End of training
Browse files- README.md +1 -1
- all_results.json +15 -0
- eval_results.json +9 -0
- runs/Nov04_13-33-34_b8b767113626/events.out.tfevents.1730790046.b8b767113626.478.1 +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-large-v3-ft-btbn-ca
|
| 18 |
|
| 19 |
-
This model is a fine-tuned version of [openai/whisper-large-v3](https://huggingface.co/openai/whisper-large-v3) on
|
| 20 |
It achieves the following results on the evaluation set:
|
| 21 |
- Loss: 0.4075
|
| 22 |
- Wer: 0.2824
|
|
|
|
| 16 |
|
| 17 |
# whisper-large-v3-ft-btbn-ca
|
| 18 |
|
| 19 |
+
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 train main, cymen-arfor/15awr train+dev+test main dataset.
|
| 20 |
It achieves the following results on the evaluation set:
|
| 21 |
- Loss: 0.4075
|
| 22 |
- Wer: 0.2824
|
all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.1938677738741617,
|
| 3 |
+
"eval_loss": 0.4074931740760803,
|
| 4 |
+
"eval_runtime": 1638.5657,
|
| 5 |
+
"eval_samples": 3901,
|
| 6 |
+
"eval_samples_per_second": 2.381,
|
| 7 |
+
"eval_steps_per_second": 0.149,
|
| 8 |
+
"eval_wer": 0.2823846789481359,
|
| 9 |
+
"total_flos": 5.435895365546803e+20,
|
| 10 |
+
"train_loss": 0.7105956988334656,
|
| 11 |
+
"train_runtime": 59674.9515,
|
| 12 |
+
"train_samples": 50095,
|
| 13 |
+
"train_samples_per_second": 2.681,
|
| 14 |
+
"train_steps_per_second": 0.084
|
| 15 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.1938677738741617,
|
| 3 |
+
"eval_loss": 0.4074931740760803,
|
| 4 |
+
"eval_runtime": 1638.5657,
|
| 5 |
+
"eval_samples": 3901,
|
| 6 |
+
"eval_samples_per_second": 2.381,
|
| 7 |
+
"eval_steps_per_second": 0.149,
|
| 8 |
+
"eval_wer": 0.2823846789481359
|
| 9 |
+
}
|
runs/Nov04_13-33-34_b8b767113626/events.out.tfevents.1730790046.b8b767113626.478.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b7168eff838fb3a36589d7f20fc726d82a301aeb640085d791f4d58b6535f765
|
| 3 |
+
size 406
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.1938677738741617,
|
| 3 |
+
"total_flos": 5.435895365546803e+20,
|
| 4 |
+
"train_loss": 0.7105956988334656,
|
| 5 |
+
"train_runtime": 59674.9515,
|
| 6 |
+
"train_samples": 50095,
|
| 7 |
+
"train_samples_per_second": 2.681,
|
| 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": 3.1938677738741617,
|
| 5 |
+
"eval_steps": 1000,
|
| 6 |
+
"global_step": 5000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.015969338869370808,
|
| 13 |
+
"grad_norm": 17.92350196838379,
|
| 14 |
+
"learning_rate": 4.800000000000001e-07,
|
| 15 |
+
"loss": 3.223,
|
| 16 |
+
"step": 25
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.031938677738741615,
|
| 20 |
+
"grad_norm": 11.101584434509277,
|
| 21 |
+
"learning_rate": 9.800000000000001e-07,
|
| 22 |
+
"loss": 2.5321,
|
| 23 |
+
"step": 50
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.047908016608112426,
|
| 27 |
+
"grad_norm": 9.936028480529785,
|
| 28 |
+
"learning_rate": 1.48e-06,
|
| 29 |
+
"loss": 2.0032,
|
| 30 |
+
"step": 75
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.06387735547748323,
|
| 34 |
+
"grad_norm": 10.770648956298828,
|
| 35 |
+
"learning_rate": 1.98e-06,
|
| 36 |
+
"loss": 1.6134,
|
| 37 |
+
"step": 100
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.07984669434685404,
|
| 41 |
+
"grad_norm": 10.68463134765625,
|
| 42 |
+
"learning_rate": 2.4800000000000004e-06,
|
| 43 |
+
"loss": 1.5694,
|
| 44 |
+
"step": 125
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.09581603321622485,
|
| 48 |
+
"grad_norm": 10.635124206542969,
|
| 49 |
+
"learning_rate": 2.9800000000000003e-06,
|
| 50 |
+
"loss": 1.4934,
|
| 51 |
+
"step": 150
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.11178537208559565,
|
| 55 |
+
"grad_norm": 9.225995063781738,
|
| 56 |
+
"learning_rate": 3.48e-06,
|
| 57 |
+
"loss": 1.518,
|
| 58 |
+
"step": 175
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.12775471095496646,
|
| 62 |
+
"grad_norm": 9.05696964263916,
|
| 63 |
+
"learning_rate": 3.980000000000001e-06,
|
| 64 |
+
"loss": 1.4258,
|
| 65 |
+
"step": 200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.14372404982433729,
|
| 69 |
+
"grad_norm": 9.599803924560547,
|
| 70 |
+
"learning_rate": 4.48e-06,
|
| 71 |
+
"loss": 1.422,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.15969338869370808,
|
| 76 |
+
"grad_norm": 9.442488670349121,
|
| 77 |
+
"learning_rate": 4.980000000000001e-06,
|
| 78 |
+
"loss": 1.3208,
|
| 79 |
+
"step": 250
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.17566272756307888,
|
| 83 |
+
"grad_norm": 10.746581077575684,
|
| 84 |
+
"learning_rate": 5.480000000000001e-06,
|
| 85 |
+
"loss": 1.3997,
|
| 86 |
+
"step": 275
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.1916320664324497,
|
| 90 |
+
"grad_norm": 9.297645568847656,
|
| 91 |
+
"learning_rate": 5.98e-06,
|
| 92 |
+
"loss": 1.3343,
|
| 93 |
+
"step": 300
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.2076014053018205,
|
| 97 |
+
"grad_norm": 10.026762962341309,
|
| 98 |
+
"learning_rate": 6.480000000000001e-06,
|
| 99 |
+
"loss": 1.3353,
|
| 100 |
+
"step": 325
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.2235707441711913,
|
| 104 |
+
"grad_norm": 9.392842292785645,
|
| 105 |
+
"learning_rate": 6.98e-06,
|
| 106 |
+
"loss": 1.3451,
|
| 107 |
+
"step": 350
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.23954008304056212,
|
| 111 |
+
"grad_norm": 7.376865386962891,
|
| 112 |
+
"learning_rate": 7.48e-06,
|
| 113 |
+
"loss": 1.2645,
|
| 114 |
+
"step": 375
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.2555094219099329,
|
| 118 |
+
"grad_norm": 10.001627922058105,
|
| 119 |
+
"learning_rate": 7.980000000000002e-06,
|
| 120 |
+
"loss": 1.271,
|
| 121 |
+
"step": 400
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.2714787607793037,
|
| 125 |
+
"grad_norm": 7.699875354766846,
|
| 126 |
+
"learning_rate": 8.48e-06,
|
| 127 |
+
"loss": 1.2116,
|
| 128 |
+
"step": 425
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.28744809964867457,
|
| 132 |
+
"grad_norm": 8.733208656311035,
|
| 133 |
+
"learning_rate": 8.98e-06,
|
| 134 |
+
"loss": 1.2222,
|
| 135 |
+
"step": 450
