End of training
Browse files- README.md +9 -6
- all_results.json +13 -0
- eval_results.json +8 -0
- runs/Sep30_17-56-08_c54d50ae23c8/events.out.tfevents.1727738632.c54d50ae23c8.1212.1 +3 -0
- train_results.json +8 -0
- trainer_state.json +1487 -0
README.md
CHANGED
|
@@ -1,21 +1,24 @@
|
|
| 1 |
---
|
|
|
|
|
|
|
| 2 |
license: apache-2.0
|
| 3 |
base_model: openai/whisper-medium
|
| 4 |
tags:
|
|
|
|
| 5 |
- generated_from_trainer
|
| 6 |
datasets:
|
| 7 |
-
- common_voice_17_0
|
| 8 |
metrics:
|
| 9 |
- wer
|
| 10 |
model-index:
|
| 11 |
-
- name:
|
| 12 |
results:
|
| 13 |
- task:
|
| 14 |
name: Automatic Speech Recognition
|
| 15 |
type: automatic-speech-recognition
|
| 16 |
dataset:
|
| 17 |
-
name: common_voice_17_0
|
| 18 |
-
type: common_voice_17_0
|
| 19 |
config: pt
|
| 20 |
split: test
|
| 21 |
args: pt
|
|
@@ -28,9 +31,9 @@ model-index:
|
|
| 28 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 29 |
should probably proofread and complete it, then remove this comment. -->
|
| 30 |
|
| 31 |
-
#
|
| 32 |
|
| 33 |
-
This model is a fine-tuned version of [openai/whisper-medium](https://huggingface.co/openai/whisper-medium) on the common_voice_17_0 dataset.
|
| 34 |
It achieves the following results on the evaluation set:
|
| 35 |
- Loss: 0.1353
|
| 36 |
- Wer: 7.1230
|
|
|
|
| 1 |
---
|
| 2 |
+
language:
|
| 3 |
+
- pt
|
| 4 |
license: apache-2.0
|
| 5 |
base_model: openai/whisper-medium
|
| 6 |
tags:
|
| 7 |
+
- whisper-event
|
| 8 |
- generated_from_trainer
|
| 9 |
datasets:
|
| 10 |
+
- mozilla-foundation/common_voice_17_0
|
| 11 |
metrics:
|
| 12 |
- wer
|
| 13 |
model-index:
|
| 14 |
+
- name: Whisper Medium Mixed-Portuguese
|
| 15 |
results:
|
| 16 |
- task:
|
| 17 |
name: Automatic Speech Recognition
|
| 18 |
type: automatic-speech-recognition
|
| 19 |
dataset:
|
| 20 |
+
name: mozilla-foundation/common_voice_17_0 pt
|
| 21 |
+
type: mozilla-foundation/common_voice_17_0
|
| 22 |
config: pt
|
| 23 |
split: test
|
| 24 |
args: pt
|
|
|
|
| 31 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 32 |
should probably proofread and complete it, then remove this comment. -->
|
| 33 |
|
| 34 |
+
# Whisper Medium Mixed-Portuguese
|
| 35 |
|
| 36 |
+
This model is a fine-tuned version of [openai/whisper-medium](https://huggingface.co/openai/whisper-medium) on the mozilla-foundation/common_voice_17_0 pt dataset.
|
| 37 |
It achieves the following results on the evaluation set:
|
| 38 |
- Loss: 0.1353
|
| 39 |
- Wer: 7.1230
|
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0622,
|
| 3 |
+
"eval_loss": 0.13527874648571014,
|
| 4 |
+
"eval_runtime": 945.6236,
|
| 5 |
+
"eval_samples_per_second": 10.011,
|
| 6 |
+
"eval_steps_per_second": 1.252,
|
| 7 |
+
"eval_wer": 7.122989865404904,
|
| 8 |
+
"total_flos": 1.63271270080512e+20,
|
| 9 |
+
"train_loss": 0.15365941531658173,
|
| 10 |
+
"train_runtime": 18539.6479,
|
| 11 |
+
"train_samples_per_second": 8.63,
|
| 12 |
+
"train_steps_per_second": 0.27
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0622,
|
| 3 |
+
"eval_loss": 0.13527874648571014,
|
| 4 |
+
"eval_runtime": 945.6236,
|
| 5 |
+
"eval_samples_per_second": 10.011,
|
| 6 |
+
"eval_steps_per_second": 1.252,
|
| 7 |
+
"eval_wer": 7.122989865404904
|
| 8 |
+
}
|
runs/Sep30_17-56-08_c54d50ae23c8/events.out.tfevents.1727738632.c54d50ae23c8.1212.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:36f88e75a5a8438591d33dbdfc0bf88097556e0b42e78741776a2632b96a6956
|
| 3 |
+
size 406
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0622,
|
| 3 |
+
"total_flos": 1.63271270080512e+20,
|
| 4 |
+
"train_loss": 0.15365941531658173,
|
| 5 |
+
"train_runtime": 18539.6479,
|
| 6 |
+
"train_samples_per_second": 8.63,
|
| 7 |
+
"train_steps_per_second": 0.27
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1487 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 7.122989865404904,
|
| 3 |
+
"best_model_checkpoint": "./whisper-medium-mix-pt/checkpoint-5000",
|
| 4 |
+
"epoch": 1.0622,
|
| 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": 0.005,
|
| 13 |
+
"grad_norm": 5.954268455505371,
|
| 14 |
+
"learning_rate": 4.6000000000000004e-07,
|
| 15 |
+
"loss": 0.7195,
|
| 16 |
+
"step": 25
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.01,
|
| 20 |
+
"grad_norm": 6.020993232727051,
|
| 21 |
+
"learning_rate": 9.600000000000001e-07,
|
| 22 |
+
"loss": 0.5893,
|
| 23 |
+
"step": 50
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.015,
|
| 27 |
+
"grad_norm": 4.6267008781433105,
|
| 28 |
+
"learning_rate": 1.46e-06,
|
| 29 |
+
"loss": 0.351,
|
| 30 |
+
"step": 75
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.02,
|
| 34 |
+
"grad_norm": 5.107844829559326,
|
| 35 |
+
"learning_rate": 1.9600000000000003e-06,
|
| 36 |
+
"loss": 0.27,
|
| 37 |
+
"step": 100
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.025,
|
| 41 |
+
"grad_norm": 4.793674468994141,
|
| 42 |
+
"learning_rate": 2.46e-06,
|
| 43 |
+
"loss": 0.5947,
|
| 44 |
+
"step": 125
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.03,
|
| 48 |
+
"grad_norm": 5.168308734893799,
|
| 49 |
+
"learning_rate": 2.96e-06,
|
| 50 |
+
"loss": 0.634,
|
| 51 |
+
"step": 150
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.035,
|
| 55 |
+
"grad_norm": 5.14207124710083,
|
| 56 |
+
"learning_rate": 3.46e-06,
|
| 57 |
+
"loss": 0.4872,
|
| 58 |
+
"step": 175
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.04,
|
| 62 |
+
"grad_norm": 4.885250568389893,
|
| 63 |
+
"learning_rate": 3.96e-06,
|
| 64 |
+
"loss": 0.3443,
|
| 65 |
+
"step": 200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.045,
|
| 69 |
+
"grad_norm": 4.922640323638916,
|
| 70 |
+
"learning_rate": 4.4600000000000005e-06,
|
| 71 |
+
"loss": 0.3077,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.05,
|
| 76 |
+
"grad_norm": 3.2033262252807617,
|
| 77 |
+
"learning_rate": 4.960000000000001e-06,
|
| 78 |
+
"loss": 0.2798,
|
| 79 |
+
"step": 250
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.055,
|
| 83 |
+
"grad_norm": 4.390773773193359,
