|
| 767 |
+
"learning_rate": 5.228888888888889e-06,
|
| 768 |
+
"loss": 1.4123,
|
| 769 |
+
"step": 2650
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 0.8720456397718012,
|
| 773 |
+
"grad_norm": 16.714982986450195,
|
| 774 |
+
"learning_rate": 5.1733333333333335e-06,
|
| 775 |
+
"loss": 1.3887,
|
| 776 |
+
"step": 2675
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 0.8801955990220048,
|
| 780 |
+
"grad_norm": 13.961587905883789,
|
| 781 |
+
"learning_rate": 5.117777777777778e-06,
|
| 782 |
+
"loss": 1.3757,
|
| 783 |
+
"step": 2700
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 0.8883455582722086,
|
| 787 |
+
"grad_norm": 14.335005760192871,
|
| 788 |
+
"learning_rate": 5.062222222222222e-06,
|
| 789 |
+
"loss": 1.3393,
|
| 790 |
+
"step": 2725
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 0.8964955175224124,
|
| 794 |
+
"grad_norm": 13.483233451843262,
|
| 795 |
+
"learning_rate": 5.006666666666667e-06,
|
| 796 |
+
"loss": 1.3782,
|
| 797 |
+
"step": 2750
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 0.9046454767726161,
|
| 801 |
+
"grad_norm": 15.308544158935547,
|
| 802 |
+
"learning_rate": 4.951111111111111e-06,
|
| 803 |
+
"loss": 1.3422,
|
| 804 |
+
"step": 2775
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 0.9127954360228199,
|
| 808 |
+
"grad_norm": 11.132999420166016,
|
| 809 |
+
"learning_rate": 4.895555555555556e-06,
|
| 810 |
+
"loss": 1.3534,
|
| 811 |
+
"step": 2800
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 0.9209453952730237,
|
| 815 |
+
"grad_norm": 15.973653793334961,
|
| 816 |
+
"learning_rate": 4.84e-06,
|
| 817 |
+
"loss": 1.4114,
|
| 818 |
+
"step": 2825
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 0.9290953545232273,
|
| 822 |
+
"grad_norm": 14.977301597595215,
|
| 823 |
+
"learning_rate": 4.784444444444445e-06,
|
| 824 |
+
"loss": 1.3283,
|
| 825 |
+
"step": 2850
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 0.9372453137734311,
|
| 829 |
+
"grad_norm": 16.090097427368164,
|
| 830 |
+
"learning_rate": 4.728888888888889e-06,
|
| 831 |
+
"loss": 1.2659,
|
| 832 |
+
"step": 2875
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 0.9453952730236349,
|
| 836 |
+
"grad_norm": 13.608882904052734,
|
| 837 |
+
"learning_rate": 4.673333333333333e-06,
|
| 838 |
+
"loss": 1.2446,
|
| 839 |
+
"step": 2900
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 0.9535452322738386,
|
| 843 |
+
"grad_norm": 15.259020805358887,
|
| 844 |
+
"learning_rate": 4.617777777777778e-06,
|
| 845 |
+
"loss": 1.2993,
|
| 846 |
+
"step": 2925
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 0.9616951915240424,
|
| 850 |
+
"grad_norm": 13.211849212646484,
|
| 851 |
+
"learning_rate": 4.562222222222222e-06,
|
| 852 |
+
"loss": 1.3024,
|
| 853 |
+
"step": 2950
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 0.9698451507742462,
|
| 857 |
+
"grad_norm": 16.002653121948242,
|
| 858 |
+
"learning_rate": 4.506666666666667e-06,
|
| 859 |
+
"loss": 1.3546,
|
| 860 |
+
"step": 2975
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 0.9779951100244498,
|
| 864 |
+
"grad_norm": 13.069381713867188,
|
| 865 |
+
"learning_rate": 4.451111111111112e-06,
|
| 866 |
+
"loss": 1.3809,
|
| 867 |
+
"step": 3000
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 0.9779951100244498,