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.30341743851804537,
|
| 139 |
+
"grad_norm": 8.286600112915039,
|
| 140 |
+
"learning_rate": 9.48e-06,
|
| 141 |
+
"loss": 1.2134,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.31938677738741617,
|
| 146 |
+
"grad_norm": 7.408142566680908,
|
| 147 |
+
"learning_rate": 9.980000000000001e-06,
|
| 148 |
+
"loss": 1.1662,
|
| 149 |
+
"step": 500
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.33535611625678696,
|
| 153 |
+
"grad_norm": 9.274364471435547,
|
| 154 |
+
"learning_rate": 9.946666666666667e-06,
|
| 155 |
+
"loss": 1.2752,
|
| 156 |
+
"step": 525
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.35132545512615776,
|
| 160 |
+
"grad_norm": 8.4090576171875,
|
| 161 |
+
"learning_rate": 9.891111111111113e-06,
|
| 162 |
+
"loss": 1.2137,
|
| 163 |
+
"step": 550
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.3672947939955286,
|
| 167 |
+
"grad_norm": 8.74751091003418,
|
| 168 |
+
"learning_rate": 9.835555555555556e-06,
|
| 169 |
+
"loss": 1.1794,
|
| 170 |
+
"step": 575
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.3832641328648994,
|
| 174 |
+
"grad_norm": 7.037300109863281,
|
| 175 |
+
"learning_rate": 9.780000000000001e-06,
|
| 176 |
+
"loss": 1.1806,
|
| 177 |
+
"step": 600
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.3992334717342702,
|
| 181 |
+
"grad_norm": 8.758708953857422,
|
| 182 |
+
"learning_rate": 9.724444444444445e-06,
|
| 183 |
+
"loss": 1.0997,
|
| 184 |
+
"step": 625
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.415202810603641,
|
| 188 |
+
"grad_norm": 7.583318710327148,
|
| 189 |
+
"learning_rate": 9.66888888888889e-06,
|
| 190 |
+
"loss": 1.1122,
|
| 191 |
+
"step": 650
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.4311721494730118,
|
| 195 |
+
"grad_norm": 7.173635959625244,
|
| 196 |
+
"learning_rate": 9.613333333333335e-06,
|
| 197 |
+
"loss": 1.1167,
|
| 198 |
+
"step": 675
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.4471414883423826,
|
| 202 |
+
"grad_norm": 6.702845096588135,
|
| 203 |
+
"learning_rate": 9.557777777777777e-06,
|
| 204 |
+
"loss": 1.0873,
|
| 205 |
+
"step": 700
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.46311082721175345,
|
| 209 |
+
"grad_norm": 7.966395854949951,
|
| 210 |
+
"learning_rate": 9.502222222222223e-06,
|
| 211 |
+
"loss": 1.0764,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.47908016608112425,
|
| 216 |
+
"grad_norm": 7.768787860870361,
|
| 217 |
+
"learning_rate": 9.446666666666667e-06,
|
| 218 |
+
"loss": 1.0985,
|
| 219 |
+
"step": 750
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.49504950495049505,
|
| 223 |
+
"grad_norm": 8.754780769348145,
|
| 224 |
+
"learning_rate": 9.391111111111111e-06,
|
| 225 |
+
"loss": 1.0646,
|
| 226 |
+
"step": 775
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.5110188438198658,
|
| 230 |
+
"grad_norm": 7.156813144683838,
|
| 231 |
+
"learning_rate": 9.335555555555557e-06,
|
| 232 |
+
"loss": 1.0149,
|
| 233 |
+
"step": 800
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.5269881826892366,
|
| 237 |
+
"grad_norm": 6.864979267120361,
|
| 238 |
+
"learning_rate": 9.280000000000001e-06,
|
| 239 |
+
"loss": 1.0514,
|
| 240 |
+
"step": 825
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.5429575215586074,
|
| 244 |
+
"grad_norm": 6.696143627166748,
|
| 245 |
+
"learning_rate": 9.224444444444445e-06,
|
| 246 |
+
"loss": 1.0376,
|
| 247 |
+
"step": 850
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.5589268604279782,
|
| 251 |
+
"grad_norm": 6.3973469734191895,
|
| 252 |
+
"learning_rate": 9.168888888888889e-06,
|
| 253 |
+
"loss": 0.992,
|
| 254 |
+
"step": 875
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.5748961992973491,
|
| 258 |
+
"grad_norm": 6.48210334777832,
|
| 259 |
+
"learning_rate": 9.113333333333335e-06,
|
| 260 |
+
"loss": 1.0463,
|
| 261 |
+
"step": 900
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.5908655381667199,
|
| 265 |
+
"grad_norm": 7.711927890777588,
|
| 266 |
+
"learning_rate": 9.057777777777779e-06,
|
| 267 |
+
"loss": 1.0023,
|
| 268 |
+
"step": 925
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.6068348770360907,
|
| 272 |
+
"grad_norm": 7.2077250480651855,
|
| 273 |
+
"learning_rate": 9.002222222222223e-06,
|
| 274 |
+
"loss": 1.0283,
|
| 275 |
+
"step": 950
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.6228042159054615,
|
| 279 |
+
"grad_norm": 7.416881561279297,
|
| 280 |
+
"learning_rate": 8.946666666666669e-06,
|
| 281 |
+
"loss": 1.0249,
|
| 282 |
+
"step": 975
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.6387735547748323,
|
| 286 |
+
"grad_norm": 6.398010730743408,
|
| 287 |
+
"learning_rate": 8.891111111111111e-06,
|
| 288 |
+
"loss": 0.9546,
|
| 289 |
+
"step": 1000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.6387735547748323,
|
| 293 |
+
"eval_loss": 0.4855521023273468,
|
| 294 |
+
"eval_runtime": 1667.9567,
|
| 295 |
+
"eval_samples_per_second": 2.339,
|
| 296 |
+
"eval_steps_per_second": 0.146,
|
| 297 |
+
"eval_wer": 0.3615418266660186,
|
| 298 |
+
"step": 1000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.6547428936442031,
|
| 302 |
+
"grad_norm": 5.443120956420898,
|
| 303 |
+
"learning_rate": 8.835555555555557e-06,
|
| 304 |
+
"loss": 1.013,
|
| 305 |
+
"step": 1025
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.6707122325135739,
|
| 309 |
+
"grad_norm": 7.361939907073975,
|
| 310 |
+
"learning_rate": 8.78e-06,
|
| 311 |
+
"loss": 0.946,
|
| 312 |
+
"step": 1050
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.6866815713829447,
|
| 316 |
+
"grad_norm": 6.80987024307251,
|
| 317 |
+
"learning_rate": 8.724444444444445e-06,
|
| 318 |
+
"loss": 0.9945,
|
| 319 |
+
"step": 1075
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.7026509102523155,
|
| 323 |
+
"grad_norm": 6.040420055389404,
|
| 324 |
+
"learning_rate": 8.66888888888889e-06,
|
| 325 |
+
"loss": 1.0078,
|
| 326 |
+
"step": 1100
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.7186202491216863,
|
| 330 |
+
"grad_norm": 7.904597282409668,
|
| 331 |
+
"learning_rate": 8.613333333333333e-06,
|
| 332 |
+
"loss": 0.9707,
|