|
| 84 |
+
"learning_rate": 5.460000000000001e-06,
|
| 85 |
+
"loss": 0.237,
|
| 86 |
+
"step": 275
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.06,
|
| 90 |
+
"grad_norm": 4.187547206878662,
|
| 91 |
+
"learning_rate": 5.9600000000000005e-06,
|
| 92 |
+
"loss": 0.2354,
|
| 93 |
+
"step": 300
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.065,
|
| 97 |
+
"grad_norm": 4.3109869956970215,
|
| 98 |
+
"learning_rate": 6.460000000000001e-06,
|
| 99 |
+
"loss": 0.2123,
|
| 100 |
+
"step": 325
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.07,
|
| 104 |
+
"grad_norm": 3.2677078247070312,
|
| 105 |
+
"learning_rate": 6.96e-06,
|
| 106 |
+
"loss": 0.1923,
|
| 107 |
+
"step": 350
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.075,
|
| 111 |
+
"grad_norm": 4.168946743011475,
|
| 112 |
+
"learning_rate": 7.4600000000000006e-06,
|
| 113 |
+
"loss": 0.204,
|
| 114 |
+
"step": 375
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.08,
|
| 118 |
+
"grad_norm": 3.5255637168884277,
|
| 119 |
+
"learning_rate": 7.960000000000002e-06,
|
| 120 |
+
"loss": 0.1864,
|
| 121 |
+
"step": 400
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.085,
|
| 125 |
+
"grad_norm": 2.9705564975738525,
|
| 126 |
+
"learning_rate": 8.46e-06,
|
| 127 |
+
"loss": 0.1935,
|
| 128 |
+
"step": 425
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.09,
|
| 132 |
+
"grad_norm": 2.4820713996887207,
|
| 133 |
+
"learning_rate": 8.96e-06,
|
| 134 |
+
"loss": 0.1751,
|
| 135 |
+
"step": 450
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.095,
|
| 139 |
+
"grad_norm": 3.700296401977539,
|
| 140 |
+
"learning_rate": 9.460000000000001e-06,
|
| 141 |
+
"loss": 0.1594,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.1,
|
| 146 |
+
"grad_norm": 3.0839247703552246,
|
| 147 |
+
"learning_rate": 9.960000000000001e-06,
|
| 148 |
+
"loss": 0.1709,
|
| 149 |
+
"step": 500
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.105,
|
| 153 |
+
"grad_norm": 3.175724983215332,
|
| 154 |
+
"learning_rate": 9.94888888888889e-06,
|
| 155 |
+
"loss": 0.1246,
|
| 156 |
+
"step": 525
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.11,
|
| 160 |
+
"grad_norm": 3.514580726623535,
|
| 161 |
+
"learning_rate": 9.893333333333334e-06,
|
| 162 |
+
"loss": 0.1323,
|
| 163 |
+
"step": 550
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.115,
|
| 167 |
+
"grad_norm": 4.5611724853515625,
|
| 168 |
+
"learning_rate": 9.837777777777778e-06,
|
| 169 |
+
"loss": 0.2125,
|
| 170 |
+
"step": 575
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.12,
|
| 174 |
+
"grad_norm": 3.8415772914886475,
|
| 175 |
+
"learning_rate": 9.782222222222222e-06,
|
| 176 |
+
"loss": 0.3451,
|
| 177 |
+
"step": 600
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.125,
|
| 181 |
+
"grad_norm": 3.0463509559631348,
|
| 182 |
+
"learning_rate": 9.726666666666668e-06,
|
| 183 |
+
"loss": 0.205,
|
| 184 |
+
"step": 625
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.13,
|
| 188 |
+
"grad_norm": 2.989981174468994,
|
| 189 |
+
"learning_rate": 9.671111111111112e-06,
|
| 190 |
+
"loss": 0.1433,
|
| 191 |
+
"step": 650
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.135,
|
| 195 |
+
"grad_norm": 2.3929381370544434,
|
| 196 |
+
"learning_rate": 9.615555555555558e-06,
|
| 197 |
+
"loss": 0.1383,
|
| 198 |
+
"step": 675
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.14,
|
| 202 |
+
"grad_norm": 2.7740297317504883,
|
| 203 |
+
"learning_rate": 9.56e-06,
|
| 204 |
+
"loss": 0.1288,
|
| 205 |
+
"step": 700
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.145,
|
| 209 |
+
"grad_norm": 3.0236458778381348,
|
| 210 |
+
"learning_rate": 9.504444444444446e-06,
|
| 211 |
+
"loss": 0.1278,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.15,
|
| 216 |
+
"grad_norm": 4.738789081573486,
|
| 217 |
+
"learning_rate": 9.44888888888889e-06,
|
| 218 |
+
"loss": 0.1316,
|
| 219 |
+
"step": 750
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.155,
|
| 223 |
+
"grad_norm": 2.820563316345215,
|
| 224 |
+
"learning_rate": 9.393333333333334e-06,
|
| 225 |
+
"loss": 0.1298,
|
| 226 |
+
"step": 775
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.16,
|
| 230 |
+
"grad_norm": 3.1976284980773926,
|
| 231 |
+
"learning_rate": 9.33777777777778e-06,
|
| 232 |
+
"loss": 0.125,
|
| 233 |
+
"step": 800
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.165,
|
| 237 |
+
"grad_norm": 3.4737789630889893,
|
| 238 |
+
"learning_rate": 9.282222222222222e-06,
|
| 239 |
+
"loss": 0.1437,
|
| 240 |
+
"step": 825
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.17,
|
| 244 |
+
"grad_norm": 2.761700391769409,
|
| 245 |
+
"learning_rate": 9.226666666666668e-06,
|
| 246 |
+
"loss": 0.1285,
|
| 247 |
+
"step": 850
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.175,
|
| 251 |
+
"grad_norm": 2.7969601154327393,
|
| 252 |
+
"learning_rate": 9.171111111111112e-06,
|
| 253 |
+
"loss": 0.1091,
|
| 254 |
+
"step": 875
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.18,
|
| 258 |
+
"grad_norm": 3.8701090812683105,
|
| 259 |
+
"learning_rate": 9.115555555555556e-06,
|
| 260 |
+
"loss": 0.1383,
|
| 261 |
+
"step": 900
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.185,
|
| 265 |
+
"grad_norm": 2.473555088043213,
|
| 266 |
+
"learning_rate": 9.060000000000001e-06,
|
| 267 |
+
"loss": 0.1067,
|
| 268 |
+
"step": 925
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.19,
|
| 272 |
+
"grad_norm": 3.0442628860473633,
|
| 273 |
+
"learning_rate": 9.004444444444445e-06,
|
| 274 |
+
"loss": 0.1265,
|
| 275 |
+
"step": 950
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.195,
|
| 279 |
+
"grad_norm": 3.2474780082702637,
|
| 280 |
+
"learning_rate": 8.94888888888889e-06,
|
| 281 |
+
"loss": 0.1254,
|
| 282 |
+
"step": 975
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.2,
|
| 286 |
+
"grad_norm": 1.9548333883285522,
|
| 287 |
+
"learning_rate": 8.893333333333333e-06,
|
| 288 |
+
"loss": 0.1116,
|
| 289 |
+
