End of training
Browse files- README.md +11 -8
- all_results.json +12 -0
- eval_results.json +8 -0
- runs/Jun29_10-06-51_durunda/events.out.tfevents.1719659156.durunda +3 -0
- train_results.json +7 -0
- trainer_state.json +1275 -0
README.md
CHANGED
|
@@ -1,39 +1,42 @@
|
|
| 1 |
---
|
|
|
|
|
|
|
| 2 |
license: apache-2.0
|
| 3 |
base_model: openai/whisper-base
|
| 4 |
tags:
|
|
|
|
| 5 |
- generated_from_trainer
|
| 6 |
datasets:
|
| 7 |
-
- common_voice_13_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_13_0
|
| 18 |
-
type: common_voice_13_0
|
| 19 |
config: pt
|
| 20 |
split: test
|
| 21 |
args: pt
|
| 22 |
metrics:
|
| 23 |
- name: Wer
|
| 24 |
type: wer
|
| 25 |
-
value: 19.
|
| 26 |
---
|
| 27 |
|
| 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-base](https://huggingface.co/openai/whisper-base) on the common_voice_13_0 dataset.
|
| 34 |
It achieves the following results on the evaluation set:
|
| 35 |
- Loss: 0.3815
|
| 36 |
-
- Wer: 19.
|
| 37 |
|
| 38 |
## Model description
|
| 39 |
|
|
|
|
| 1 |
---
|
| 2 |
+
language:
|
| 3 |
+
- pt
|
| 4 |
license: apache-2.0
|
| 5 |
base_model: openai/whisper-base
|
| 6 |
tags:
|
| 7 |
+
- whisper-event
|
| 8 |
- generated_from_trainer
|
| 9 |
datasets:
|
| 10 |
+
- mozilla-foundation/common_voice_13_0
|
| 11 |
metrics:
|
| 12 |
- wer
|
| 13 |
model-index:
|
| 14 |
+
- name: Whisper Base Portuguese
|
| 15 |
results:
|
| 16 |
- task:
|
| 17 |
name: Automatic Speech Recognition
|
| 18 |
type: automatic-speech-recognition
|
| 19 |
dataset:
|
| 20 |
+
name: mozilla-foundation/common_voice_13_0 pt
|
| 21 |
+
type: mozilla-foundation/common_voice_13_0
|
| 22 |
config: pt
|
| 23 |
split: test
|
| 24 |
args: pt
|
| 25 |
metrics:
|
| 26 |
- name: Wer
|
| 27 |
type: wer
|
| 28 |
+
value: 19.28991555219663
|
| 29 |
---
|
| 30 |
|
| 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 Base Portuguese
|
| 35 |
|
| 36 |
+
This model is a fine-tuned version of [openai/whisper-base](https://huggingface.co/openai/whisper-base) on the mozilla-foundation/common_voice_13_0 pt dataset.
|
| 37 |
It achieves the following results on the evaluation set:
|
| 38 |
- Loss: 0.3815
|
| 39 |
+
- Wer: 19.2899
|
| 40 |
|
| 41 |
## Model description
|
| 42 |
|
all_results.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 35.21,
|
| 3 |
+
"eval_loss": 0.38145098090171814,
|
| 4 |
+
"eval_runtime": 138.9291,
|
| 5 |
+
"eval_samples_per_second": 65.3,
|
| 6 |
+
"eval_steps_per_second": 1.022,
|
| 7 |
+
"eval_wer": 19.28991555219663,
|
| 8 |
+
"train_loss": 0.30019377393722535,
|
| 9 |
+
"train_runtime": 10549.3759,
|
| 10 |
+
"train_samples_per_second": 60.667,
|
| 11 |
+
"train_steps_per_second": 0.474
|
| 12 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 35.21,
|
| 3 |
+
"eval_loss": 0.38145098090171814,
|
| 4 |
+
"eval_runtime": 138.9291,
|
| 5 |
+
"eval_samples_per_second": 65.3,
|
| 6 |
+
"eval_steps_per_second": 1.022,
|
| 7 |
+
"eval_wer": 19.28991555219663
|
| 8 |
+
}
|
runs/Jun29_10-06-51_durunda/events.out.tfevents.1719659156.durunda
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fa93ff8f4155d3476d757f893a4030ec19235b9da2828577c930bd88bf14e9c6
|
| 3 |
+
size 40
|
train_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 35.21,
|
| 3 |
+
"train_loss": 0.30019377393722535,
|
| 4 |
+
"train_runtime": 10549.3759,
|
| 5 |
+
"train_samples_per_second": 60.667,
|
| 6 |
+
"train_steps_per_second": 0.474
|
| 7 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1275 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 19.28991555219663,
|
| 3 |
+
"best_model_checkpoint": "./checkpoint-4000",
|
| 4 |
+
"epoch": 35.2112676056338,
|
| 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": 0.18,
|
| 13 |
+
"learning_rate": 4.4e-08,
|
| 14 |
+
"loss": 1.2187,
|
| 15 |
+
"step": 25
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.35,
|
| 19 |
+
"learning_rate": 9.4e-08,
|
| 20 |
+
"loss": 1.2095,
|
| 21 |
+
"step": 50
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.53,
|
| 25 |
+
"learning_rate": 1.44e-07,
|
| 26 |
+
"loss": 1.1872,
|
| 27 |
+
"step": 75
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.7,
|
| 31 |
+
"learning_rate": 1.94e-07,
|
| 32 |
+
"loss": 1.1709,
|
| 33 |
+
"step": 100
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.88,
|
| 37 |
+
"learning_rate": 2.4399999999999996e-07,
|
| 38 |
+
"loss": 1.1096,
|