| 333 |
+
"step": 1125
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.7345895879910572,
|
| 337 |
+
"grad_norm": 6.9557204246521,
|
| 338 |
+
"learning_rate": 8.557777777777778e-06,
|
| 339 |
+
"loss": 0.9712,
|
| 340 |
+
"step": 1150
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.750558926860428,
|
| 344 |
+
"grad_norm": 5.4510178565979,
|
| 345 |
+
"learning_rate": 8.502222222222223e-06,
|
| 346 |
+
"loss": 0.9659,
|
| 347 |
+
"step": 1175
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 0.7665282657297988,
|
| 351 |
+
"grad_norm": 5.72000789642334,
|
| 352 |
+
"learning_rate": 8.446666666666668e-06,
|
| 353 |
+
"loss": 0.9368,
|
| 354 |
+
"step": 1200
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.7824976045991696,
|
| 358 |
+
"grad_norm": 5.816184043884277,
|
| 359 |
+
"learning_rate": 8.391111111111112e-06,
|
| 360 |
+
"loss": 0.9089,
|
| 361 |
+
"step": 1225
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 0.7984669434685404,
|
| 365 |
+
"grad_norm": 6.033445835113525,
|
| 366 |
+
"learning_rate": 8.335555555555556e-06,
|
| 367 |
+
"loss": 0.9502,
|
| 368 |
+
"step": 1250
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 0.8144362823379112,
|
| 372 |
+
"grad_norm": 5.273908615112305,
|
| 373 |
+
"learning_rate": 8.28e-06,
|
| 374 |
+
"loss": 0.9151,
|
| 375 |
+
"step": 1275
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.830405621207282,
|
| 379 |
+
"grad_norm": 6.122651100158691,
|
| 380 |
+
"learning_rate": 8.224444444444444e-06,
|
| 381 |
+
"loss": 0.9517,
|
| 382 |
+
"step": 1300
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 0.8463749600766528,
|
| 386 |
+
"grad_norm": 6.673498630523682,
|
| 387 |
+
"learning_rate": 8.16888888888889e-06,
|
| 388 |
+
"loss": 1.0318,
|
| 389 |
+
"step": 1325
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 0.8623442989460236,
|
| 393 |
+
"grad_norm": 7.197372913360596,
|
| 394 |
+
"learning_rate": 8.113333333333334e-06,
|
| 395 |
+
"loss": 0.937,
|
| 396 |
+
"step": 1350
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 0.8783136378153944,
|
| 400 |
+
"grad_norm": 6.718713283538818,
|
| 401 |
+
"learning_rate": 8.057777777777778e-06,
|
| 402 |
+
"loss": 0.9325,
|
| 403 |
+
"step": 1375
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.8942829766847652,
|
| 407 |
+
"grad_norm": 7.090008735656738,
|
| 408 |
+
"learning_rate": 8.002222222222222e-06,
|
| 409 |
+
"loss": 0.9044,
|
| 410 |
+
"step": 1400
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.9102523155541361,
|
| 414 |
+
"grad_norm": 8.890588760375977,
|
| 415 |
+
"learning_rate": 7.946666666666666e-06,
|
| 416 |
+
"loss": 0.8935,
|
| 417 |
+
"step": 1425
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.9262216544235069,
|
| 421 |
+
"grad_norm": 5.521778583526611,
|
| 422 |
+
"learning_rate": 7.891111111111112e-06,
|
| 423 |
+
"loss": 0.8857,
|
| 424 |
+
"step": 1450
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 0.9421909932928777,
|
| 428 |
+
"grad_norm": 6.829771041870117,
|
| 429 |
+
"learning_rate": 7.835555555555556e-06,
|
| 430 |
+
"loss": 0.9411,
|
| 431 |
+
"step": 1475
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.9581603321622485,
|
| 435 |
+
"grad_norm": 6.48971700668335,
|
| 436 |
+
"learning_rate": 7.78e-06,
|
| 437 |
+
"loss": 0.8616,
|
| 438 |
+
"step": 1500
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 0.9741296710316193,
|
| 442 |
+
"grad_norm": 5.809508323669434,
|
| 443 |
+
"learning_rate": 7.724444444444446e-06,
|
| 444 |
+
"loss": 0.8976,
|
| 445 |
+
"step": 1525
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 0.9900990099009901,
|
| 449 |
+
"grad_norm": 6.918030738830566,
|
| 450 |
+
"learning_rate": 7.66888888888889e-06,
|
| 451 |
+
"loss": 0.9213,
|
| 452 |
+
"step": 1550
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 1.006068348770361,
|
| 456 |
+
"grad_norm": 5.847545623779297,
|
| 457 |
+
"learning_rate": 7.613333333333334e-06,
|
| 458 |
+
"loss": 0.813,
|
| 459 |
+
"step": 1575
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 1.0220376876397317,
|
| 463 |
+
"grad_norm": 5.331517219543457,
|
| 464 |
+
"learning_rate": 7.557777777777779e-06,
|
| 465 |
+
"loss": 0.6469,
|
| 466 |
+
"step": 1600
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 1.0380070265091026,
|
| 470 |
+
"grad_norm": 7.251556396484375,
|
| 471 |
+
"learning_rate": 7.502222222222223e-06,
|
| 472 |
+
"loss": 0.646,
|
| 473 |
+
"step": 1625
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 1.0539763653784733,
|
| 477 |
+
"grad_norm": 5.755091667175293,
|
| 478 |
+
"learning_rate": 7.446666666666668e-06,
|
| 479 |
+
"loss": 0.6496,
|
| 480 |
+
"step": 1650
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 1.0699457042478442,
|
| 484 |
+
"grad_norm": 5.709854602813721,
|
| 485 |
+
"learning_rate": 7.3911111111111125e-06,
|
| 486 |
+
"loss": 0.6459,
|
| 487 |
+
"step": 1675
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 1.0859150431172149,
|
| 491 |
+
"grad_norm": 5.866008281707764,
|
| 492 |
+
"learning_rate": 7.335555555555556e-06,
|
| 493 |
+
"loss": 0.6629,
|
| 494 |
+
"step": 1700
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 1.1018843819865858,
|
| 498 |
+
"grad_norm": 5.575136184692383,
|
| 499 |
+
"learning_rate": 7.280000000000001e-06,
|
| 500 |
+
"loss": 0.6724,
|
| 501 |
+
"step": 1725
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 1.1178537208559565,
|
| 505 |
+
"grad_norm": 5.722553730010986,
|
| 506 |
+
"learning_rate": 7.224444444444445e-06,
|
| 507 |
+
"loss": 0.6946,
|
| 508 |
+
"step": 1750
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 1.1338230597253274,
|
| 512 |
+
"grad_norm": 5.85053825378418,
|
| 513 |
+
"learning_rate": 7.1688888888888895e-06,
|
| 514 |
+
"loss": 0.6676,
|
| 515 |
+
"step": 1775
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 1.1497923985946983,
|
| 519 |
+
"grad_norm": 6.432461261749268,
|
| 520 |
+
"learning_rate": 7.113333333333334e-06,
|
| 521 |
+
"loss": 0.6381,
|
| 522 |
+
"step": 1800
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 1.165761737464069,
|
| 526 |
+
"grad_norm": 6.030885696411133,
|
| 527 |
+
"learning_rate": 7.057777777777778e-06,