"step": 1000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.2,
|
| 293 |
+
"eval_loss": 0.15698818862438202,
|
| 294 |
+
"eval_runtime": 947.4383,
|
| 295 |
+
"eval_samples_per_second": 9.992,
|
| 296 |
+
"eval_steps_per_second": 1.25,
|
| 297 |
+
"eval_wer": 8.582435499856134,
|
| 298 |
+
"step": 1000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.205,
|
| 302 |
+
"grad_norm": 2.7346816062927246,
|
| 303 |
+
"learning_rate": 8.83777777777778e-06,
|
| 304 |
+
"loss": 0.1005,
|
| 305 |
+
"step": 1025
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.21,
|
| 309 |
+
"grad_norm": 3.2607483863830566,
|
| 310 |
+
"learning_rate": 8.782222222222223e-06,
|
| 311 |
+
"loss": 0.0931,
|
| 312 |
+
"step": 1050
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.215,
|
| 316 |
+
"grad_norm": 2.8954336643218994,
|
| 317 |
+
"learning_rate": 8.726666666666667e-06,
|
| 318 |
+
"loss": 0.1111,
|
| 319 |
+
"step": 1075
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.22,
|
| 323 |
+
"grad_norm": 3.1846327781677246,
|
| 324 |
+
"learning_rate": 8.671111111111113e-06,
|
| 325 |
+
"loss": 0.1003,
|
| 326 |
+
"step": 1100
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.225,
|
| 330 |
+
"grad_norm": 2.284499168395996,
|
| 331 |
+
"learning_rate": 8.615555555555555e-06,
|
| 332 |
+
"loss": 0.1018,
|
| 333 |
+
"step": 1125
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.23,
|
| 337 |
+
"grad_norm": 2.8060786724090576,
|
| 338 |
+
"learning_rate": 8.560000000000001e-06,
|
| 339 |
+
"loss": 0.1145,
|
| 340 |
+
"step": 1150
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.235,
|
| 344 |
+
"grad_norm": 2.9943974018096924,
|
| 345 |
+
"learning_rate": 8.504444444444445e-06,
|
| 346 |
+
"loss": 0.1247,
|
| 347 |
+
"step": 1175
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 0.24,
|
| 351 |
+
"grad_norm": 2.2265496253967285,
|
| 352 |
+
"learning_rate": 8.448888888888889e-06,
|
| 353 |
+
"loss": 0.1193,
|
| 354 |
+
"step": 1200
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.245,
|
| 358 |
+
"grad_norm": 2.4557271003723145,
|
| 359 |
+
"learning_rate": 8.393333333333335e-06,
|
| 360 |
+
"loss": 0.1209,
|
| 361 |
+
"step": 1225
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 0.25,
|
| 365 |
+
"grad_norm": 2.332017183303833,
|
| 366 |
+
"learning_rate": 8.337777777777777e-06,
|
| 367 |
+
"loss": 0.114,
|
| 368 |
+
"step": 1250
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 0.255,
|
| 372 |
+
"grad_norm": 3.056497812271118,
|
| 373 |
+
"learning_rate": 8.282222222222223e-06,
|
| 374 |
+
"loss": 0.1043,
|
| 375 |
+
"step": 1275
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.26,
|
| 379 |
+
"grad_norm": 3.024203062057495,
|
| 380 |
+
"learning_rate": 8.226666666666667e-06,
|
| 381 |
+
"loss": 0.1247,
|
| 382 |
+
"step": 1300
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 0.265,
|
| 386 |
+
"grad_norm": 2.543893814086914,
|
| 387 |
+
"learning_rate": 8.171111111111113e-06,
|
| 388 |
+
"loss": 0.1347,
|
| 389 |
+
"step": 1325
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 0.27,
|
| 393 |
+
"grad_norm": 2.2330307960510254,
|
| 394 |
+
"learning_rate": 8.115555555555557e-06,
|
| 395 |
+
"loss": 0.119,
|
| 396 |
+
"step": 1350
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 0.275,
|
| 400 |
+
"grad_norm": 2.819495677947998,
|
| 401 |
+
"learning_rate": 8.06e-06,
|
| 402 |
+
"loss": 0.0998,
|
| 403 |
+
"step": 1375
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.28,
|
| 407 |
+
"grad_norm": 2.6235690116882324,
|
| 408 |
+
"learning_rate": 8.004444444444445e-06,
|
| 409 |
+
"loss": 0.1037,
|
| 410 |
+
"step": 1400
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.285,
|
| 414 |
+
"grad_norm": 3.0939559936523438,
|
| 415 |
+
"learning_rate": 7.948888888888889e-06,
|
| 416 |
+
"loss": 0.1027,
|
| 417 |
+
"step": 1425
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.29,
|
| 421 |
+
"grad_norm": 2.9563097953796387,
|
| 422 |
+
"learning_rate": 7.893333333333335e-06,
|
| 423 |
+
"loss": 0.106,
|
| 424 |
+
"step": 1450
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 0.295,
|
| 428 |
+
"grad_norm": 2.6161818504333496,
|
| 429 |
+
"learning_rate": 7.837777777777779e-06,
|
| 430 |
+
"loss": 0.1112,
|
| 431 |
+
"step": 1475
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.3,
|
| 435 |
+
"grad_norm": 2.455970525741577,
|
| 436 |
+
"learning_rate": 7.782222222222223e-06,
|
| 437 |
+
"loss": 0.1138,
|
| 438 |
+
"step": 1500
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 0.305,
|
| 442 |
+
"grad_norm": 2.8548126220703125,
|
| 443 |
+
"learning_rate": 7.726666666666667e-06,
|
| 444 |
+
"loss": 0.0962,
|
| 445 |
+
"step": 1525
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 0.31,
|
| 449 |
+
"grad_norm": 2.414487838745117,
|
| 450 |
+
"learning_rate": 7.67111111111111e-06,
|
| 451 |
+
"loss": 0.105,
|
| 452 |
+
"step": 1550
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 0.315,
|
| 456 |
+
"grad_norm": 2.747642993927002,
|
| 457 |
+
"learning_rate": 7.6155555555555564e-06,
|
| 458 |
+
"loss": 0.1111,
|
| 459 |
+
"step": 1575
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.32,
|
| 463 |
+
"grad_norm": 2.5105972290039062,
|
| 464 |
+
"learning_rate": 7.5600000000000005e-06,
|
| 465 |
+
"loss": 0.1066,
|
| 466 |
+
"step": 1600
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 0.325,
|
| 470 |
+
"grad_norm": 2.765288829803467,
|
| 471 |
+
"learning_rate": 7.504444444444445e-06,
|
| 472 |
+
"loss": 0.1423,
|
| 473 |
+
"step": 1625
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 0.33,
|
| 477 |
+
"grad_norm": 3.616039991378784,
|
| 478 |
+
"learning_rate": 7.44888888888889e-06,
|
| 479 |
+
"loss": 0.1527,
|
| 480 |
+
"step": 1650
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 0.335,
|
| 484 |
+
"grad_norm": 3.961867094039917,
|
| 485 |
+
"learning_rate": 7.393333333333333e-06,
|
| 486 |
+
"loss": 0.1409,
|
| 487 |
+
"step": 1675
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 0.34,
|
| 491 |
+