| 39 |
+
"step": 125
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 1.06,
|
| 43 |
+
"learning_rate": 2.9399999999999996e-07,
|
| 44 |
+
"loss": 1.0602,
|
| 45 |
+
"step": 150
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 1.23,
|
| 49 |
+
"learning_rate": 3.4399999999999996e-07,
|
| 50 |
+
"loss": 0.9914,
|
| 51 |
+
"step": 175
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 1.41,
|
| 55 |
+
"learning_rate": 3.94e-07,
|
| 56 |
+
"loss": 0.9003,
|
| 57 |
+
"step": 200
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 1.58,
|
| 61 |
+
"learning_rate": 4.44e-07,
|
| 62 |
+
"loss": 0.8196,
|
| 63 |
+
"step": 225
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 1.76,
|
| 67 |
+
"learning_rate": 4.94e-07,
|
| 68 |
+
"loss": 0.6752,
|
| 69 |
+
"step": 250
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 1.94,
|
| 73 |
+
"learning_rate": 5.44e-07,
|
| 74 |
+
"loss": 0.5487,
|
| 75 |
+
"step": 275
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 2.11,
|
| 79 |
+
"learning_rate": 5.939999999999999e-07,
|
| 80 |
+
"loss": 0.5037,
|
| 81 |
+
"step": 300
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 2.29,
|
| 85 |
+
"learning_rate": 6.44e-07,
|
| 86 |
+
"loss": 0.4675,
|
| 87 |
+
"step": 325
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 2.46,
|
| 91 |
+
"learning_rate": 6.939999999999999e-07,
|
| 92 |
+
"loss": 0.4716,
|
| 93 |
+
"step": 350
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 2.64,
|
| 97 |
+
"learning_rate": 7.44e-07,
|
| 98 |
+
"loss": 0.4602,
|
| 99 |
+
"step": 375
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 2.82,
|
| 103 |
+
"learning_rate": 7.94e-07,
|
| 104 |
+
"loss": 0.432,
|
| 105 |
+
"step": 400
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 2.99,
|
| 109 |
+
"learning_rate": 8.439999999999999e-07,
|
| 110 |
+
"loss": 0.453,
|
| 111 |
+
"step": 425
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 3.17,
|
| 115 |
+
"learning_rate": 8.939999999999999e-07,
|
| 116 |
+
"loss": 0.4161,
|
| 117 |
+
"step": 450
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 3.35,
|
| 121 |
+
"learning_rate": 9.439999999999999e-07,
|
| 122 |
+
"loss": 0.4119,
|
| 123 |
+
"step": 475
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 3.52,
|
| 127 |
+
"learning_rate": 9.94e-07,
|
| 128 |
+
"loss": 0.4225,
|
| 129 |
+
"step": 500
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 3.7,
|
| 133 |
+
"learning_rate": 9.95111111111111e-07,
|
| 134 |
+
"loss": 0.4105,
|
| 135 |
+
"step": 525
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 3.87,
|
| 139 |
+
"learning_rate": 9.895555555555554e-07,
|
| 140 |
+
"loss": 0.4202,
|
| 141 |
+
"step": 550
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 4.05,
|
| 145 |
+
"learning_rate": 9.84e-07,
|
| 146 |
+
"loss": 0.3985,
|
| 147 |
+
"step": 575
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 4.23,
|
| 151 |
+
"learning_rate": 9.784444444444444e-07,
|
| 152 |
+
"loss": 0.3901,
|
| 153 |
+
"step": 600
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 4.4,
|
| 157 |
+
"learning_rate": 9.728888888888888e-07,
|
| 158 |
+
"loss": 0.3751,
|
| 159 |
+
"step": 625
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 4.58,
|
| 163 |
+
"learning_rate": 9.673333333333332e-07,
|
| 164 |
+
"loss": 0.3857,
|
| 165 |
+
"step": 650
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 4.75,
|
| 169 |
+
"learning_rate": 9.617777777777776e-07,
|
| 170 |
+
"loss": 0.3814,
|
| 171 |
+
"step": 675
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 4.93,
|
| 175 |
+
"learning_rate": 9.562222222222223e-07,
|
| 176 |
+
"loss": 0.3884,
|
| 177 |
+
"step": 700
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 5.11,
|
| 181 |
+
"learning_rate": 9.506666666666667e-07,
|
| 182 |
+
"loss": 0.3731,
|
| 183 |
+
"step": 725
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 5.28,
|
| 187 |
+
"learning_rate": 9.451111111111111e-07,
|
| 188 |
+
"loss": 0.3576,
|
| 189 |
+
"step": 750
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 5.46,
|
| 193 |
+
"learning_rate": 9.395555555555556e-07,
|
| 194 |
+
"loss": 0.3611,
|
| 195 |
+
"step": 775
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 5.63,
|
| 199 |
+
"learning_rate": 9.34e-07,
|
| 200 |
+
"loss": 0.3531,
|
| 201 |
+
"step": 800
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 5.81,
|
| 205 |
+