|
| 528 |
+
"loss": 0.683,
|
| 529 |
+
"step": 1825
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 1.1817310763334399,
|
| 533 |
+
"grad_norm": 5.215100288391113,
|
| 534 |
+
"learning_rate": 7.0022222222222225e-06,
|
| 535 |
+
"loss": 0.6623,
|
| 536 |
+
"step": 1850
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 1.1977004152028106,
|
| 540 |
+
"grad_norm": 5.460089206695557,
|
| 541 |
+
"learning_rate": 6.946666666666667e-06,
|
| 542 |
+
"loss": 0.6519,
|
| 543 |
+
"step": 1875
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 1.2136697540721815,
|
| 547 |
+
"grad_norm": 5.882613182067871,
|
| 548 |
+
"learning_rate": 6.891111111111111e-06,
|
| 549 |
+
"loss": 0.6788,
|
| 550 |
+
"step": 1900
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 1.2296390929415522,
|
| 554 |
+
"grad_norm": 5.0330915451049805,
|
| 555 |
+
"learning_rate": 6.835555555555556e-06,
|
| 556 |
+
"loss": 0.6654,
|
| 557 |
+
"step": 1925
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.245608431810923,
|
| 561 |
+
"grad_norm": 4.624293804168701,
|
| 562 |
+
"learning_rate": 6.780000000000001e-06,
|
| 563 |
+
"loss": 0.6394,
|
| 564 |
+
"step": 1950
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.2615777706802938,
|
| 568 |
+
"grad_norm": 5.252276420593262,
|
| 569 |
+
"learning_rate": 6.724444444444444e-06,
|
| 570 |
+
"loss": 0.6522,
|
| 571 |
+
"step": 1975
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.2775471095496647,
|
| 575 |
+
"grad_norm": 5.393728733062744,
|
| 576 |
+
"learning_rate": 6.668888888888889e-06,
|
| 577 |
+
"loss": 0.6673,
|
| 578 |
+
"step": 2000
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.2775471095496647,
|
| 582 |
+
"eval_loss": 0.41851192712783813,
|
| 583 |
+
"eval_runtime": 1656.9911,
|
| 584 |
+
"eval_samples_per_second": 2.354,
|
| 585 |
+
"eval_steps_per_second": 0.147,
|
| 586 |
+
"eval_wer": 0.31497593933796725,
|
| 587 |
+
"step": 2000
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 1.2935164484190356,
|
| 591 |
+
"grad_norm": 7.740047454833984,
|
| 592 |
+
"learning_rate": 6.613333333333334e-06,
|
| 593 |
+
"loss": 0.643,
|
| 594 |
+
"step": 2025
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 1.3094857872884063,
|
| 598 |
+
"grad_norm": 4.754881381988525,
|
| 599 |
+
"learning_rate": 6.557777777777778e-06,
|
| 600 |
+
"loss": 0.6545,
|
| 601 |
+
"step": 2050
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 1.325455126157777,
|
| 605 |
+
"grad_norm": 5.5493693351745605,
|
| 606 |
+
"learning_rate": 6.502222222222223e-06,
|
| 607 |
+
"loss": 0.6741,
|
| 608 |
+
"step": 2075
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 1.3414244650271478,
|
| 612 |
+
"grad_norm": 5.743963241577148,
|
| 613 |
+
"learning_rate": 6.446666666666668e-06,
|
| 614 |
+
"loss": 0.65,
|
| 615 |
+
"step": 2100
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 1.3573938038965188,
|
| 619 |
+
"grad_norm": 6.479578495025635,
|
| 620 |
+
"learning_rate": 6.391111111111111e-06,
|
| 621 |
+
"loss": 0.6345,
|
| 622 |
+
"step": 2125
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 1.3733631427658894,
|
| 626 |
+
"grad_norm": 5.8687286376953125,
|
| 627 |
+
"learning_rate": 6.335555555555556e-06,
|
| 628 |
+
"loss": 0.6612,
|
| 629 |
+
"step": 2150
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 1.3893324816352604,
|
| 633 |
+
"grad_norm": 4.9471893310546875,
|
| 634 |
+
"learning_rate": 6.280000000000001e-06,
|
| 635 |
+
"loss": 0.6616,
|
| 636 |
+
"step": 2175
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 1.405301820504631,
|
| 640 |
+
"grad_norm": 4.968109130859375,
|
| 641 |
+
"learning_rate": 6.224444444444445e-06,
|
| 642 |
+
"loss": 0.6405,
|
| 643 |
+
"step": 2200
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 1.421271159374002,
|
| 647 |
+
"grad_norm": 6.063320159912109,
|
| 648 |
+
"learning_rate": 6.16888888888889e-06,
|
| 649 |
+
"loss": 0.6615,
|
| 650 |
+
"step": 2225
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 1.4372404982433729,
|
| 654 |
+
"grad_norm": 5.78763484954834,
|
| 655 |
+
"learning_rate": 6.113333333333333e-06,
|
| 656 |
+
"loss": 0.6308,
|
| 657 |
+
"step": 2250
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.4532098371127435,
|
| 661 |
+
"grad_norm": 5.714231014251709,
|
| 662 |
+
"learning_rate": 6.057777777777778e-06,
|
| 663 |
+
"loss": 0.6619,
|
| 664 |
+
"step": 2275
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 1.4691791759821142,
|
| 668 |
+
"grad_norm": 7.075171947479248,
|
| 669 |
+
"learning_rate": 6.002222222222223e-06,
|
| 670 |
+
"loss": 0.6748,
|
| 671 |
+
"step": 2300
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 1.4851485148514851,
|
| 675 |
+
"grad_norm": 6.3147969245910645,
|
| 676 |
+
"learning_rate": 5.946666666666668e-06,
|
| 677 |
+
"loss": 0.6441,
|
| 678 |
+
"step": 2325
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 1.501117853720856,
|
| 682 |
+
"grad_norm": 6.079730033874512,
|
| 683 |
+
"learning_rate": 5.891111111111112e-06,
|
| 684 |
+
"loss": 0.6303,
|
| 685 |
+
"step": 2350
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 1.5170871925902267,
|
| 689 |
+
"grad_norm": 5.132657527923584,
|
| 690 |
+
"learning_rate": 5.8355555555555565e-06,
|
| 691 |
+
"loss": 0.6718,
|
| 692 |
+
"step": 2375
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 1.5330565314595974,
|
| 696 |
+
"grad_norm": 5.566254138946533,
|
| 697 |
+
"learning_rate": 5.78e-06,
|
| 698 |
+
"loss": 0.6686,
|
| 699 |
+
"step": 2400
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 1.5490258703289683,
|
| 703 |
+
"grad_norm": 6.299161911010742,
|
| 704 |
+
"learning_rate": 5.724444444444445e-06,
|
| 705 |
+
"loss": 0.6485,
|
| 706 |
+
"step": 2425
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 1.5649952091983392,
|
| 710 |
+
"grad_norm": 5.586335182189941,
|
| 711 |
+
"learning_rate": 5.6688888888888895e-06,
|
| 712 |
+
"loss": 0.6031,
|
| 713 |
+
"step": 2450
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 1.5809645480677101,
|
| 717 |
+
"grad_norm": 5.396686553955078,
|
| 718 |
+
"learning_rate": 5.613333333333334e-06,
|
| 719 |
+
"loss": 0.6407,
|
| 720 |
+