"grad_norm": 3.5502121448516846,
|
| 492 |
+
"learning_rate": 7.337777777777778e-06,
|
| 493 |
+
"loss": 0.1375,
|
| 494 |
+
"step": 1700
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.345,
|
| 498 |
+
"grad_norm": 2.9982211589813232,
|
| 499 |
+
"learning_rate": 7.282222222222222e-06,
|
| 500 |
+
"loss": 0.1208,
|
| 501 |
+
"step": 1725
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.35,
|
| 505 |
+
"grad_norm": 3.1248340606689453,
|
| 506 |
+
"learning_rate": 7.226666666666667e-06,
|
| 507 |
+
"loss": 0.1375,
|
| 508 |
+
"step": 1750
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 0.355,
|
| 512 |
+
"grad_norm": 2.1703038215637207,
|
| 513 |
+
"learning_rate": 7.171111111111112e-06,
|
| 514 |
+
"loss": 0.1513,
|
| 515 |
+
"step": 1775
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 0.36,
|
| 519 |
+
"grad_norm": 2.672173500061035,
|
| 520 |
+
"learning_rate": 7.115555555555557e-06,
|
| 521 |
+
"loss": 0.1169,
|
| 522 |
+
"step": 1800
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.365,
|
| 526 |
+
"grad_norm": 2.1961538791656494,
|
| 527 |
+
"learning_rate": 7.06e-06,
|
| 528 |
+
"loss": 0.1072,
|
| 529 |
+
"step": 1825
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 0.37,
|
| 533 |
+
"grad_norm": 2.129289388656616,
|
| 534 |
+
"learning_rate": 7.004444444444445e-06,
|
| 535 |
+
"loss": 0.1022,
|
| 536 |
+
"step": 1850
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 0.375,
|
| 540 |
+
"grad_norm": 2.4798481464385986,
|
| 541 |
+
"learning_rate": 6.948888888888889e-06,
|
| 542 |
+
"loss": 0.1129,
|
| 543 |
+
"step": 1875
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 0.38,
|
| 547 |
+
"grad_norm": 2.9369380474090576,
|
| 548 |
+
"learning_rate": 6.893333333333334e-06,
|
| 549 |
+
"loss": 0.1149,
|
| 550 |
+
"step": 1900
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 0.385,
|
| 554 |
+
"grad_norm": 2.7497618198394775,
|
| 555 |
+
"learning_rate": 6.837777777777779e-06,
|
| 556 |
+
"loss": 0.1181,
|
| 557 |
+
"step": 1925
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 0.39,
|
| 561 |
+
"grad_norm": 2.4043569564819336,
|
| 562 |
+
"learning_rate": 6.782222222222222e-06,
|
| 563 |
+
"loss": 0.0998,
|
| 564 |
+
"step": 1950
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 0.395,
|
| 568 |
+
"grad_norm": 2.9279298782348633,
|
| 569 |
+
"learning_rate": 6.726666666666667e-06,
|
| 570 |
+
"loss": 0.1035,
|
| 571 |
+
"step": 1975
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 0.4,
|
| 575 |
+
"grad_norm": 2.7362992763519287,
|
| 576 |
+
"learning_rate": 6.671111111111112e-06,
|
| 577 |
+
"loss": 0.105,
|
| 578 |
+
"step": 2000
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 0.4,
|
| 582 |
+
"eval_loss": 0.14844202995300293,
|
| 583 |
+
"eval_runtime": 940.7954,
|
| 584 |
+
"eval_samples_per_second": 10.063,
|
| 585 |
+
"eval_steps_per_second": 1.259,
|
| 586 |
+
"eval_wer": 7.93983183605614,
|
| 587 |
+
"step": 2000
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 0.405,
|
| 591 |
+
"grad_norm": 2.0498945713043213,
|
| 592 |
+
"learning_rate": 6.615555555555556e-06,
|
| 593 |
+
"loss": 0.1086,
|
| 594 |
+
"step": 2025
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 0.41,
|
| 598 |
+
"grad_norm": 2.7259154319763184,
|
| 599 |
+
"learning_rate": 6.560000000000001e-06,
|
| 600 |
+
"loss": 0.1176,
|
| 601 |
+
"step": 2050
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 0.415,
|
| 605 |
+
"grad_norm": 2.4379186630249023,
|
| 606 |
+
"learning_rate": 6.504444444444446e-06,
|
| 607 |
+
"loss": 0.1157,
|
| 608 |
+
"step": 2075
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 0.42,
|
| 612 |
+
"grad_norm": 2.397427558898926,
|
| 613 |
+
"learning_rate": 6.448888888888889e-06,
|
| 614 |
+
"loss": 0.1229,
|
| 615 |
+
"step": 2100
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 0.425,
|
| 619 |
+
"grad_norm": 2.2779042720794678,
|
| 620 |
+
"learning_rate": 6.393333333333334e-06,
|
| 621 |
+
"loss": 0.0934,
|
| 622 |
+
"step": 2125
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 0.43,
|
| 626 |
+
"grad_norm": 1.9669123888015747,
|
| 627 |
+
"learning_rate": 6.3377777777777786e-06,
|
| 628 |
+
"loss": 0.0968,
|
| 629 |
+
"step": 2150
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 0.435,
|
| 633 |
+
"grad_norm": 2.051048755645752,
|
| 634 |
+
"learning_rate": 6.282222222222223e-06,
|
| 635 |
+
"loss": 0.0876,
|
| 636 |
+
"step": 2175
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 0.44,
|
| 640 |
+
"grad_norm": 2.082129716873169,
|
| 641 |
+
"learning_rate": 6.2266666666666675e-06,
|
| 642 |
+
"loss": 0.0832,
|
| 643 |
+
"step": 2200
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 0.445,
|
| 647 |
+
"grad_norm": 2.0666728019714355,
|
| 648 |
+
"learning_rate": 6.171111111111112e-06,
|
| 649 |
+
"loss": 0.081,
|
| 650 |
+
"step": 2225
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 0.45,
|
| 654 |
+
"grad_norm": 2.1512069702148438,
|
| 655 |
+
"learning_rate": 6.1155555555555555e-06,
|
| 656 |
+
"loss": 0.0795,
|
| 657 |
+
"step": 2250
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.455,
|
| 661 |
+
"grad_norm": 2.393537998199463,
|
| 662 |
+
"learning_rate": 6.0600000000000004e-06,
|
| 663 |
+
"loss": 0.0797,
|
| 664 |
+
"step": 2275
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 0.46,
|
| 668 |
+
"grad_norm": 2.4059464931488037,
|
| 669 |
+
"learning_rate": 6.004444444444445e-06,
|
| 670 |
+
"loss": 0.086,
|
| 671 |
+
"step": 2300
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 0.465,
|
| 675 |
+
"grad_norm": 2.3822548389434814,
|
| 676 |
+
"learning_rate": 5.948888888888889e-06,
|
| 677 |
+
"loss": 0.0883,
|
| 678 |
+
"step": 2325
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 0.47,
|
| 682 |
+
"grad_norm": 2.063124656677246,
|
| 683 |
+
"learning_rate": 5.893333333333334e-06,
|
| 684 |
+
"loss": 0.082,
|
| 685 |
+
"step": 2350
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 0.475,
|
| 689 |
+
"grad_norm": 2.988023281097412,
|
| 690 |
+