"learning_rate": 9.284444444444444e-07,
|
| 206 |
+
"loss": 0.3601,
|
| 207 |
+
"step": 825
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 5.99,
|
| 211 |
+
"learning_rate": 9.228888888888888e-07,
|
| 212 |
+
"loss": 0.3636,
|
| 213 |
+
"step": 850
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 6.16,
|
| 217 |
+
"learning_rate": 9.173333333333333e-07,
|
| 218 |
+
"loss": 0.3504,
|
| 219 |
+
"step": 875
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 6.34,
|
| 223 |
+
"learning_rate": 9.117777777777778e-07,
|
| 224 |
+
"loss": 0.3394,
|
| 225 |
+
"step": 900
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 6.51,
|
| 229 |
+
"learning_rate": 9.062222222222222e-07,
|
| 230 |
+
"loss": 0.3536,
|
| 231 |
+
"step": 925
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 6.69,
|
| 235 |
+
"learning_rate": 9.006666666666666e-07,
|
| 236 |
+
"loss": 0.345,
|
| 237 |
+
"step": 950
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 6.87,
|
| 241 |
+
"learning_rate": 8.95111111111111e-07,
|
| 242 |
+
"loss": 0.3403,
|
| 243 |
+
"step": 975
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 7.04,
|
| 247 |
+
"learning_rate": 8.895555555555555e-07,
|
| 248 |
+
"loss": 0.3261,
|
| 249 |
+
"step": 1000
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 7.04,
|
| 253 |
+
"eval_loss": 0.40974706411361694,
|
| 254 |
+
"eval_runtime": 301.6855,
|
| 255 |
+
"eval_samples_per_second": 30.071,
|
| 256 |
+
"eval_steps_per_second": 0.471,
|
| 257 |
+
"eval_wer": 20.676568198994513,
|
| 258 |
+
"step": 1000
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 7.22,
|
| 262 |
+
"learning_rate": 8.839999999999999e-07,
|
| 263 |
+
"loss": 0.3246,
|
| 264 |
+
"step": 1025
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 7.39,
|
| 268 |
+
"learning_rate": 8.784444444444444e-07,
|
| 269 |
+
"loss": 0.3261,
|
| 270 |
+
"step": 1050
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 7.57,
|
| 274 |
+
"learning_rate": 8.728888888888889e-07,
|
| 275 |
+
"loss": 0.3307,
|
| 276 |
+
"step": 1075
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 7.75,
|
| 280 |
+
"learning_rate": 8.673333333333332e-07,
|
| 281 |
+
"loss": 0.3244,
|
| 282 |
+
"step": 1100
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 7.92,
|
| 286 |
+
"learning_rate": 8.617777777777777e-07,
|
| 287 |
+
"loss": 0.3277,
|
| 288 |
+
"step": 1125
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 8.1,
|
| 292 |
+
"learning_rate": 8.562222222222222e-07,
|
| 293 |
+
"loss": 0.3232,
|
| 294 |
+
"step": 1150
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 8.27,
|
| 298 |
+
"learning_rate": 8.506666666666667e-07,
|
| 299 |
+
"loss": 0.3197,
|
| 300 |
+
"step": 1175
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 8.45,
|
| 304 |
+
"learning_rate": 8.451111111111111e-07,
|
| 305 |
+
"loss": 0.3073,
|
| 306 |
+
"step": 1200
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 8.63,
|
| 310 |
+
"learning_rate": 8.395555555555556e-07,
|
| 311 |
+
"loss": 0.3146,
|
| 312 |
+
"step": 1225
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 8.8,
|
| 316 |
+
"learning_rate": 8.34e-07,
|
| 317 |
+
"loss": 0.3195,
|
| 318 |
+
"step": 1250
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 8.98,
|
| 322 |
+
"learning_rate": 8.284444444444444e-07,
|
| 323 |
+
"loss": 0.3117,
|
| 324 |
+
"step": 1275
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 9.15,
|
| 328 |
+
"learning_rate": 8.228888888888889e-07,
|
| 329 |
+
"loss": 0.3093,
|
| 330 |
+
"step": 1300
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 9.33,
|
| 334 |
+
"learning_rate": 8.173333333333333e-07,
|
| 335 |
+
"loss": 0.2997,
|
| 336 |
+
"step": 1325
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 9.51,
|
| 340 |
+
"learning_rate": 8.117777777777778e-07,
|
| 341 |
+
"loss": 0.3,
|
| 342 |
+
"step": 1350
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 9.68,
|
| 346 |
+
"learning_rate": 8.062222222222221e-07,
|
| 347 |
+
"loss": 0.3048,
|
| 348 |
+
"step": 1375
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 9.86,
|
| 352 |
+
"learning_rate": 8.006666666666666e-07,
|
| 353 |
+
"loss": 0.3,
|
| 354 |
+
"step": 1400
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 10.04,
|
| 358 |
+
"learning_rate": 7.95111111111111e-07,
|
| 359 |
+
"loss": 0.3062,
|
| 360 |
+
"step": 1425
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 10.21,