|
| 871 |
+
"eval_loss": 0.7566213607788086,
|
| 872 |
+
"eval_runtime": 438.5427,
|
| 873 |
+
"eval_samples_per_second": 8.895,
|
| 874 |
+
"eval_steps_per_second": 0.556,
|
| 875 |
+
"eval_wer": 0.5274075147300299,
|
| 876 |
+
"step": 3000
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 0.9861450692746536,
|
| 880 |
+
"grad_norm": 14.120579719543457,
|
| 881 |
+
"learning_rate": 4.395555555555556e-06,
|
| 882 |
+
"loss": 1.3394,
|
| 883 |
+
"step": 3025
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 0.9942950285248574,
|
| 887 |
+
"grad_norm": 20.488895416259766,
|
| 888 |
+
"learning_rate": 4.34e-06,
|
| 889 |
+
"loss": 1.3911,
|
| 890 |
+
"step": 3050
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 1.0024449877750612,
|
| 894 |
+
"grad_norm": 13.28896713256836,
|
| 895 |
+
"learning_rate": 4.284444444444445e-06,
|
| 896 |
+
"loss": 1.2861,
|
| 897 |
+
"step": 3075
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 1.0105949470252649,
|
| 901 |
+
"grad_norm": 13.970645904541016,
|
| 902 |
+
"learning_rate": 4.228888888888889e-06,
|
| 903 |
+
"loss": 1.146,
|
| 904 |
+
"step": 3100
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 1.0187449062754685,
|
| 908 |
+
"grad_norm": 13.756807327270508,
|
| 909 |
+
"learning_rate": 4.173333333333334e-06,
|
| 910 |
+
"loss": 1.1688,
|
| 911 |
+
"step": 3125
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 1.0268948655256724,
|
| 915 |
+
"grad_norm": 14.576444625854492,
|
| 916 |
+
"learning_rate": 4.117777777777779e-06,
|
| 917 |
+
"loss": 1.1595,
|
| 918 |
+
"step": 3150
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 1.0350448247758761,
|
| 922 |
+
"grad_norm": 12.942944526672363,
|
| 923 |
+
"learning_rate": 4.062222222222223e-06,
|
| 924 |
+
"loss": 1.2039,
|
| 925 |
+
"step": 3175
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 1.0431947840260798,
|
| 929 |
+
"grad_norm": 13.682048797607422,
|
| 930 |
+
"learning_rate": 4.006666666666667e-06,
|
| 931 |
+
"loss": 1.18,
|
| 932 |
+
"step": 3200
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 1.0513447432762837,
|
| 936 |
+
"grad_norm": 14.890068054199219,
|
| 937 |
+
"learning_rate": 3.951111111111112e-06,
|
| 938 |
+
"loss": 1.1905,
|
| 939 |
+
"step": 3225
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 1.0594947025264874,
|
| 943 |
+
"grad_norm": 14.280926704406738,
|
| 944 |
+
"learning_rate": 3.895555555555556e-06,
|
| 945 |
+
"loss": 1.1572,
|
| 946 |
+
"step": 3250
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 1.067644661776691,
|
| 950 |
+
"grad_norm": 12.664419174194336,
|
| 951 |
+
"learning_rate": 3.8400000000000005e-06,
|
| 952 |
+
"loss": 1.1405,
|
| 953 |
+
"step": 3275
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 1.075794621026895,
|
| 957 |
+
"grad_norm": 13.912672996520996,
|
| 958 |
+
"learning_rate": 3.784444444444445e-06,
|
| 959 |
+
"loss": 1.1779,
|
| 960 |
+
"step": 3300
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 1.0839445802770986,
|
| 964 |
+
"grad_norm": 13.749354362487793,
|
| 965 |
+
"learning_rate": 3.728888888888889e-06,
|
| 966 |
+
"loss": 1.1534,
|
| 967 |
+
"step": 3325
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 1.0920945395273023,
|
| 971 |
+
"grad_norm": 13.6649169921875,
|
| 972 |
+