"learning_rate": 5.837777777777777e-06,
|
| 691 |
+
"loss": 0.194,
|
| 692 |
+
"step": 2375
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 0.48,
|
| 696 |
+
"grad_norm": 3.959836959838867,
|
| 697 |
+
"learning_rate": 5.782222222222222e-06,
|
| 698 |
+
"loss": 0.2993,
|
| 699 |
+
"step": 2400
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.485,
|
| 703 |
+
"grad_norm": 2.79730486869812,
|
| 704 |
+
"learning_rate": 5.726666666666667e-06,
|
| 705 |
+
"loss": 0.2067,
|
| 706 |
+
"step": 2425
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 0.49,
|
| 710 |
+
"grad_norm": 1.7770684957504272,
|
| 711 |
+
"learning_rate": 5.671111111111112e-06,
|
| 712 |
+
"loss": 0.1769,
|
| 713 |
+
"step": 2450
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 0.495,
|
| 717 |
+
"grad_norm": 2.2204787731170654,
|
| 718 |
+
"learning_rate": 5.615555555555556e-06,
|
| 719 |
+
"loss": 0.0911,
|
| 720 |
+
"step": 2475
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 0.5,
|
| 724 |
+
"grad_norm": 3.053565263748169,
|
| 725 |
+
"learning_rate": 5.560000000000001e-06,
|
| 726 |
+
"loss": 0.1141,
|
| 727 |
+
"step": 2500
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 0.505,
|
| 731 |
+
"grad_norm": 5.307312488555908,
|
| 732 |
+
"learning_rate": 5.504444444444444e-06,
|
| 733 |
+
"loss": 0.4682,
|
| 734 |
+
"step": 2525
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 0.51,
|
| 738 |
+
"grad_norm": 4.99273157119751,
|
| 739 |
+
"learning_rate": 5.448888888888889e-06,
|
| 740 |
+
"loss": 0.4198,
|
| 741 |
+
"step": 2550
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 0.515,
|
| 745 |
+
"grad_norm": 4.5542707443237305,
|
| 746 |
+
"learning_rate": 5.393333333333334e-06,
|
| 747 |
+
"loss": 0.3805,
|
| 748 |
+
"step": 2575
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 0.52,
|
| 752 |
+
"grad_norm": 3.3522355556488037,
|
| 753 |
+
"learning_rate": 5.337777777777779e-06,
|
| 754 |
+
"loss": 0.3213,
|
| 755 |
+
"step": 2600
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 0.525,
|
| 759 |
+
"grad_norm": 3.8808069229125977,
|
| 760 |
+
"learning_rate": 5.282222222222223e-06,
|
| 761 |
+
"loss": 0.2504,
|
| 762 |
+
"step": 2625
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 0.53,
|
| 766 |
+
"grad_norm": 3.6517879962921143,
|
| 767 |
+
"learning_rate": 5.226666666666667e-06,
|
| 768 |
+
"loss": 0.2508,
|
| 769 |
+
"step": 2650
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 0.535,
|
| 773 |
+
"grad_norm": 5.0575151443481445,
|
| 774 |
+
"learning_rate": 5.171111111111111e-06,
|
| 775 |
+
"loss": 0.288,
|
| 776 |
+
"step": 2675
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 0.54,
|
| 780 |
+
"grad_norm": 4.049450874328613,
|
| 781 |
+
"learning_rate": 5.115555555555556e-06,
|
| 782 |
+
"loss": 0.2526,
|
| 783 |
+
"step": 2700
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 0.545,
|
| 787 |
+
"grad_norm": 3.075061798095703,
|
| 788 |
+
"learning_rate": 5.060000000000001e-06,
|
| 789 |
+
"loss": 0.25,
|
| 790 |
+
"step": 2725
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 0.55,
|
| 794 |
+
"grad_norm": 3.80968976020813,
|
| 795 |
+
"learning_rate": 5.004444444444445e-06,
|
| 796 |
+
"loss": 0.2388,
|
| 797 |
+
"step": 2750
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 0.555,
|
| 801 |
+
"grad_norm": 3.044834852218628,
|
| 802 |
+
"learning_rate": 4.94888888888889e-06,
|
| 803 |
+
"loss": 0.1718,
|
| 804 |
+
"step": 2775
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 0.56,
|
| 808 |
+
"grad_norm": 2.3228962421417236,
|
| 809 |
+
"learning_rate": 4.893333333333334e-06,
|
| 810 |
+
"loss": 0.148,
|
| 811 |
+
"step": 2800
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 0.565,
|
| 815 |
+
"grad_norm": 2.9311017990112305,
|
| 816 |
+
"learning_rate": 4.837777777777778e-06,
|
| 817 |
+
"loss": 0.1308,
|
| 818 |
+
"step": 2825
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 0.57,
|
| 822 |
+
"grad_norm": 2.485396146774292,
|
| 823 |
+
"learning_rate": 4.7822222222222226e-06,
|
| 824 |
+
"loss": 0.1197,
|
| 825 |
+
"step": 2850
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 0.575,
|
| 829 |
+
"grad_norm": 3.5410890579223633,
|
| 830 |
+
"learning_rate": 4.7266666666666674e-06,
|
| 831 |
+
"loss": 0.0878,
|
| 832 |
+
"step": 2875
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 0.58,
|
| 836 |
+
"grad_norm": 2.246903657913208,
|
| 837 |
+
"learning_rate": 4.6711111111111115e-06,
|
| 838 |
+
"loss": 0.0793,
|
| 839 |
+
"step": 2900
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 0.585,
|
| 843 |
+
"grad_norm": 2.251746416091919,
|
| 844 |
+
"learning_rate": 4.6155555555555555e-06,
|
| 845 |
+
"loss": 0.0795,
|
| 846 |
+
"step": 2925
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 0.59,
|
| 850 |
+
"grad_norm": 1.791506052017212,
|
| 851 |
+
"learning_rate": 4.56e-06,
|
| 852 |
+
"loss": 0.0918,
|
| 853 |
+
"step": 2950
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 0.595,
|
| 857 |
+
"grad_norm": 1.3495018482208252,
|
| 858 |
+
"learning_rate": 4.504444444444444e-06,
|
| 859 |
+
"loss": 0.077,
|
| 860 |
+
"step": 2975
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 0.6,
|
| 864 |
+
"grad_norm": 1.8840142488479614,
|
| 865 |
+
"learning_rate": 4.448888888888889e-06,
|
| 866 |
+
"loss": 0.0783,
|
| 867 |
+
"step": 3000
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 0.6,
|
| 871 |
+
"eval_loss": 0.13742496073246002,
|
| 872 |
+
"eval_runtime": 944.2672,
|
| 873 |
+
"eval_samples_per_second": 10.026,
|
| 874 |
+
"eval_steps_per_second": 1.254,
|
| 875 |
+
"eval_wer": 7.447488730458135,
|
| 876 |
+
"step": 3000
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 0.605,
|
| 880 |
+
"grad_norm": 1.7588751316070557,
|
| 881 |
+
"learning_rate": 4.393333333333334e-06,
|
| 882 |
+
"loss": 0.0747,
|
| 883 |
+
"step": 3025
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 0.61,
|
| 887 |
+
"grad_norm": 1.9886692762374878,
|
| 888 |
+
"learning_rate": 4.337777777777778e-06,
|
| 889 |
+
"loss": 0.076,