"step": 2475
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 1.5969338869370808,
|
| 724 |
+
"grad_norm": 6.483650207519531,
|
| 725 |
+
"learning_rate": 5.557777777777778e-06,
|
| 726 |
+
"loss": 0.6703,
|
| 727 |
+
"step": 2500
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 1.6129032258064515,
|
| 731 |
+
"grad_norm": 5.946717739105225,
|
| 732 |
+
"learning_rate": 5.5022222222222224e-06,
|
| 733 |
+
"loss": 0.6079,
|
| 734 |
+
"step": 2525
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 1.6288725646758224,
|
| 738 |
+
"grad_norm": 5.817110061645508,
|
| 739 |
+
"learning_rate": 5.4466666666666665e-06,
|
| 740 |
+
"loss": 0.6275,
|
| 741 |
+
"step": 2550
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 1.6448419035451933,
|
| 745 |
+
"grad_norm": 5.888713836669922,
|
| 746 |
+
"learning_rate": 5.391111111111111e-06,
|
| 747 |
+
"loss": 0.64,
|
| 748 |
+
"step": 2575
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 1.660811242414564,
|
| 752 |
+
"grad_norm": 5.296940803527832,
|
| 753 |
+
"learning_rate": 5.335555555555556e-06,
|
| 754 |
+
"loss": 0.6278,
|
| 755 |
+
"step": 2600
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 1.6767805812839347,
|
| 759 |
+
"grad_norm": 5.4542388916015625,
|
| 760 |
+
"learning_rate": 5.28e-06,
|
| 761 |
+
"loss": 0.6236,
|
| 762 |
+
"step": 2625
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 1.6927499201533056,
|
| 766 |
+
"grad_norm": 5.980891227722168,
|
| 767 |
+
"learning_rate": 5.224444444444445e-06,
|
| 768 |
+
"loss": 0.6607,
|
| 769 |
+
"step": 2650
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 1.7087192590226765,
|
| 773 |
+
"grad_norm": 7.3297014236450195,
|
| 774 |
+
"learning_rate": 5.168888888888889e-06,
|
| 775 |
+
"loss": 0.673,
|
| 776 |
+
"step": 2675
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 1.7246885978920474,
|
| 780 |
+
"grad_norm": 4.739117622375488,
|
| 781 |
+
"learning_rate": 5.113333333333333e-06,
|
| 782 |
+
"loss": 0.5893,
|
| 783 |
+
"step": 2700
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 1.7406579367614181,
|
| 787 |
+
"grad_norm": 4.754851818084717,
|
| 788 |
+
"learning_rate": 5.057777777777778e-06,
|
| 789 |
+
"loss": 0.6459,
|
| 790 |
+
"step": 2725
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 1.7566272756307888,
|
| 794 |
+
"grad_norm": 4.72427225112915,
|
| 795 |
+
"learning_rate": 5.002222222222223e-06,
|
| 796 |
+
"loss": 0.6093,
|
| 797 |
+
"step": 2750
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 1.7725966145001597,
|
| 801 |
+
"grad_norm": 5.871730327606201,
|
| 802 |
+
"learning_rate": 4.946666666666667e-06,
|
| 803 |
+
"loss": 0.6088,
|
| 804 |
+
"step": 2775
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 1.7885659533695306,
|
| 808 |
+
"grad_norm": 5.640221118927002,
|
| 809 |
+
"learning_rate": 4.891111111111111e-06,
|
| 810 |
+
"loss": 0.6214,
|
| 811 |
+
"step": 2800
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 1.8045352922389013,
|
| 815 |
+
"grad_norm": 5.749293327331543,
|
| 816 |
+
"learning_rate": 4.835555555555556e-06,
|
| 817 |
+
"loss": 0.6105,
|
| 818 |
+
"step": 2825
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 1.820504631108272,
|
| 822 |
+
"grad_norm": 6.414817810058594,
|
| 823 |
+
"learning_rate": 4.78e-06,
|
| 824 |
+
"loss": 0.6523,
|
| 825 |
+
"step": 2850
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 1.836473969977643,
|
| 829 |
+
"grad_norm": 5.655066967010498,
|
| 830 |
+
"learning_rate": 4.724444444444445e-06,
|
| 831 |
+
"loss": 0.6348,
|
| 832 |
+
"step": 2875
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 1.8524433088470138,
|
| 836 |
+
"grad_norm": 5.395172119140625,
|
| 837 |
+
"learning_rate": 4.66888888888889e-06,
|
| 838 |
+
"loss": 0.634,
|
| 839 |
+
"step": 2900
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 1.8684126477163845,
|
| 843 |
+
"grad_norm": 4.894425868988037,
|
| 844 |
+
"learning_rate": 4.613333333333334e-06,
|
| 845 |
+
"loss": 0.595,
|
| 846 |
+
"step": 2925
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 1.8843819865857554,
|
| 850 |
+
"grad_norm": 5.686845779418945,
|
| 851 |
+
"learning_rate": 4.557777777777778e-06,
|
| 852 |
+
"loss": 0.6113,
|
| 853 |
+
"step": 2950
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 1.900351325455126,
|
| 857 |
+
"grad_norm": 5.444372177124023,
|
| 858 |
+
"learning_rate": 4.502222222222223e-06,
|
| 859 |
+
"loss": 0.6277,
|
| 860 |
+
"step": 2975
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 1.916320664324497,
|
| 864 |
+
"grad_norm": 5.224650859832764,
|
| 865 |
+
"learning_rate": 4.446666666666667e-06,
|
| 866 |
+
"loss": 0.6601,
|
| 867 |
+
"step": 3000
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 1.916320664324497,
|
| 871 |
+
"eval_loss": 0.389792263507843,
|
| 872 |
+
"eval_runtime": 1666.4367,
|
| 873 |
+
"eval_samples_per_second": 2.341,
|
| 874 |
+
"eval_steps_per_second": 0.146,
|
| 875 |
+
"eval_wer": 0.2912798327905507,
|
| 876 |
+
"step": 3000
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 1.932290003193868,
|
| 880 |
+
"grad_norm": 4.736299514770508,
|
| 881 |
+
"learning_rate": 4.391111111111112e-06,
|
| 882 |
+
"loss": 0.6351,
|
| 883 |
+
"step": 3025
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 1.9482593420632386,
|
| 887 |
+
"grad_norm": 4.797982692718506,
|
| 888 |
+
"learning_rate": 4.3355555555555565e-06,
|
| 889 |
+
"loss": 0.6017,
|
| 890 |
+
"step": 3050
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 1.9642286809326093,
|
| 894 |
+
"grad_norm": 5.4142022132873535,
|
| 895 |
+
"learning_rate": 4.2800000000000005e-06,
|
| 896 |
+
"loss": 0.6124,
|
| 897 |
+
"step": 3075
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 1.9801980198019802,
|
| 901 |
+
"grad_norm": 5.374582767486572,
|
| 902 |
+
"learning_rate": 4.2244444444444446e-06,
|
| 903 |
+
"loss": 0.6243,
|
| 904 |
+
"step": 3100
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 1.996167358671351,
|
| 908 |
+
"grad_norm": 4.480958938598633,
|
| 909 |
+
"learning_rate": 4.168888888888889e-06,
|
| 910 |
+
"loss": 0.5981,
|
| 911 |
+
"step": 3125
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 2.012136697540722,