|
| 364 |
+
"learning_rate": 7.895555555555555e-07,
|
| 365 |
+
"loss": 0.2855,
|
| 366 |
+
"step": 1450
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 10.39,
|
| 370 |
+
"learning_rate": 7.84e-07,
|
| 371 |
+
"loss": 0.2954,
|
| 372 |
+
"step": 1475
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 10.56,
|
| 376 |
+
"learning_rate": 7.784444444444444e-07,
|
| 377 |
+
"loss": 0.2931,
|
| 378 |
+
"step": 1500
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 10.74,
|
| 382 |
+
"learning_rate": 7.728888888888888e-07,
|
| 383 |
+
"loss": 0.2873,
|
| 384 |
+
"step": 1525
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 10.92,
|
| 388 |
+
"learning_rate": 7.673333333333332e-07,
|
| 389 |
+
"loss": 0.2949,
|
| 390 |
+
"step": 1550
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 11.09,
|
| 394 |
+
"learning_rate": 7.617777777777778e-07,
|
| 395 |
+
"loss": 0.2882,
|
| 396 |
+
"step": 1575
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 11.27,
|
| 400 |
+
"learning_rate": 7.562222222222222e-07,
|
| 401 |
+
"loss": 0.2797,
|
| 402 |
+
"step": 1600
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 11.44,
|
| 406 |
+
"learning_rate": 7.506666666666667e-07,
|
| 407 |
+
"loss": 0.281,
|
| 408 |
+
"step": 1625
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 11.62,
|
| 412 |
+
"learning_rate": 7.451111111111111e-07,
|
| 413 |
+
"loss": 0.2901,
|
| 414 |
+
"step": 1650
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 11.8,
|
| 418 |
+
"learning_rate": 7.395555555555555e-07,
|
| 419 |
+
"loss": 0.2768,
|
| 420 |
+
"step": 1675
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 11.97,
|
| 424 |
+
"learning_rate": 7.34e-07,
|
| 425 |
+
"loss": 0.284,
|
| 426 |
+
"step": 1700
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 12.15,
|
| 430 |
+
"learning_rate": 7.284444444444444e-07,
|
| 431 |
+
"loss": 0.2822,
|
| 432 |
+
"step": 1725
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 12.32,
|
| 436 |
+
"learning_rate": 7.228888888888889e-07,
|
| 437 |
+
"loss": 0.2735,
|
| 438 |
+
"step": 1750
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 12.5,
|
| 442 |
+
"learning_rate": 7.173333333333333e-07,
|
| 443 |
+
"loss": 0.2735,
|
| 444 |
+
"step": 1775
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 12.68,
|
| 448 |
+
"learning_rate": 7.117777777777777e-07,
|
| 449 |
+
"loss": 0.2736,
|
| 450 |
+
"step": 1800
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 12.85,
|
| 454 |
+
"learning_rate": 7.062222222222222e-07,
|
| 455 |
+
"loss": 0.2715,
|
| 456 |
+
"step": 1825
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 13.03,
|
| 460 |
+
"learning_rate": 7.006666666666666e-07,
|
| 461 |
+
"loss": 0.2722,
|
| 462 |
+
"step": 1850
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"epoch": 13.2,
|
| 466 |
+
"learning_rate": 6.951111111111111e-07,
|
| 467 |
+
"loss": 0.2666,
|
| 468 |
+
"step": 1875
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 13.38,
|
| 472 |
+
"learning_rate": 6.895555555555555e-07,
|
| 473 |
+
"loss": 0.2716,
|
| 474 |
+
"step": 1900
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 13.56,
|
| 478 |
+
"learning_rate": 6.84e-07,
|
| 479 |
+
"loss": 0.2628,
|
| 480 |
+
"step": 1925
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 13.73,
|
| 484 |
+
"learning_rate": 6.784444444444443e-07,
|
| 485 |
+
"loss": 0.2654,
|
| 486 |
+
"step": 1950
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 13.91,
|
| 490 |
+
"learning_rate": 6.728888888888888e-07,
|
| 491 |
+
"loss": 0.2616,
|
| 492 |
+
"step": 1975
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 14.08,
|
| 496 |
+
"learning_rate": 6.673333333333334e-07,
|
| 497 |
+
"loss": 0.2632,
|
| 498 |
+
"step": 2000
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 14.08,
|
| 502 |
+
"eval_loss": 0.3883862793445587,
|
| 503 |
+
"eval_runtime": 138.6173,
|
| 504 |
+
"eval_samples_per_second": 65.446,
|
| 505 |
+
"eval_steps_per_second": 1.024,
|
| 506 |
+
"eval_wer": 19.51007130417639,
|
| 507 |
+
"step": 2000
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 14.26,
|
| 511 |
+
"learning_rate": 6.617777777777778e-07,
|
| 512 |
+
"loss": 0.259,
|
| 513 |
+
"step": 2025
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 14.44,
|
| 517 |
+
"learning_rate": 6.562222222222223e-07,
|
| 518 |
+
"loss": 0.2604,
|
| 519 |
+