|
| 890 |
+
"step": 3050
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 0.615,
|
| 894 |
+
"grad_norm": 2.5392279624938965,
|
| 895 |
+
"learning_rate": 4.282222222222222e-06,
|
| 896 |
+
"loss": 0.0796,
|
| 897 |
+
"step": 3075
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 0.62,
|
| 901 |
+
"grad_norm": 2.3111472129821777,
|
| 902 |
+
"learning_rate": 4.226666666666667e-06,
|
| 903 |
+
"loss": 0.0898,
|
| 904 |
+
"step": 3100
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 0.625,
|
| 908 |
+
"grad_norm": 1.8798396587371826,
|
| 909 |
+
"learning_rate": 4.171111111111111e-06,
|
| 910 |
+
"loss": 0.0742,
|
| 911 |
+
"step": 3125
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 0.63,
|
| 915 |
+
"grad_norm": 1.8901393413543701,
|
| 916 |
+
"learning_rate": 4.115555555555556e-06,
|
| 917 |
+
"loss": 0.0815,
|
| 918 |
+
"step": 3150
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 0.635,
|
| 922 |
+
"grad_norm": 3.506927490234375,
|
| 923 |
+
"learning_rate": 4.060000000000001e-06,
|
| 924 |
+
"loss": 0.1055,
|
| 925 |
+
"step": 3175
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 0.64,
|
| 929 |
+
"grad_norm": 3.018944025039673,
|
| 930 |
+
"learning_rate": 4.004444444444445e-06,
|
| 931 |
+
"loss": 0.1541,
|
| 932 |
+
"step": 3200
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 0.645,
|
| 936 |
+
"grad_norm": 2.304964303970337,
|
| 937 |
+
"learning_rate": 3.948888888888889e-06,
|
| 938 |
+
"loss": 0.1213,
|
| 939 |
+
"step": 3225
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 0.65,
|
| 943 |
+
"grad_norm": 2.425555944442749,
|
| 944 |
+
"learning_rate": 3.893333333333333e-06,
|
| 945 |
+
"loss": 0.0935,
|
| 946 |
+
"step": 3250
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 0.655,
|
| 950 |
+
"grad_norm": 1.5498875379562378,
|
| 951 |
+
"learning_rate": 3.837777777777778e-06,
|
| 952 |
+
"loss": 0.0838,
|
| 953 |
+
"step": 3275
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 0.66,
|
| 957 |
+
"grad_norm": 2.360494613647461,
|
| 958 |
+
"learning_rate": 3.782222222222223e-06,
|
| 959 |
+
"loss": 0.0777,
|
| 960 |
+
"step": 3300
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 0.665,
|
| 964 |
+
"grad_norm": 1.5479514598846436,
|
| 965 |
+
"learning_rate": 3.726666666666667e-06,
|
| 966 |
+
"loss": 0.0778,
|
| 967 |
+
"step": 3325
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 0.67,
|
| 971 |
+
"grad_norm": 2.920868158340454,
|
| 972 |
+
"learning_rate": 3.6711111111111113e-06,
|
| 973 |
+
"loss": 0.0837,
|
| 974 |
+
"step": 3350
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 0.675,
|
| 978 |
+
"grad_norm": 1.8797603845596313,
|
| 979 |
+
"learning_rate": 3.615555555555556e-06,
|
| 980 |
+
"loss": 0.0727,
|
| 981 |
+
"step": 3375
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 0.68,
|
| 985 |
+
"grad_norm": 2.5125744342803955,
|
| 986 |
+
"learning_rate": 3.5600000000000002e-06,
|
| 987 |
+
"loss": 0.0798,
|
| 988 |
+
"step": 3400
|
| 989 |
+
},
|
| 990 |
+
{
|
| 991 |
+
"epoch": 0.685,
|
| 992 |
+
"grad_norm": 2.6805381774902344,
|
| 993 |
+
"learning_rate": 3.5044444444444447e-06,
|
| 994 |
+
"loss": 0.0825,
|
| 995 |
+
"step": 3425
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 0.69,
|
| 999 |
+
"grad_norm": 2.2335715293884277,
|
| 1000 |
+
"learning_rate": 3.4488888888888896e-06,
|
| 1001 |
+
"loss": 0.1145,
|
| 1002 |
+
"step": 3450
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 0.695,
|
| 1006 |
+
"grad_norm": 3.3753836154937744,
|
| 1007 |
+
"learning_rate": 3.3933333333333336e-06,
|
| 1008 |
+
"loss": 0.1083,
|
| 1009 |
+
"step": 3475
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 0.7,
|
| 1013 |
+
"grad_norm": 3.470496654510498,
|
| 1014 |
+
"learning_rate": 3.337777777777778e-06,
|
| 1015 |
+
"loss": 0.3921,
|
| 1016 |
+
"step": 3500
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 0.705,
|
| 1020 |
+
"grad_norm": 1.5432311296463013,
|
| 1021 |
+
"learning_rate": 3.282222222222223e-06,
|
| 1022 |
+
"loss": 0.161,
|
| 1023 |
+
"step": 3525
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 0.71,
|
| 1027 |
+
"grad_norm": 2.0430617332458496,
|
| 1028 |
+
"learning_rate": 3.226666666666667e-06,
|
| 1029 |
+
"loss": 0.1171,
|
| 1030 |
+
"step": 3550
|
| 1031 |
+
},
|
| 1032 |
+
{
|
| 1033 |
+
"epoch": 0.715,
|
| 1034 |
+
"grad_norm": 1.5512393712997437,
|
| 1035 |
+
"learning_rate": 3.1711111111111114e-06,
|
| 1036 |
+
"loss": 0.08,
|
| 1037 |
+
"step": 3575
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 0.72,
|
| 1041 |
+
"grad_norm": 1.4717549085617065,
|
| 1042 |
+
"learning_rate": 3.1155555555555555e-06,
|
| 1043 |
+
"loss": 0.0726,
|
| 1044 |
+
"step": 3600
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 0.725,
|
| 1048 |
+
"grad_norm": 1.5445371866226196,
|
| 1049 |
+
"learning_rate": 3.0600000000000003e-06,
|
| 1050 |
+
"loss": 0.0589,
|
| 1051 |
+
"step": 3625
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 0.73,
|
| 1055 |
+
"grad_norm": 1.693588376045227,
|
| 1056 |
+
"learning_rate": 3.004444444444445e-06,
|
| 1057 |
+
"loss": 0.0606,
|
| 1058 |
+
"step": 3650
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 0.735,
|
| 1062 |
+
"grad_norm": 1.875542163848877,
|
| 1063 |
+
"learning_rate": 2.948888888888889e-06,
|
| 1064 |
+
"loss": 0.0548,
|
| 1065 |
+
"step": 3675
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 0.74,
|
| 1069 |
+
"grad_norm": 1.2488154172897339,
|
| 1070 |
+
"learning_rate": 2.8933333333333337e-06,
|
| 1071 |
+
"loss": 0.0517,
|
| 1072 |
+
"step": 3700
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"epoch": 0.745,
|
| 1076 |
+
"grad_norm": 1.5682026147842407,
|
| 1077 |
+
"learning_rate": 2.837777777777778e-06,
|
| 1078 |
+
"loss": 0.0571,
|
| 1079 |
+
"step": 3725
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 0.75,
|
| 1083 |
+
"grad_norm": 1.1433475017547607,
|
| 1084 |
+
"learning_rate": 2.7822222222222222e-06,
|
| 1085 |
+
"loss": 0.0513,
|
| 1086 |
+
"step": 3750
|
| 1087 |
+
},
|
| 1088 |