|
| 915 |
+
"grad_norm": 5.1775736808776855,
|
| 916 |
+
"learning_rate": 4.1133333333333335e-06,
|
| 917 |
+
"loss": 0.4607,
|
| 918 |
+
"step": 3150
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 2.0281060364100925,
|
| 922 |
+
"grad_norm": 4.18350887298584,
|
| 923 |
+
"learning_rate": 4.057777777777778e-06,
|
| 924 |
+
"loss": 0.3907,
|
| 925 |
+
"step": 3175
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 2.0440753752794634,
|
| 929 |
+
"grad_norm": 4.902673721313477,
|
| 930 |
+
"learning_rate": 4.002222222222222e-06,
|
| 931 |
+
"loss": 0.4004,
|
| 932 |
+
"step": 3200
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 2.0600447141488343,
|
| 936 |
+
"grad_norm": 4.528185844421387,
|
| 937 |
+
"learning_rate": 3.946666666666667e-06,
|
| 938 |
+
"loss": 0.3936,
|
| 939 |
+
"step": 3225
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 2.076014053018205,
|
| 943 |
+
"grad_norm": 5.0421061515808105,
|
| 944 |
+
"learning_rate": 3.891111111111111e-06,
|
| 945 |
+
"loss": 0.4242,
|
| 946 |
+
"step": 3250
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 2.0919833918875756,
|
| 950 |
+
"grad_norm": 4.861361026763916,
|
| 951 |
+
"learning_rate": 3.835555555555555e-06,
|
| 952 |
+
"loss": 0.3916,
|
| 953 |
+
"step": 3275
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 2.1079527307569466,
|
| 957 |
+
"grad_norm": 6.367924213409424,
|
| 958 |
+
"learning_rate": 3.7800000000000002e-06,
|
| 959 |
+
"loss": 0.413,
|
| 960 |
+
"step": 3300
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 2.1239220696263175,
|
| 964 |
+
"grad_norm": 3.6156017780303955,
|
| 965 |
+
"learning_rate": 3.724444444444445e-06,
|
| 966 |
+
"loss": 0.4103,
|
| 967 |
+
"step": 3325
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 2.1398914084956884,
|
| 971 |
+
"grad_norm": 5.132371425628662,
|
| 972 |
+
"learning_rate": 3.668888888888889e-06,
|
| 973 |
+
"loss": 0.407,
|
| 974 |
+
"step": 3350
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 2.1558607473650593,
|
| 978 |
+
"grad_norm": 4.446967601776123,
|
| 979 |
+
"learning_rate": 3.6133333333333336e-06,
|
| 980 |
+
"loss": 0.4082,
|
| 981 |
+
"step": 3375
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 2.1718300862344297,
|
| 985 |
+
"grad_norm": 5.18086576461792,
|
| 986 |
+
"learning_rate": 3.5577777777777785e-06,
|
| 987 |
+
"loss": 0.4053,
|
| 988 |
+
"step": 3400
|
| 989 |
+
},
|
| 990 |
+
{
|
| 991 |
+
"epoch": 2.1877994251038007,
|
| 992 |
+
"grad_norm": 4.955840587615967,
|
| 993 |
+
"learning_rate": 3.5022222222222225e-06,
|
| 994 |
+
"loss": 0.3995,
|
| 995 |
+
"step": 3425
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 2.2037687639731716,
|
| 999 |
+
"grad_norm": 4.59506368637085,
|
| 1000 |
+
"learning_rate": 3.446666666666667e-06,
|
| 1001 |
+
"loss": 0.4064,
|
| 1002 |
+
"step": 3450
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 2.2197381028425425,
|
| 1006 |
+
"grad_norm": 5.527169227600098,
|
| 1007 |
+
"learning_rate": 3.391111111111111e-06,
|
| 1008 |
+
"loss": 0.3946,
|
| 1009 |
+
"step": 3475
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 2.235707441711913,
|
| 1013 |
+
"grad_norm": 5.034458637237549,
|
| 1014 |
+
"learning_rate": 3.335555555555556e-06,
|
| 1015 |
+
"loss": 0.4088,
|
| 1016 |
+
"step": 3500
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 2.251676780581284,
|
| 1020 |
+
"grad_norm": 4.814046382904053,
|
| 1021 |
+
"learning_rate": 3.2800000000000004e-06,
|
| 1022 |
+
"loss": 0.4032,
|
| 1023 |
+
"step": 3525
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 2.2676461194506548,
|
| 1027 |
+
"grad_norm": 5.115776062011719,
|
| 1028 |
+
"learning_rate": 3.2244444444444444e-06,
|
| 1029 |
+
"loss": 0.4009,
|
| 1030 |
+
"step": 3550
|
| 1031 |
+
},
|
| 1032 |
+
{
|
| 1033 |
+
"epoch": 2.2836154583200257,
|
| 1034 |
+
"grad_norm": 4.4629130363464355,
|
| 1035 |
+
"learning_rate": 3.1688888888888893e-06,
|
| 1036 |
+
"loss": 0.408,
|
| 1037 |
+
"step": 3575
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 2.2995847971893966,
|
| 1041 |
+
"grad_norm": 4.914193153381348,
|
| 1042 |
+
"learning_rate": 3.1133333333333337e-06,
|
| 1043 |
+
"loss": 0.3867,
|
| 1044 |
+
"step": 3600
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 2.315554136058767,
|
| 1048 |
+
"grad_norm": 4.358288288116455,
|
| 1049 |
+
"learning_rate": 3.0577777777777778e-06,
|
| 1050 |
+
"loss": 0.4485,
|
| 1051 |
+
"step": 3625
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 2.331523474928138,
|
| 1055 |
+
"grad_norm": 4.8634161949157715,
|
| 1056 |
+
"learning_rate": 3.0022222222222227e-06,
|
| 1057 |
+
"loss": 0.389,
|
| 1058 |
+
"step": 3650
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 2.347492813797509,
|
| 1062 |
+
"grad_norm": 5.283102989196777,
|
| 1063 |
+
"learning_rate": 2.946666666666667e-06,
|
| 1064 |
+
"loss": 0.395,
|
| 1065 |
+
"step": 3675
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 2.3634621526668798,
|
| 1069 |
+
"grad_norm": 4.393892765045166,
|
| 1070 |
+
"learning_rate": 2.891111111111111e-06,
|
| 1071 |
+
"loss": 0.4047,
|
| 1072 |
+
"step": 3700
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"epoch": 2.37943149153625,
|
| 1076 |
+
"grad_norm": 5.246931552886963,
|
| 1077 |
+
"learning_rate": 2.835555555555556e-06,
|
| 1078 |
+
"loss": 0.3957,
|
| 1079 |
+
"step": 3725
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 2.395400830405621,
|
| 1083 |
+
"grad_norm": 4.599190711975098,
|
| 1084 |
+
"learning_rate": 2.7800000000000005e-06,
|
| 1085 |
+
"loss": 0.3829,
|
| 1086 |
+
"step": 3750
|
| 1087 |
+
},
|
| 1088 |
+
{
|
| 1089 |
+
"epoch": 2.411370169274992,
|
| 1090 |
+
"grad_norm": 5.136366367340088,
|
| 1091 |
+
"learning_rate": 2.7244444444444445e-06,
|
| 1092 |
+
"loss": 0.4029,
|
| 1093 |
+
"step": 3775
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"epoch": 2.427339508144363,
|
| 1097 |
+
"grad_norm": 4.448164463043213,
|
| 1098 |
+
"learning_rate": 2.6688888888888894e-06,
|
| 1099 |
+
"loss": 0.3869,
|
| 1100 |
+
"step": 3800
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 2.443308847013734,
|
| 1104 |
+
"grad_norm": 5.434977054595947,