"learning_rate": 3.673333333333334e-06,
|
| 973 |
+
"loss": 1.1381,
|
| 974 |
+
"step": 3350
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 1.1002444987775062,
|
| 978 |
+
"grad_norm": 13.478617668151855,
|
| 979 |
+
"learning_rate": 3.617777777777778e-06,
|
| 980 |
+
"loss": 1.2095,
|
| 981 |
+
"step": 3375
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 1.1083944580277099,
|
| 985 |
+
"grad_norm": 13.082009315490723,
|
| 986 |
+
"learning_rate": 3.5622222222222224e-06,
|
| 987 |
+
"loss": 1.1965,
|
| 988 |
+
"step": 3400
|
| 989 |
+
},
|
| 990 |
+
{
|
| 991 |
+
"epoch": 1.1165444172779135,
|
| 992 |
+
"grad_norm": 13.995879173278809,
|
| 993 |
+
"learning_rate": 3.5066666666666673e-06,
|
| 994 |
+
"loss": 1.1418,
|
| 995 |
+
"step": 3425
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 1.1246943765281174,
|
| 999 |
+
"grad_norm": 13.161800384521484,
|
| 1000 |
+
"learning_rate": 3.4511111111111113e-06,
|
| 1001 |
+
"loss": 1.1888,
|
| 1002 |
+
"step": 3450
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 1.132844335778321,
|
| 1006 |
+
"grad_norm": 12.4346342086792,
|
| 1007 |
+
"learning_rate": 3.3955555555555558e-06,
|
| 1008 |
+
"loss": 1.1425,
|
| 1009 |
+
"step": 3475
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 1.1409942950285248,
|
| 1013 |
+
"grad_norm": 14.341651916503906,
|
| 1014 |
+
"learning_rate": 3.3400000000000006e-06,
|
| 1015 |
+
"loss": 1.136,
|
| 1016 |
+
"step": 3500
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 1.1491442542787287,
|
| 1020 |
+
"grad_norm": 11.676952362060547,
|
| 1021 |
+
"learning_rate": 3.2844444444444447e-06,
|
| 1022 |
+
"loss": 1.1345,
|
| 1023 |
+
"step": 3525
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 1.1572942135289324,
|
| 1027 |
+
"grad_norm": 14.452948570251465,
|
| 1028 |
+
"learning_rate": 3.228888888888889e-06,
|
| 1029 |
+
"loss": 1.1602,
|
| 1030 |
+
"step": 3550
|
| 1031 |
+
},
|
| 1032 |
+
{
|
| 1033 |
+
"epoch": 1.165444172779136,
|
| 1034 |
+
"grad_norm": 13.384020805358887,
|
| 1035 |
+
"learning_rate": 3.173333333333334e-06,
|
| 1036 |
+
"loss": 1.159,
|
| 1037 |
+
"step": 3575
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 1.17359413202934,
|
| 1041 |
+
"grad_norm": 13.295368194580078,
|
| 1042 |
+
"learning_rate": 3.117777777777778e-06,
|
| 1043 |
+
"loss": 1.1146,
|
| 1044 |
+
"step": 3600
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 1.1817440912795436,
|
| 1048 |
+
"grad_norm": 13.34139633178711,
|
| 1049 |
+
"learning_rate": 3.0622222222222225e-06,
|
| 1050 |
+
"loss": 1.1496,
|
| 1051 |
+
"step": 3625
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 1.1898940505297473,
|
| 1055 |
+
"grad_norm": 15.428239822387695,
|
| 1056 |
+
"learning_rate": 3.0066666666666674e-06,
|
| 1057 |
+
"loss": 1.2243,
|
| 1058 |
+
"step": 3650
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 1.1980440097799512,
|
| 1062 |
+
"grad_norm": 13.666820526123047,
|
| 1063 |
+
"learning_rate": 2.9511111111111114e-06,
|
| 1064 |
+
"loss": 1.1792,
|
| 1065 |
+
"step": 3675
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 1.2061939690301549,
|
| 1069 |
+
"grad_norm": 11.906725883483887,
|
| 1070 |
+
"learning_rate": 2.895555555555556e-06,
|
| 1071 |
+
"loss": 1.135,
|
| 1072 |
+
"step": 3700