+
{
|
| 1089 |
+
"epoch": 0.755,
|
| 1090 |
+
"grad_norm": 1.7756175994873047,
|
| 1091 |
+
"learning_rate": 2.726666666666667e-06,
|
| 1092 |
+
"loss": 0.047,
|
| 1093 |
+
"step": 3775
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"epoch": 0.76,
|
| 1097 |
+
"grad_norm": 1.6270946264266968,
|
| 1098 |
+
"learning_rate": 2.6711111111111116e-06,
|
| 1099 |
+
"loss": 0.0528,
|
| 1100 |
+
"step": 3800
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 0.765,
|
| 1104 |
+
"grad_norm": 1.3709731101989746,
|
| 1105 |
+
"learning_rate": 2.6155555555555556e-06,
|
| 1106 |
+
"loss": 0.0507,
|
| 1107 |
+
"step": 3825
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 0.77,
|
| 1111 |
+
"grad_norm": 1.4345024824142456,
|
| 1112 |
+
"learning_rate": 2.56e-06,
|
| 1113 |
+
"loss": 0.0455,
|
| 1114 |
+
"step": 3850
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 0.775,
|
| 1118 |
+
"grad_norm": 2.159369707107544,
|
| 1119 |
+
"learning_rate": 2.504444444444445e-06,
|
| 1120 |
+
"loss": 0.0548,
|
| 1121 |
+
"step": 3875
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 0.78,
|
| 1125 |
+
"grad_norm": 2.050881862640381,
|
| 1126 |
+
"learning_rate": 2.448888888888889e-06,
|
| 1127 |
+
"loss": 0.0833,
|
| 1128 |
+
"step": 3900
|
| 1129 |
+
},
|
| 1130 |
+
{
|
| 1131 |
+
"epoch": 0.785,
|
| 1132 |
+
"grad_norm": 2.428863763809204,
|
| 1133 |
+
"learning_rate": 2.3933333333333334e-06,
|
| 1134 |
+
"loss": 0.1491,
|
| 1135 |
+
"step": 3925
|
| 1136 |
+
},
|
| 1137 |
+
{
|
| 1138 |
+
"epoch": 0.79,
|
| 1139 |
+
"grad_norm": 2.655684471130371,
|
| 1140 |
+
"learning_rate": 2.337777777777778e-06,
|
| 1141 |
+
"loss": 0.1254,
|
| 1142 |
+
"step": 3950
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 0.795,
|
| 1146 |
+
"grad_norm": 1.752144455909729,
|
| 1147 |
+
"learning_rate": 2.2822222222222223e-06,
|
| 1148 |
+
"loss": 0.1352,
|
| 1149 |
+
"step": 3975
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 0.8,
|
| 1153 |
+
"grad_norm": 2.89431095123291,
|
| 1154 |
+
"learning_rate": 2.226666666666667e-06,
|
| 1155 |
+
"loss": 0.1703,
|
| 1156 |
+
"step": 4000
|
| 1157 |
+
},
|
| 1158 |
+
{
|
| 1159 |
+
"epoch": 0.8,
|
| 1160 |
+
"eval_loss": 0.13704591989517212,
|
| 1161 |
+
"eval_runtime": 942.2184,
|
| 1162 |
+
"eval_samples_per_second": 10.048,
|
| 1163 |
+
"eval_steps_per_second": 1.257,
|
| 1164 |
+
"eval_wer": 7.241280092074555,
|
| 1165 |
+
"step": 4000
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 0.805,
|
| 1169 |
+
"grad_norm": 2.1112782955169678,
|
| 1170 |
+
"learning_rate": 2.1711111111111113e-06,
|
| 1171 |
+
"loss": 0.137,
|
| 1172 |
+
"step": 4025
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 0.81,
|
| 1176 |
+
"grad_norm": 2.241307258605957,
|
| 1177 |
+
"learning_rate": 2.1155555555555557e-06,
|
| 1178 |
+
"loss": 0.0809,
|
| 1179 |
+
"step": 4050
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.815,
|
| 1183 |
+
"grad_norm": 1.5932738780975342,
|
| 1184 |
+
"learning_rate": 2.06e-06,
|
| 1185 |
+
"loss": 0.0605,
|
| 1186 |
+
"step": 4075
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 0.82,
|
| 1190 |
+
"grad_norm": 1.965742588043213,
|
| 1191 |
+
"learning_rate": 2.006666666666667e-06,
|
| 1192 |
+
"loss": 0.0549,
|
| 1193 |
+
"step": 4100
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 0.825,
|
| 1197 |
+
"grad_norm": 3.738175868988037,
|
| 1198 |
+
"learning_rate": 1.9511111111111113e-06,
|
| 1199 |
+
"loss": 0.0557,
|
| 1200 |
+
"step": 4125
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 0.83,
|
| 1204 |
+
"grad_norm": 1.9460804462432861,
|
| 1205 |
+
"learning_rate": 1.8955555555555557e-06,
|
| 1206 |
+
"loss": 0.0433,
|
| 1207 |
+
"step": 4150
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 0.835,
|
| 1211 |
+
"grad_norm": 1.6488566398620605,
|
| 1212 |
+
"learning_rate": 1.8400000000000002e-06,
|
| 1213 |
+
"loss": 0.0655,
|
| 1214 |
+
"step": 4175
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 0.84,
|
| 1218 |
+
"grad_norm": 1.511797308921814,
|
| 1219 |
+
"learning_rate": 1.7844444444444444e-06,
|
| 1220 |
+
"loss": 0.0787,
|
| 1221 |
+
"step": 4200
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 0.845,
|
| 1225 |
+
"grad_norm": 2.361323833465576,
|
| 1226 |
+
"learning_rate": 1.728888888888889e-06,
|
| 1227 |
+
"loss": 0.0685,
|
| 1228 |
+
"step": 4225
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 0.85,
|
| 1232 |
+
"grad_norm": 1.1696654558181763,
|
| 1233 |
+
"learning_rate": 1.6733333333333335e-06,
|
| 1234 |
+
"loss": 0.0518,
|
| 1235 |
+
"step": 4250
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 0.855,
|
| 1239 |
+
"grad_norm": 1.6606045961380005,
|
| 1240 |
+
"learning_rate": 1.6177777777777778e-06,
|
| 1241 |
+
"loss": 0.0509,
|
| 1242 |
+
"step": 4275
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 0.86,
|
| 1246 |
+
"grad_norm": 1.3564594984054565,
|
| 1247 |
+
"learning_rate": 1.5622222222222225e-06,
|
| 1248 |
+
"loss": 0.0545,
|
| 1249 |
+
"step": 4300
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 0.865,
|
| 1253 |
+
"grad_norm": 2.545974016189575,
|
| 1254 |
+
"learning_rate": 1.506666666666667e-06,
|
| 1255 |
+
"loss": 0.0953,
|
| 1256 |
+
"step": 4325
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 0.87,
|
| 1260 |
+
"grad_norm": 5.036618232727051,
|
| 1261 |
+
"learning_rate": 1.4511111111111112e-06,
|
| 1262 |
+
"loss": 0.1465,
|
| 1263 |
+
"step": 4350
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 0.875,
|
| 1267 |
+
"grad_norm": 3.5834786891937256,
|
| 1268 |
+
"learning_rate": 1.3955555555555556e-06,
|
| 1269 |
+
"loss": 0.4963,
|
| 1270 |
+
"step": 4375
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 0.88,
|
| 1274 |
+
"grad_norm": 4.646224498748779,
|
| 1275 |
+
"learning_rate": 1.34e-06,
|
| 1276 |
+
"loss": 0.5522,
|
| 1277 |
+
"step": 4400
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 0.885,
|
| 1281 |
+
"grad_norm": 4.311796188354492,
|
| 1282 |
+
"learning_rate": 1.2844444444444445e-06,
|
| 1283 |
+
"loss": 0.585,