"step": 2050
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 14.61,
|
| 523 |
+
"learning_rate": 6.506666666666666e-07,
|
| 524 |
+
"loss": 0.2608,
|
| 525 |
+
"step": 2075
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 14.79,
|
| 529 |
+
"learning_rate": 6.451111111111111e-07,
|
| 530 |
+
"loss": 0.2545,
|
| 531 |
+
"step": 2100
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 14.96,
|
| 535 |
+
"learning_rate": 6.395555555555555e-07,
|
| 536 |
+
"loss": 0.256,
|
| 537 |
+
"step": 2125
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 15.14,
|
| 541 |
+
"learning_rate": 6.34e-07,
|
| 542 |
+
"loss": 0.2562,
|
| 543 |
+
"step": 2150
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 15.32,
|
| 547 |
+
"learning_rate": 6.284444444444445e-07,
|
| 548 |
+
"loss": 0.251,
|
| 549 |
+
"step": 2175
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 15.49,
|
| 553 |
+
"learning_rate": 6.228888888888889e-07,
|
| 554 |
+
"loss": 0.2457,
|
| 555 |
+
"step": 2200
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 15.67,
|
| 559 |
+
"learning_rate": 6.173333333333333e-07,
|
| 560 |
+
"loss": 0.2542,
|
| 561 |
+
"step": 2225
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 15.85,
|
| 565 |
+
"learning_rate": 6.117777777777777e-07,
|
| 566 |
+
"loss": 0.2568,
|
| 567 |
+
"step": 2250
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 16.02,
|
| 571 |
+
"learning_rate": 6.062222222222222e-07,
|
| 572 |
+
"loss": 0.2504,
|
| 573 |
+
"step": 2275
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 16.2,
|
| 577 |
+
"learning_rate": 6.006666666666666e-07,
|
| 578 |
+
"loss": 0.2415,
|
| 579 |
+
"step": 2300
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 16.37,
|
| 583 |
+
"learning_rate": 5.951111111111111e-07,
|
| 584 |
+
"loss": 0.247,
|
| 585 |
+
"step": 2325
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 16.55,
|
| 589 |
+
"learning_rate": 5.895555555555555e-07,
|
| 590 |
+
"loss": 0.2471,
|
| 591 |
+
"step": 2350
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 16.73,
|
| 595 |
+
"learning_rate": 5.839999999999999e-07,
|
| 596 |
+
"loss": 0.2474,
|
| 597 |
+
"step": 2375
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 16.9,
|
| 601 |
+
"learning_rate": 5.784444444444444e-07,
|
| 602 |
+
"loss": 0.2483,
|
| 603 |
+
"step": 2400
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 17.08,
|
| 607 |
+
"learning_rate": 5.728888888888888e-07,
|
| 608 |
+
"loss": 0.2423,
|
| 609 |
+
"step": 2425
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 17.25,
|
| 613 |
+
"learning_rate": 5.673333333333334e-07,
|
| 614 |
+
"loss": 0.2427,
|
| 615 |
+
"step": 2450
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 17.43,
|
| 619 |
+
"learning_rate": 5.617777777777778e-07,
|
| 620 |
+
"loss": 0.242,
|
| 621 |
+
"step": 2475
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 17.61,
|
| 625 |
+
"learning_rate": 5.562222222222222e-07,
|
| 626 |
+
"loss": 0.2438,
|
| 627 |
+
"step": 2500
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 17.78,
|
| 631 |
+
"learning_rate": 5.506666666666666e-07,
|
| 632 |
+
"loss": 0.2319,
|
| 633 |
+
"step": 2525
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 17.96,
|
| 637 |
+
"learning_rate": 5.451111111111111e-07,
|
| 638 |
+
"loss": 0.2404,
|
| 639 |
+
"step": 2550
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 18.13,
|
| 643 |
+
"learning_rate": 5.395555555555556e-07,
|
| 644 |
+
"loss": 0.2331,
|
| 645 |
+
"step": 2575
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 18.31,
|
| 649 |
+
"learning_rate": 5.34e-07,
|
| 650 |
+
"loss": 0.2315,
|
| 651 |
+
"step": 2600
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 18.49,
|
| 655 |
+
"learning_rate": 5.284444444444445e-07,
|
| 656 |
+
"loss": 0.2337,
|
| 657 |
+
"step": 2625
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 18.66,
|
| 661 |
+
"learning_rate": 5.228888888888888e-07,
|
| 662 |
+
"loss": 0.2412,
|
| 663 |
+
"step": 2650
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 18.84,
|
| 667 |
+
"learning_rate": 5.173333333333333e-07,
|
| 668 |
+
"loss": 0.2362,
|
| 669 |
+
"step": 2675
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 19.01,
|
| 673 |
+
"learning_rate": 5.117777777777777e-07,
|
| 674 |
+
"loss": 0.2353,
|
| 675 |
+
"step": 2700
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 19.19,
|
| 679 |