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"epoch": 1.2143439282803585,
|
| 1076 |
+
"grad_norm": 14.56628131866455,
|
| 1077 |
+
"learning_rate": 2.84e-06,
|
| 1078 |
+
"loss": 1.134,
|
| 1079 |
+
"step": 3725
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 1.2224938875305624,
|
| 1083 |
+
"grad_norm": 14.18961238861084,
|
| 1084 |
+
"learning_rate": 2.784444444444445e-06,
|
| 1085 |
+
"loss": 1.1115,
|
| 1086 |
+
"step": 3750
|
| 1087 |
+
},
|
| 1088 |
+
{
|
| 1089 |
+
"epoch": 1.230643846780766,
|
| 1090 |
+
"grad_norm": 14.074260711669922,
|
| 1091 |
+
"learning_rate": 2.7288888888888893e-06,
|
| 1092 |
+
"loss": 1.1592,
|
| 1093 |
+
"step": 3775
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"epoch": 1.2387938060309698,
|
| 1097 |
+
"grad_norm": 14.50068473815918,
|
| 1098 |
+
"learning_rate": 2.6733333333333333e-06,
|
| 1099 |
+
"loss": 1.1568,
|
| 1100 |
+
"step": 3800
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 1.2469437652811737,
|
| 1104 |
+
"grad_norm": 14.552042007446289,
|
| 1105 |
+
"learning_rate": 2.617777777777778e-06,
|
| 1106 |
+
"loss": 1.1456,
|
| 1107 |
+
"step": 3825
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 1.2550937245313774,
|
| 1111 |
+
"grad_norm": 13.851117134094238,
|
| 1112 |
+
"learning_rate": 2.5622222222222226e-06,
|
| 1113 |
+
"loss": 1.1469,
|
| 1114 |
+
"step": 3850
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 1.263243683781581,
|
| 1118 |
+
"grad_norm": 13.306814193725586,
|
| 1119 |
+
"learning_rate": 2.5066666666666667e-06,
|
| 1120 |
+
"loss": 1.1595,
|
| 1121 |
+
"step": 3875
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 1.271393643031785,
|
| 1125 |
+
"grad_norm": 14.614799499511719,
|
| 1126 |
+
"learning_rate": 2.451111111111111e-06,
|
| 1127 |
+
"loss": 1.148,
|
| 1128 |
+
"step": 3900
|
| 1129 |
+
},
|
| 1130 |
+
{
|
| 1131 |
+
"epoch": 1.2795436022819886,
|
| 1132 |
+
"grad_norm": 13.00090217590332,
|
| 1133 |
+
"learning_rate": 2.3955555555555556e-06,
|
| 1134 |
+
"loss": 1.1127,
|
| 1135 |
+
"step": 3925
|
| 1136 |
+
},
|
| 1137 |
+
{
|
| 1138 |
+
"epoch": 1.2876935615321923,
|
| 1139 |
+
"grad_norm": 12.845821380615234,
|
| 1140 |
+
"learning_rate": 2.3400000000000005e-06,
|
| 1141 |
+
"loss": 1.108,
|
| 1142 |
+
"step": 3950
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 1.295843520782396,
|
| 1146 |
+
"grad_norm": 13.80776596069336,
|
| 1147 |
+
"learning_rate": 2.2844444444444445e-06,
|
| 1148 |
+
"loss": 1.1851,
|
| 1149 |
+
"step": 3975
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 1.3039934800325998,
|
| 1153 |
+
"grad_norm": 12.853269577026367,
|
| 1154 |
+
"learning_rate": 2.228888888888889e-06,
|
| 1155 |
+
"loss": 1.0882,
|
| 1156 |
+
"step": 4000
|
| 1157 |
+
},
|
| 1158 |
+
{
|
| 1159 |
+
"epoch": 1.3039934800325998,
|
| 1160 |
+
"eval_loss": 0.7265446186065674,
|
| 1161 |
+
"eval_runtime": 440.5369,
|
| 1162 |
+
"eval_samples_per_second": 8.855,
|
| 1163 |
+
"eval_steps_per_second": 0.554,
|
| 1164 |
+
"eval_wer": 0.5286148942335555,
|
| 1165 |
+
"step": 4000
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 1.3121434392828035,
|
| 1169 |
+
"grad_norm": 13.121480941772461,
|
| 1170 |
+
"learning_rate": 2.1733333333333334e-06,
|
| 1171 |
+
"loss": 1.1081,
|
| 1172 |
+
"step": 4025