|
| 1105 |
+
"learning_rate": 2.6133333333333334e-06,
|
| 1106 |
+
"loss": 0.3871,
|
| 1107 |
+
"step": 3825
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 2.4592781858831043,
|
| 1111 |
+
"grad_norm": 5.272222518920898,
|
| 1112 |
+
"learning_rate": 2.557777777777778e-06,
|
| 1113 |
+
"loss": 0.388,
|
| 1114 |
+
"step": 3850
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 2.4752475247524752,
|
| 1118 |
+
"grad_norm": 5.163560390472412,
|
| 1119 |
+
"learning_rate": 2.5022222222222224e-06,
|
| 1120 |
+
"loss": 0.4156,
|
| 1121 |
+
"step": 3875
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 2.491216863621846,
|
| 1125 |
+
"grad_norm": 4.091150283813477,
|
| 1126 |
+
"learning_rate": 2.446666666666667e-06,
|
| 1127 |
+
"loss": 0.4093,
|
| 1128 |
+
"step": 3900
|
| 1129 |
+
},
|
| 1130 |
+
{
|
| 1131 |
+
"epoch": 2.5071862024912166,
|
| 1132 |
+
"grad_norm": 4.858378887176514,
|
| 1133 |
+
"learning_rate": 2.3911111111111113e-06,
|
| 1134 |
+
"loss": 0.4068,
|
| 1135 |
+
"step": 3925
|
| 1136 |
+
},
|
| 1137 |
+
{
|
| 1138 |
+
"epoch": 2.5231555413605875,
|
| 1139 |
+
"grad_norm": 5.535120487213135,
|
| 1140 |
+
"learning_rate": 2.3355555555555557e-06,
|
| 1141 |
+
"loss": 0.3992,
|
| 1142 |
+
"step": 3950
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 2.5391248802299584,
|
| 1146 |
+
"grad_norm": 4.70937442779541,
|
| 1147 |
+
"learning_rate": 2.28e-06,
|
| 1148 |
+
"loss": 0.4009,
|
| 1149 |
+
"step": 3975
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 2.5550942190993293,
|
| 1153 |
+
"grad_norm": 5.273413181304932,
|
| 1154 |
+
"learning_rate": 2.2244444444444447e-06,
|
| 1155 |
+
"loss": 0.3887,
|
| 1156 |
+
"step": 4000
|
| 1157 |
+
},
|
| 1158 |
+
{
|
| 1159 |
+
"epoch": 2.5550942190993293,
|
| 1160 |
+
"eval_loss": 0.3953465223312378,
|
| 1161 |
+
"eval_runtime": 1634.3629,
|
| 1162 |
+
"eval_samples_per_second": 2.387,
|
| 1163 |
+
"eval_steps_per_second": 0.149,
|
| 1164 |
+
"eval_wer": 0.2815826568803772,
|
| 1165 |
+
"step": 4000
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 2.5710635579687002,
|
| 1169 |
+
"grad_norm": 5.2189483642578125,
|
| 1170 |
+
"learning_rate": 2.168888888888889e-06,
|
| 1171 |
+
"loss": 0.3789,
|
| 1172 |
+
"step": 4025
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 2.587032896838071,
|
| 1176 |
+
"grad_norm": 5.335792064666748,
|
| 1177 |
+
"learning_rate": 2.1133333333333336e-06,
|
| 1178 |
+
"loss": 0.3911,
|
| 1179 |
+
"step": 4050
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 2.6030022357074416,
|
| 1183 |
+
"grad_norm": 4.469089508056641,
|
| 1184 |
+
"learning_rate": 2.057777777777778e-06,
|
| 1185 |
+
"loss": 0.388,
|
| 1186 |
+
"step": 4075
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 2.6189715745768125,
|
| 1190 |
+
"grad_norm": 4.454526424407959,
|
| 1191 |
+
"learning_rate": 2.0022222222222225e-06,
|
| 1192 |
+
"loss": 0.3829,
|
| 1193 |
+
"step": 4100
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 2.6349409134461834,
|
| 1197 |
+
"grad_norm": 5.536740303039551,
|
| 1198 |
+
"learning_rate": 1.9466666666666665e-06,
|
| 1199 |
+
"loss": 0.3846,
|
| 1200 |
+
"step": 4125
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 2.650910252315554,
|
| 1204 |
+
"grad_norm": 4.877845764160156,
|
| 1205 |
+
"learning_rate": 1.8911111111111114e-06,
|
| 1206 |
+
"loss": 0.4116,
|
| 1207 |
+
"step": 4150
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 2.666879591184925,
|
| 1211 |
+
"grad_norm": 5.3322014808654785,
|
| 1212 |
+
"learning_rate": 1.8355555555555557e-06,
|
| 1213 |
+
"loss": 0.3857,
|
| 1214 |
+
"step": 4175
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 2.6828489300542957,
|
| 1218 |
+
"grad_norm": 4.9132866859436035,
|
| 1219 |
+
"learning_rate": 1.7800000000000001e-06,
|
| 1220 |
+
"loss": 0.3922,
|
| 1221 |
+
"step": 4200
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 2.6988182689236666,
|
| 1225 |
+
"grad_norm": 5.380650043487549,
|
| 1226 |
+
"learning_rate": 1.7244444444444448e-06,
|
| 1227 |
+
"loss": 0.3757,
|
| 1228 |
+
"step": 4225
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 2.7147876077930375,
|
| 1232 |
+
"grad_norm": 4.5387282371521,
|
| 1233 |
+
"learning_rate": 1.668888888888889e-06,
|
| 1234 |
+
"loss": 0.3919,
|
| 1235 |
+
"step": 4250
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 2.7307569466624084,
|
| 1239 |
+
"grad_norm": 5.215519428253174,
|
| 1240 |
+
"learning_rate": 1.6133333333333335e-06,
|
| 1241 |
+
"loss": 0.4104,
|
| 1242 |
+
"step": 4275
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 2.746726285531779,
|
| 1246 |
+
"grad_norm": 6.126536846160889,
|
| 1247 |
+
"learning_rate": 1.5577777777777777e-06,
|
| 1248 |
+
"loss": 0.3792,
|
| 1249 |
+
"step": 4300
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 2.76269562440115,
|
| 1253 |
+
"grad_norm": 5.707647323608398,
|
| 1254 |
+
"learning_rate": 1.5022222222222224e-06,
|
| 1255 |
+
"loss": 0.4041,
|
| 1256 |
+
"step": 4325
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 2.7786649632705207,
|
| 1260 |
+
"grad_norm": 5.024244785308838,
|
| 1261 |
+
"learning_rate": 1.4466666666666669e-06,
|
| 1262 |
+
"loss": 0.4045,
|
| 1263 |
+
"step": 4350
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 2.794634302139891,
|
| 1267 |
+
"grad_norm": 5.240878105163574,
|
| 1268 |
+
"learning_rate": 1.3911111111111111e-06,
|
| 1269 |
+
"loss": 0.4022,
|
| 1270 |
+
"step": 4375
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 2.810603641009262,
|
| 1274 |
+
"grad_norm": 5.824533939361572,
|
| 1275 |
+
"learning_rate": 1.3355555555555558e-06,
|
| 1276 |
+
"loss": 0.3947,
|
| 1277 |
+
"step": 4400
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 2.826572979878633,
|
| 1281 |
+
"grad_norm": 4.228843688964844,
|
| 1282 |
+
"learning_rate": 1.28e-06,
|
| 1283 |
+
"loss": 0.3725,
|
| 1284 |
+
"step": 4425
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 2.842542318748004,
|
| 1288 |
+
"grad_norm": 4.704908847808838,
|
| 1289 |
+
"learning_rate": 1.2244444444444445e-06,
|
| 1290 |
+
"loss": 0.4001,
|
| 1291 |
+
"step": 4450
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 2.858511657617375,