+
"learning_rate": 5.062222222222222e-07,
|
| 680 |
+
"loss": 0.2287,
|
| 681 |
+
"step": 2725
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 19.37,
|
| 685 |
+
"learning_rate": 5.006666666666667e-07,
|
| 686 |
+
"loss": 0.2265,
|
| 687 |
+
"step": 2750
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 19.54,
|
| 691 |
+
"learning_rate": 4.951111111111111e-07,
|
| 692 |
+
"loss": 0.2387,
|
| 693 |
+
"step": 2775
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 19.72,
|
| 697 |
+
"learning_rate": 4.895555555555555e-07,
|
| 698 |
+
"loss": 0.2288,
|
| 699 |
+
"step": 2800
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 19.89,
|
| 703 |
+
"learning_rate": 4.839999999999999e-07,
|
| 704 |
+
"loss": 0.2281,
|
| 705 |
+
"step": 2825
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 20.07,
|
| 709 |
+
"learning_rate": 4.784444444444444e-07,
|
| 710 |
+
"loss": 0.2283,
|
| 711 |
+
"step": 2850
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 20.25,
|
| 715 |
+
"learning_rate": 4.728888888888889e-07,
|
| 716 |
+
"loss": 0.2252,
|
| 717 |
+
"step": 2875
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 20.42,
|
| 721 |
+
"learning_rate": 4.673333333333333e-07,
|
| 722 |
+
"loss": 0.228,
|
| 723 |
+
"step": 2900
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 20.6,
|
| 727 |
+
"learning_rate": 4.6177777777777777e-07,
|
| 728 |
+
"loss": 0.2246,
|
| 729 |
+
"step": 2925
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 20.77,
|
| 733 |
+
"learning_rate": 4.5622222222222217e-07,
|
| 734 |
+
"loss": 0.2249,
|
| 735 |
+
"step": 2950
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 20.95,
|
| 739 |
+
"learning_rate": 4.506666666666666e-07,
|
| 740 |
+
"loss": 0.2245,
|
| 741 |
+
"step": 2975
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 21.13,
|
| 745 |
+
"learning_rate": 4.451111111111111e-07,
|
| 746 |
+
"loss": 0.2241,
|
| 747 |
+
"step": 3000
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 21.13,
|
| 751 |
+
"eval_loss": 0.3826531767845154,
|
| 752 |
+
"eval_runtime": 141.4024,
|
| 753 |
+
"eval_samples_per_second": 64.157,
|
| 754 |
+
"eval_steps_per_second": 1.004,
|
| 755 |
+
"eval_wer": 19.468997469851807,
|
| 756 |
+
"step": 3000
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 21.3,
|
| 760 |
+
"learning_rate": 4.3955555555555554e-07,
|
| 761 |
+
"loss": 0.2228,
|
| 762 |
+
"step": 3025
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 21.48,
|
| 766 |
+
"learning_rate": 4.34e-07,
|
| 767 |
+
"loss": 0.2223,
|
| 768 |
+
"step": 3050
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 21.65,
|
| 772 |
+
"learning_rate": 4.2844444444444445e-07,
|
| 773 |
+
"loss": 0.2263,
|
| 774 |
+
"step": 3075
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 21.83,
|
| 778 |
+
"learning_rate": 4.2288888888888886e-07,
|
| 779 |
+
"loss": 0.2201,
|
| 780 |
+
"step": 3100
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 22.01,
|
| 784 |
+
"learning_rate": 4.173333333333333e-07,
|
| 785 |
+
"loss": 0.2166,
|
| 786 |
+
"step": 3125
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 22.18,
|
| 790 |
+
"learning_rate": 4.1177777777777777e-07,
|
| 791 |
+
"loss": 0.2212,
|
| 792 |
+
"step": 3150
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 22.36,
|
| 796 |
+
"learning_rate": 4.0622222222222217e-07,
|
| 797 |
+
"loss": 0.2161,
|
| 798 |
+
"step": 3175
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 22.54,
|
| 802 |
+
"learning_rate": 4.0066666666666663e-07,
|
| 803 |
+
"loss": 0.2116,
|
| 804 |
+
"step": 3200
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 22.71,
|
| 808 |
+
"learning_rate": 3.9511111111111114e-07,
|
| 809 |
+
"loss": 0.2204,
|
| 810 |
+
"step": 3225
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 22.89,
|
| 814 |
+
"learning_rate": 3.8955555555555554e-07,
|
| 815 |
+
"loss": 0.2173,
|
| 816 |
+
"step": 3250
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 23.06,
|
| 820 |
+
"learning_rate": 3.84e-07,
|
| 821 |
+
"loss": 0.2238,
|
| 822 |
+
"step": 3275
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 23.24,
|
| 826 |
+
"learning_rate": 3.7844444444444445e-07,
|
| 827 |
+
"loss": 0.2102,
|
| 828 |
+
"step": 3300
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 23.42,
|
| 832 |
+
"learning_rate": 3.7288888888888886e-07,
|
| 833 |
+
"loss": 0.2184,
|
| 834 |
+
"step": 3325