|
| 1284 |
+
"step": 4425
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 0.89,
|
| 1288 |
+
"grad_norm": 2.739590883255005,
|
| 1289 |
+
"learning_rate": 1.228888888888889e-06,
|
| 1290 |
+
"loss": 0.5151,
|
| 1291 |
+
"step": 4450
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 0.895,
|
| 1295 |
+
"grad_norm": 2.0715537071228027,
|
| 1296 |
+
"learning_rate": 1.1733333333333335e-06,
|
| 1297 |
+
"loss": 0.2399,
|
| 1298 |
+
"step": 4475
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 0.9,
|
| 1302 |
+
"grad_norm": 1.7919174432754517,
|
| 1303 |
+
"learning_rate": 1.117777777777778e-06,
|
| 1304 |
+
"loss": 0.1013,
|
| 1305 |
+
"step": 4500
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 0.905,
|
| 1309 |
+
"grad_norm": 1.8934131860733032,
|
| 1310 |
+
"learning_rate": 1.0622222222222222e-06,
|
| 1311 |
+
"loss": 0.0792,
|
| 1312 |
+
"step": 4525
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 0.91,
|
| 1316 |
+
"grad_norm": 1.463447093963623,
|
| 1317 |
+
"learning_rate": 1.0066666666666668e-06,
|
| 1318 |
+
"loss": 0.0594,
|
| 1319 |
+
"step": 4550
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 0.915,
|
| 1323 |
+
"grad_norm": 2.0955190658569336,
|
| 1324 |
+
"learning_rate": 9.511111111111111e-07,
|
| 1325 |
+
"loss": 0.0641,
|
| 1326 |
+
"step": 4575
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 0.92,
|
| 1330 |
+
"grad_norm": 1.8596012592315674,
|
| 1331 |
+
"learning_rate": 8.955555555555557e-07,
|
| 1332 |
+
"loss": 0.065,
|
| 1333 |
+
"step": 4600
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 0.925,
|
| 1337 |
+
"grad_norm": 2.5963196754455566,
|
| 1338 |
+
"learning_rate": 8.400000000000001e-07,
|
| 1339 |
+
"loss": 0.0604,
|
| 1340 |
+
"step": 4625
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 0.93,
|
| 1344 |
+
"grad_norm": 1.361854076385498,
|
| 1345 |
+
"learning_rate": 7.844444444444445e-07,
|
| 1346 |
+
"loss": 0.0695,
|
| 1347 |
+
"step": 4650
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 0.935,
|
| 1351 |
+
"grad_norm": 1.8713892698287964,
|
| 1352 |
+
"learning_rate": 7.28888888888889e-07,
|
| 1353 |
+
"loss": 0.0501,
|
| 1354 |
+
"step": 4675
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 1.0022,
|
| 1358 |
+
"grad_norm": 1.7594069242477417,
|
| 1359 |
+
"learning_rate": 6.755555555555555e-07,
|
| 1360 |
+
"loss": 0.0811,
|
| 1361 |
+
"step": 4700
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 1.0072,
|
| 1365 |
+
"grad_norm": 2.2666430473327637,
|
| 1366 |
+
"learning_rate": 6.200000000000001e-07,
|
| 1367 |
+
"loss": 0.0794,
|
| 1368 |
+
"step": 4725
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 1.0122,
|
| 1372 |
+
"grad_norm": 1.4609644412994385,
|
| 1373 |
+
"learning_rate": 5.644444444444445e-07,
|
| 1374 |
+
"loss": 0.0913,
|
| 1375 |
+
"step": 4750
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 1.0172,
|
| 1379 |
+
"grad_norm": 2.5821077823638916,
|
| 1380 |
+
"learning_rate": 5.088888888888889e-07,
|
| 1381 |
+
"loss": 0.0743,
|
| 1382 |
+
"step": 4775
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 1.0222,
|
| 1386 |
+
"grad_norm": 3.746551036834717,
|
| 1387 |
+
"learning_rate": 4.533333333333334e-07,
|
| 1388 |
+
"loss": 0.2048,
|
| 1389 |
+
"step": 4800
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 1.0272,
|
| 1393 |
+
"grad_norm": 4.535083293914795,
|
| 1394 |
+
"learning_rate": 3.9777777777777783e-07,
|
| 1395 |
+
"loss": 0.4846,
|
| 1396 |
+
"step": 4825
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 1.0322,
|
| 1400 |
+
"grad_norm": 4.492420196533203,
|
| 1401 |
+
"learning_rate": 3.422222222222223e-07,
|
| 1402 |
+
"loss": 0.3757,
|
| 1403 |
+
"step": 4850
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 1.0372,
|
| 1407 |
+
"grad_norm": 3.43796706199646,
|
| 1408 |
+
"learning_rate": 2.866666666666667e-07,
|
| 1409 |
+
"loss": 0.2405,
|
| 1410 |
+
"step": 4875
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 1.0422,
|
| 1414 |
+
"grad_norm": 3.234219551086426,
|
| 1415 |
+
"learning_rate": 2.3111111111111112e-07,
|
| 1416 |
+
"loss": 0.1906,
|
| 1417 |
+
"step": 4900
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 1.0472,
|
| 1421 |
+
"grad_norm": 2.204688549041748,
|
| 1422 |
+
"learning_rate": 1.7555555555555558e-07,
|
| 1423 |
+
"loss": 0.1335,
|
| 1424 |
+
"step": 4925
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 1.0522,
|
| 1428 |
+
"grad_norm": 3.1717331409454346,
|
| 1429 |
+
"learning_rate": 1.2000000000000002e-07,
|
| 1430 |
+
"loss": 0.1198,
|
| 1431 |
+
"step": 4950
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 1.0572,
|
| 1435 |
+
"grad_norm": 2.4640631675720215,
|
| 1436 |
+
"learning_rate": 6.444444444444445e-08,
|
| 1437 |
+
"loss": 0.106,
|
| 1438 |
+
"step": 4975
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 1.0622,
|
| 1442 |
+
"grad_norm": 2.141594648361206,
|
| 1443 |
+
"learning_rate": 8.88888888888889e-09,
|
| 1444 |
+
"loss": 0.0977,
|
| 1445 |
+
"step": 5000
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 1.0622,
|
| 1449 |
+
"eval_loss": 0.13527874648571014,
|
| 1450 |
+
"eval_runtime": 940.5129,
|
| 1451 |
+
"eval_samples_per_second": 10.066,
|
| 1452 |
+
"eval_steps_per_second": 1.259,
|
| 1453 |
+
"eval_wer": 7.122989865404904,
|
| 1454 |
+
"step": 5000
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 1.0622,
|
| 1458 |
+
"step": 5000,
|
| 1459 |
+
"total_flos": 1.63271270080512e+20,
|
| 1460 |
+
"train_loss": 0.15365941531658173,
|
| 1461 |
+
"train_runtime": 18539.6479,
|
| 1462 |
+
"train_samples_per_second": 8.63,
|
| 1463 |
+
"train_steps_per_second": 0.27
|
| 1464 |
+
}
|
| 1465 |
+
],
|
| 1466 |
+
"logging_steps": 25,
|
| 1467 |
+
"max_steps": 5000,
|
| 1468 |
+
"num_input_tokens_seen": 0,
|
| 1469 |
+
"num_train_epochs": 9223372036854775807,
|
| 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": 1.63271270080512e+20,
|
| 1484 |
+
"train_batch_size": 32,
|
| 1485 |
+
"trial_name": null,
|
| 1486 |
+
"trial_params": null
|
| 1487 |
+
}
|