|
| 1295 |
+
"grad_norm": 4.397124767303467,
|
| 1296 |
+
"learning_rate": 1.168888888888889e-06,
|
| 1297 |
+
"loss": 0.381,
|
| 1298 |
+
"step": 4475
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 2.8744809964867457,
|
| 1302 |
+
"grad_norm": 4.7757062911987305,
|
| 1303 |
+
"learning_rate": 1.1133333333333334e-06,
|
| 1304 |
+
"loss": 0.3689,
|
| 1305 |
+
"step": 4500
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 2.890450335356116,
|
| 1309 |
+
"grad_norm": 4.855478763580322,
|
| 1310 |
+
"learning_rate": 1.0577777777777779e-06,
|
| 1311 |
+
"loss": 0.373,
|
| 1312 |
+
"step": 4525
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 2.906419674225487,
|
| 1316 |
+
"grad_norm": 5.458982467651367,
|
| 1317 |
+
"learning_rate": 1.0022222222222223e-06,
|
| 1318 |
+
"loss": 0.3919,
|
| 1319 |
+
"step": 4550
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 2.922389013094858,
|
| 1323 |
+
"grad_norm": 3.9116828441619873,
|
| 1324 |
+
"learning_rate": 9.466666666666667e-07,
|
| 1325 |
+
"loss": 0.4143,
|
| 1326 |
+
"step": 4575
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 2.9383583519642285,
|
| 1330 |
+
"grad_norm": 4.8182878494262695,
|
| 1331 |
+
"learning_rate": 8.911111111111112e-07,
|
| 1332 |
+
"loss": 0.3726,
|
| 1333 |
+
"step": 4600
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 2.9543276908335994,
|
| 1337 |
+
"grad_norm": 5.018452167510986,
|
| 1338 |
+
"learning_rate": 8.355555555555556e-07,
|
| 1339 |
+
"loss": 0.4018,
|
| 1340 |
+
"step": 4625
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 2.9702970297029703,
|
| 1344 |
+
"grad_norm": 5.315720081329346,
|
| 1345 |
+
"learning_rate": 7.8e-07,
|
| 1346 |
+
"loss": 0.3648,
|
| 1347 |
+
"step": 4650
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 2.986266368572341,
|
| 1351 |
+
"grad_norm": 4.65333890914917,
|
| 1352 |
+
"learning_rate": 7.266666666666668e-07,
|
| 1353 |
+
"loss": 0.3865,
|
| 1354 |
+
"step": 4675
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 3.002235707441712,
|
| 1358 |
+
"grad_norm": 2.9384920597076416,
|
| 1359 |
+
"learning_rate": 6.711111111111111e-07,
|
| 1360 |
+
"loss": 0.3793,
|
| 1361 |
+
"step": 4700
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 3.0182050463110826,
|
| 1365 |
+
"grad_norm": 4.572001934051514,
|
| 1366 |
+
"learning_rate": 6.155555555555556e-07,
|
| 1367 |
+
"loss": 0.2557,
|
| 1368 |
+
"step": 4725
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 3.0341743851804535,
|
| 1372 |
+
"grad_norm": 3.237912654876709,
|
| 1373 |
+
"learning_rate": 5.6e-07,
|
| 1374 |
+
"loss": 0.252,
|
| 1375 |
+
"step": 4750
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 3.0501437240498244,
|
| 1379 |
+
"grad_norm": 3.8456506729125977,
|
| 1380 |
+
"learning_rate": 5.044444444444445e-07,
|
| 1381 |
+
"loss": 0.2556,
|
| 1382 |
+
"step": 4775
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 3.0661130629191953,
|
| 1386 |
+
"grad_norm": 4.627549171447754,
|
| 1387 |
+
"learning_rate": 4.488888888888889e-07,
|
| 1388 |
+
"loss": 0.2618,
|
| 1389 |
+
"step": 4800
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 3.0820824017885657,
|
| 1393 |
+
"grad_norm": 4.322127342224121,
|
| 1394 |
+
"learning_rate": 3.9333333333333336e-07,
|
| 1395 |
+
"loss": 0.2715,
|
| 1396 |
+
"step": 4825
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 3.0980517406579366,
|
| 1400 |
+
"grad_norm": 4.883068084716797,
|
| 1401 |
+
"learning_rate": 3.3777777777777777e-07,
|
| 1402 |
+
"loss": 0.2585,
|
| 1403 |
+
"step": 4850
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 3.1140210795273076,
|
| 1407 |
+
"grad_norm": 4.687808513641357,
|
| 1408 |
+
"learning_rate": 2.822222222222222e-07,
|
| 1409 |
+
"loss": 0.2593,
|
| 1410 |
+
"step": 4875
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 3.1299904183966785,
|
| 1414 |
+
"grad_norm": 4.111570835113525,
|
| 1415 |
+
"learning_rate": 2.266666666666667e-07,
|
| 1416 |
+
"loss": 0.2645,
|
| 1417 |
+
"step": 4900
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 3.1459597572660494,
|
| 1421 |
+
"grad_norm": 4.777975082397461,
|
| 1422 |
+
"learning_rate": 1.7111111111111114e-07,
|
| 1423 |
+
"loss": 0.2594,
|
| 1424 |
+
"step": 4925
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 3.16192909613542,
|
| 1428 |
+
"grad_norm": 4.395523548126221,
|
| 1429 |
+
"learning_rate": 1.1555555555555556e-07,
|
| 1430 |
+
"loss": 0.2665,
|
| 1431 |
+
"step": 4950
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 3.1778984350047907,
|
| 1435 |
+
"grad_norm": 5.332032203674316,
|
| 1436 |
+
"learning_rate": 6.000000000000001e-08,
|
| 1437 |
+
"loss": 0.2772,
|
| 1438 |
+
"step": 4975
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 3.1938677738741617,
|
| 1442 |
+
"grad_norm": 4.1832733154296875,
|
| 1443 |
+
"learning_rate": 4.444444444444445e-09,
|
| 1444 |
+
"loss": 0.2637,
|
| 1445 |
+
"step": 5000
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 3.1938677738741617,
|
| 1449 |
+
"eval_loss": 0.4074931740760803,
|
| 1450 |
+
"eval_runtime": 1670.4337,
|
| 1451 |
+
"eval_samples_per_second": 2.335,
|
| 1452 |
+
"eval_steps_per_second": 0.146,
|
| 1453 |
+
"eval_wer": 0.2823846789481359,
|
| 1454 |
+
"step": 5000
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 3.1938677738741617,
|
| 1458 |
+
"step": 5000,
|
| 1459 |
+
"total_flos": 5.435895365546803e+20,
|
| 1460 |
+
"train_loss": 0.7105956988334656,
|
| 1461 |
+
"train_runtime": 59674.9515,
|
| 1462 |
+
"train_samples_per_second": 2.681,
|
| 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": 4,
|
| 1470 |
+
"save_steps": 1000,
|
| 1471 |
+
"stateful_callbacks": {
|
| 1472 |
+
"TrainerControl": {
|
| 1473 |
+
"args": {
|
| 1474 |
+
"should_epoch_stop": false,
|
| 1475 |
+
"should_evaluate": false,
|
| 1476 |
+
"should_log": false,
|
| 1477 |
+
"should_save": true,
|
| 1478 |
+
"should_training_stop": true
|
| 1479 |
+
},
|
| 1480 |
+
"attributes": {}
|
| 1481 |
+
}
|
| 1482 |
+
},
|
| 1483 |
+
"total_flos": 5.435895365546803e+20,
|
| 1484 |
+
"train_batch_size": 16,
|
| 1485 |
+
"trial_name": null,
|
| 1486 |
+
"trial_params": null
|
| 1487 |
+
}
|