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 1.3202933985330074,
|
| 1176 |
+
"grad_norm": 14.560317039489746,
|
| 1177 |
+
"learning_rate": 2.117777777777778e-06,
|
| 1178 |
+
"loss": 1.1418,
|
| 1179 |
+
"step": 4050
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 1.328443357783211,
|
| 1183 |
+
"grad_norm": 13.478865623474121,
|
| 1184 |
+
"learning_rate": 2.0622222222222223e-06,
|
| 1185 |
+
"loss": 1.1296,
|
| 1186 |
+
"step": 4075
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 1.3365933170334148,
|
| 1190 |
+
"grad_norm": 12.93165397644043,
|
| 1191 |
+
"learning_rate": 2.006666666666667e-06,
|
| 1192 |
+
"loss": 1.1609,
|
| 1193 |
+
"step": 4100
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 1.3447432762836184,
|
| 1197 |
+
"grad_norm": 13.513744354248047,
|
| 1198 |
+
"learning_rate": 1.9511111111111113e-06,
|
| 1199 |
+
"loss": 1.099,
|
| 1200 |
+
"step": 4125
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 1.3528932355338223,
|
| 1204 |
+
"grad_norm": 13.625751495361328,
|
| 1205 |
+
"learning_rate": 1.8955555555555557e-06,
|
| 1206 |
+
"loss": 1.0996,
|
| 1207 |
+
"step": 4150
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 1.361043194784026,
|
| 1211 |
+
"grad_norm": 13.141559600830078,
|
| 1212 |
+
"learning_rate": 1.8400000000000002e-06,
|
| 1213 |
+
"loss": 1.1522,
|
| 1214 |
+
"step": 4175
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 1.36919315403423,
|
| 1218 |
+
"grad_norm": 13.436763763427734,
|
| 1219 |
+
"learning_rate": 1.7844444444444444e-06,
|
| 1220 |
+
"loss": 1.188,
|
| 1221 |
+
"step": 4200
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 1.3773431132844336,
|
| 1225 |
+
"grad_norm": 13.628213882446289,
|
| 1226 |
+
"learning_rate": 1.728888888888889e-06,
|
| 1227 |
+
"loss": 1.0837,
|
| 1228 |
+
"step": 4225
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 1.3854930725346373,
|
| 1232 |
+
"grad_norm": 12.989063262939453,
|
| 1233 |
+
"learning_rate": 1.6733333333333335e-06,
|
| 1234 |
+
"loss": 1.1027,
|
| 1235 |
+
"step": 4250
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 1.393643031784841,
|
| 1239 |
+
"grad_norm": 13.773276329040527,
|
| 1240 |
+
"learning_rate": 1.6177777777777778e-06,
|
| 1241 |
+
"loss": 1.1276,
|
| 1242 |
+
"step": 4275
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 1.4017929910350448,
|
| 1246 |
+
"grad_norm": 13.921293258666992,
|
| 1247 |
+
"learning_rate": 1.5622222222222225e-06,
|
| 1248 |
+
"loss": 1.0993,
|
| 1249 |
+
"step": 4300
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 1.4099429502852485,
|
| 1253 |
+
"grad_norm": 13.18332290649414,
|
| 1254 |
+
"learning_rate": 1.506666666666667e-06,
|
| 1255 |
+
"loss": 1.0903,
|
| 1256 |
+
"step": 4325
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 1.4180929095354524,
|
| 1260 |
+
"grad_norm": 14.134808540344238,
|
| 1261 |
+
"learning_rate": 1.4511111111111112e-06,
|
| 1262 |
+
"loss": 1.0599,
|
| 1263 |
+
"step": 4350
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 1.426242868785656,
|
| 1267 |
+
"grad_norm": 13.665396690368652,
|
| 1268 |
+
"learning_rate": 1.3955555555555556e-06,
|
| 1269 |
+
"loss": 1.0808,
|
| 1270 |
+
"step": 4375
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 1.4343928280358598,
|
| 1274 |
+
"grad_norm": 14.74458122253418,
|
| 1275 |
+