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 23.59,
|
| 838 |
+
"learning_rate": 3.673333333333333e-07,
|
| 839 |
+
"loss": 0.2131,
|
| 840 |
+
"step": 3350
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 23.77,
|
| 844 |
+
"learning_rate": 3.617777777777777e-07,
|
| 845 |
+
"loss": 0.2123,
|
| 846 |
+
"step": 3375
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 23.94,
|
| 850 |
+
"learning_rate": 3.5622222222222223e-07,
|
| 851 |
+
"loss": 0.2151,
|
| 852 |
+
"step": 3400
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 24.12,
|
| 856 |
+
"learning_rate": 3.506666666666667e-07,
|
| 857 |
+
"loss": 0.2117,
|
| 858 |
+
"step": 3425
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 24.3,
|
| 862 |
+
"learning_rate": 3.451111111111111e-07,
|
| 863 |
+
"loss": 0.2127,
|
| 864 |
+
"step": 3450
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 24.47,
|
| 868 |
+
"learning_rate": 3.3955555555555554e-07,
|
| 869 |
+
"loss": 0.211,
|
| 870 |
+
"step": 3475
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 24.65,
|
| 874 |
+
"learning_rate": 3.34e-07,
|
| 875 |
+
"loss": 0.2037,
|
| 876 |
+
"step": 3500
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 24.82,
|
| 880 |
+
"learning_rate": 3.284444444444444e-07,
|
| 881 |
+
"loss": 0.211,
|
| 882 |
+
"step": 3525
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 25.0,
|
| 886 |
+
"learning_rate": 3.2288888888888886e-07,
|
| 887 |
+
"loss": 0.2169,
|
| 888 |
+
"step": 3550
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 25.18,
|
| 892 |
+
"learning_rate": 3.173333333333333e-07,
|
| 893 |
+
"loss": 0.2068,
|
| 894 |
+
"step": 3575
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 25.35,
|
| 898 |
+
"learning_rate": 3.1177777777777777e-07,
|
| 899 |
+
"loss": 0.2089,
|
| 900 |
+
"step": 3600
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 25.53,
|
| 904 |
+
"learning_rate": 3.0622222222222223e-07,
|
| 905 |
+
"loss": 0.2099,
|
| 906 |
+
"step": 3625
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 25.7,
|
| 910 |
+
"learning_rate": 3.006666666666667e-07,
|
| 911 |
+
"loss": 0.2092,
|
| 912 |
+
"step": 3650
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 25.88,
|
| 916 |
+
"learning_rate": 2.951111111111111e-07,
|
| 917 |
+
"loss": 0.2107,
|
| 918 |
+
"step": 3675
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 26.06,
|
| 922 |
+
"learning_rate": 2.8955555555555555e-07,
|
| 923 |
+
"loss": 0.2026,
|
| 924 |
+
"step": 3700
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 26.23,
|
| 928 |
+
"learning_rate": 2.8399999999999995e-07,
|
| 929 |
+
"loss": 0.2065,
|
| 930 |
+
"step": 3725
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 26.41,
|
| 934 |
+
"learning_rate": 2.784444444444444e-07,
|
| 935 |
+
"loss": 0.1987,
|
| 936 |
+
"step": 3750
|
| 937 |
+
},
|
| 938 |
+
{
|
| 939 |
+
"epoch": 26.58,
|
| 940 |
+
"learning_rate": 2.728888888888889e-07,
|
| 941 |
+
"loss": 0.2115,
|
| 942 |
+
"step": 3775
|
| 943 |
+
},
|
| 944 |
+
{
|
| 945 |
+
"epoch": 26.76,
|
| 946 |
+
"learning_rate": 2.673333333333333e-07,
|
| 947 |
+
"loss": 0.2062,
|
| 948 |
+
"step": 3800
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 26.94,
|
| 952 |
+
"learning_rate": 2.617777777777778e-07,
|
| 953 |
+
"loss": 0.2074,
|
| 954 |
+
"step": 3825
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 27.11,
|
| 958 |
+
"learning_rate": 2.5622222222222223e-07,
|
| 959 |
+
"loss": 0.2022,
|
| 960 |
+
"step": 3850
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 27.29,
|
| 964 |
+
"learning_rate": 2.5066666666666663e-07,
|
| 965 |
+
"loss": 0.2109,
|
| 966 |
+
"step": 3875
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 27.46,
|
| 970 |
+
"learning_rate": 2.451111111111111e-07,
|
| 971 |
+
"loss": 0.2001,
|
| 972 |
+
"step": 3900
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"epoch": 27.64,
|
| 976 |
+
"learning_rate": 2.3955555555555555e-07,
|
| 977 |
+
"loss": 0.2069,
|
| 978 |
+
"step": 3925
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 27.82,
|
| 982 |
+
"learning_rate": 2.34e-07,
|
| 983 |
+
"loss": 0.2032,
|
| 984 |
+
"step": 3950
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 27.99,
|
| 988 |
+
"learning_rate": 2.2844444444444443e-07,
|
| 989 |
+
"loss": 0.2013,
|
| 990 |
+
"step": 3975
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 28.17,
|