"learning_rate": 1.34e-06,
|
| 1276 |
+
"loss": 1.1597,
|
| 1277 |
+
"step": 4400
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 1.4425427872860634,
|
| 1281 |
+
"grad_norm": 13.300642013549805,
|
| 1282 |
+
"learning_rate": 1.2844444444444445e-06,
|
| 1283 |
+
"loss": 1.1345,
|
| 1284 |
+
"step": 4425
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 1.4506927465362673,
|
| 1288 |
+
"grad_norm": 12.414579391479492,
|
| 1289 |
+
"learning_rate": 1.228888888888889e-06,
|
| 1290 |
+
"loss": 1.1019,
|
| 1291 |
+
"step": 4450
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 1.458842705786471,
|
| 1295 |
+
"grad_norm": 16.401771545410156,
|
| 1296 |
+
"learning_rate": 1.1733333333333335e-06,
|
| 1297 |
+
"loss": 1.1428,
|
| 1298 |
+
"step": 4475
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 1.466992665036675,
|
| 1302 |
+
"grad_norm": 14.687612533569336,
|
| 1303 |
+
"learning_rate": 1.117777777777778e-06,
|
| 1304 |
+
"loss": 1.1126,
|
| 1305 |
+
"step": 4500
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 1.4751426242868786,
|
| 1309 |
+
"grad_norm": 12.354652404785156,
|
| 1310 |
+
"learning_rate": 1.0622222222222222e-06,
|
| 1311 |
+
"loss": 1.145,
|
| 1312 |
+
"step": 4525
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 1.4832925835370823,
|
| 1316 |
+
"grad_norm": 14.474491119384766,
|
| 1317 |
+
"learning_rate": 1.0066666666666668e-06,
|
| 1318 |
+
"loss": 1.1362,
|
| 1319 |
+
"step": 4550
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 1.491442542787286,
|
| 1323 |
+
"grad_norm": 13.144413948059082,
|
| 1324 |
+
"learning_rate": 9.511111111111111e-07,
|
| 1325 |
+
"loss": 1.1293,
|
| 1326 |
+
"step": 4575
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 1.4995925020374898,
|
| 1330 |
+
"grad_norm": 12.695611953735352,
|
| 1331 |
+
"learning_rate": 8.955555555555557e-07,
|
| 1332 |
+
"loss": 1.0901,
|
| 1333 |
+
"step": 4600
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 1.5077424612876935,
|
| 1337 |
+
"grad_norm": 16.430057525634766,
|
| 1338 |
+
"learning_rate": 8.400000000000001e-07,
|
| 1339 |
+
"loss": 1.1373,
|
| 1340 |
+
"step": 4625
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 1.5158924205378974,
|
| 1344 |
+
"grad_norm": 13.695208549499512,
|
| 1345 |
+
"learning_rate": 7.844444444444445e-07,
|
| 1346 |
+
"loss": 1.12,
|
| 1347 |
+
"step": 4650
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 1.524042379788101,
|
| 1351 |
+
"grad_norm": 14.039958000183105,
|
| 1352 |
+
"learning_rate": 7.28888888888889e-07,
|
| 1353 |
+
"loss": 1.1394,
|
| 1354 |
+
"step": 4675
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 1.5321923390383048,
|
| 1358 |
+
"grad_norm": 14.492539405822754,
|
| 1359 |
+
"learning_rate": 6.733333333333334e-07,
|
| 1360 |
+
"loss": 1.1385,
|
| 1361 |
+
"step": 4700
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 1.5403422982885084,
|
| 1365 |
+
"grad_norm": 15.331225395202637,
|
| 1366 |
+
"learning_rate": 6.177777777777778e-07,
|
| 1367 |
+
"loss": 1.119,
|
| 1368 |
+
"step": 4725
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 1.5484922575387123,
|
| 1372 |
+
"grad_norm": 12.952714920043945,
|
| 1373 |
+
"learning_rate": 5.622222222222223e-07,
|
| 1374 |
+
"loss": 1.1479,
|
| 1375 |
+
"step": 4750
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 1.556642216788916,
|
| 1379 |
+
"grad_norm": 13.869738578796387,
|
| 1380 |
+
"learning_rate": 5.066666666666667e-07,
|
| 1381 |
+
"loss": 1.0884,
|
| 1382 |
+
"step": 4775
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 1.56479217603912,
|
| 1386 |
+
"grad_norm": 13.278375625610352,
|
| 1387 |
+
"learning_rate": 4.511111111111111e-07,
|
| 1388 |
+
"loss": 1.1049,
|
| 1389 |
+
"step": 4800
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 1.5729421352893236,
|
| 1393 |
+
"grad_norm": 12.596354484558105,
|
| 1394 |
+
"learning_rate": 3.9555555555555557e-07,
|
| 1395 |
+
"loss": 1.0696,
|
| 1396 |
+
"step": 4825
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 1.5810920945395273,
|
| 1400 |
+
"grad_norm": 13.701985359191895,
|
| 1401 |
+
"learning_rate": 3.4000000000000003e-07,
|
| 1402 |
+
"loss": 1.1376,
|
| 1403 |
+
"step": 4850
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 1.589242053789731,
|
| 1407 |
+
"grad_norm": 12.712823867797852,
|
| 1408 |
+
"learning_rate": 2.844444444444445e-07,
|
| 1409 |
+
"loss": 1.1158,
|
| 1410 |
+
"step": 4875
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 1.5973920130399348,
|
| 1414 |
+
"grad_norm": 12.259002685546875,
|
| 1415 |
+
"learning_rate": 2.2888888888888892e-07,
|
| 1416 |
+
"loss": 1.094,
|
| 1417 |
+
"step": 4900
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 1.6055419722901385,
|
| 1421 |
+
"grad_norm": 14.35509204864502,
|
| 1422 |
+
"learning_rate": 1.7333333333333335e-07,
|
| 1423 |
+
"loss": 1.1516,
|
| 1424 |
+
"step": 4925
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 1.6136919315403424,
|
| 1428 |
+
"grad_norm": 12.94543170928955,
|
| 1429 |
+
"learning_rate": 1.1777777777777778e-07,
|
| 1430 |
+
"loss": 1.0555,
|
| 1431 |
+
"step": 4950
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 1.621841890790546,
|
| 1435 |
+
"grad_norm": 14.701851844787598,
|
| 1436 |
+
"learning_rate": 6.222222222222223e-08,
|
| 1437 |
+
"loss": 1.0818,
|
| 1438 |
+
"step": 4975
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 1.6299918500407498,
|
| 1442 |
+
"grad_norm": 13.011249542236328,
|
| 1443 |
+
"learning_rate": 6.666666666666667e-09,
|
| 1444 |
+
"loss": 1.0903,
|
| 1445 |
+
"step": 5000
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 1.6299918500407498,
|
| 1449 |
+
"eval_loss": 0.7135367393493652,
|
| 1450 |
+
"eval_runtime": 442.8745,
|
| 1451 |
+
"eval_samples_per_second": 8.808,
|
| 1452 |
+
"eval_steps_per_second": 0.551,
|
| 1453 |
+
"eval_wer": 0.508886313145948,
|
| 1454 |
+
"step": 5000
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 1.6299918500407498,
|
| 1458 |
+
"step": 5000,
|
| 1459 |
+
"total_flos": 1.037708050563072e+19,
|
| 1460 |
+
"train_loss": 1.5516241161346436,
|
| 1461 |
+
"train_runtime": 17280.0888,
|
| 1462 |
+
"train_samples_per_second": 9.259,
|
| 1463 |
+
"train_steps_per_second": 0.289
|
| 1464 |
+
}
|
| 1465 |
+
],
|
| 1466 |
+
"logging_steps": 25,
|
| 1467 |
+
"max_steps": 5000,
|
| 1468 |
+
"num_input_tokens_seen": 0,
|
| 1469 |
+
"num_train_epochs": 2,
|
| 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": 1.037708050563072e+19,
|
| 1484 |
+
"train_batch_size": 16,
|
| 1485 |
+
"trial_name": null,
|
| 1486 |
+
"trial_params": null
|
| 1487 |
+
}
|