| 994 |
+
"learning_rate": 2.2288888888888886e-07,
|
| 995 |
+
"loss": 0.2048,
|
| 996 |
+
"step": 4000
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 28.17,
|
| 1000 |
+
"eval_loss": 0.38145098090171814,
|
| 1001 |
+
"eval_runtime": 140.8417,
|
| 1002 |
+
"eval_samples_per_second": 64.413,
|
| 1003 |
+
"eval_steps_per_second": 1.008,
|
| 1004 |
+
"eval_wer": 19.28991555219663,
|
| 1005 |
+
"step": 4000
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 28.35,
|
| 1009 |
+
"learning_rate": 2.1733333333333332e-07,
|
| 1010 |
+
"loss": 0.1993,
|
| 1011 |
+
"step": 4025
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 28.52,
|
| 1015 |
+
"learning_rate": 2.1177777777777778e-07,
|
| 1016 |
+
"loss": 0.2037,
|
| 1017 |
+
"step": 4050
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 28.7,
|
| 1021 |
+
"learning_rate": 2.0644444444444445e-07,
|
| 1022 |
+
"loss": 0.2028,
|
| 1023 |
+
"step": 4075
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 28.87,
|
| 1027 |
+
"learning_rate": 2.0088888888888888e-07,
|
| 1028 |
+
"loss": 0.1994,
|
| 1029 |
+
"step": 4100
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 29.05,
|
| 1033 |
+
"learning_rate": 1.953333333333333e-07,
|
| 1034 |
+
"loss": 0.1988,
|
| 1035 |
+
"step": 4125
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 29.23,
|
| 1039 |
+
"learning_rate": 1.8977777777777777e-07,
|
| 1040 |
+
"loss": 0.2023,
|
| 1041 |
+
"step": 4150
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 29.4,
|
| 1045 |
+
"learning_rate": 1.8422222222222222e-07,
|
| 1046 |
+
"loss": 0.199,
|
| 1047 |
+
"step": 4175
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 29.58,
|
| 1051 |
+
"learning_rate": 1.7866666666666665e-07,
|
| 1052 |
+
"loss": 0.205,
|
| 1053 |
+
"step": 4200
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 29.75,
|
| 1057 |
+
"learning_rate": 1.731111111111111e-07,
|
| 1058 |
+
"loss": 0.1975,
|
| 1059 |
+
"step": 4225
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 29.93,
|
| 1063 |
+
"learning_rate": 1.6755555555555556e-07,
|
| 1064 |
+
"loss": 0.1989,
|
| 1065 |
+
"step": 4250
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 30.11,
|
| 1069 |
+
"learning_rate": 1.62e-07,
|
| 1070 |
+
"loss": 0.1995,
|
| 1071 |
+
"step": 4275
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 30.28,
|
| 1075 |
+
"learning_rate": 1.5644444444444442e-07,
|
| 1076 |
+
"loss": 0.1992,
|
| 1077 |
+
"step": 4300
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 30.46,
|
| 1081 |
+
"learning_rate": 1.5088888888888888e-07,
|
| 1082 |
+
"loss": 0.2002,
|
| 1083 |
+
"step": 4325
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 30.63,
|
| 1087 |
+
"learning_rate": 1.4533333333333334e-07,
|
| 1088 |
+
"loss": 0.194,
|
| 1089 |
+
"step": 4350
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 30.81,
|
| 1093 |
+
"learning_rate": 1.3977777777777777e-07,
|
| 1094 |
+
"loss": 0.199,
|
| 1095 |
+
"step": 4375
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 30.99,
|
| 1099 |
+
"learning_rate": 1.342222222222222e-07,
|
| 1100 |
+
"loss": 0.1993,
|
| 1101 |
+
"step": 4400
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 31.16,
|
| 1105 |
+
"learning_rate": 1.2866666666666668e-07,
|
| 1106 |
+
"loss": 0.194,
|
| 1107 |
+
"step": 4425
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 31.34,
|
| 1111 |
+
"learning_rate": 1.231111111111111e-07,
|
| 1112 |
+
"loss": 0.1967,
|
| 1113 |
+
"step": 4450
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 31.51,
|
| 1117 |
+
"learning_rate": 1.1755555555555554e-07,
|
| 1118 |
+
"loss": 0.2012,
|
| 1119 |
+
"step": 4475
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 31.69,
|
| 1123 |
+
"learning_rate": 1.12e-07,
|
| 1124 |
+
"loss": 0.1941,
|
| 1125 |
+
"step": 4500
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 31.87,
|
| 1129 |
+
"learning_rate": 1.0644444444444444e-07,
|
| 1130 |
+
"loss": 0.1997,
|
| 1131 |
+
"step": 4525
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 32.04,
|
| 1135 |
+
"learning_rate": 1.0088888888888888e-07,
|
| 1136 |
+
"loss": 0.195,
|
| 1137 |
+
"step": 4550
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 32.22,
|
| 1141 |
+
"learning_rate": 9.533333333333334e-08,
|
| 1142 |
+
"loss": 0.1945,
|
| 1143 |
+
"step": 4575
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 32.39,
|
| 1147 |
+
"learning_rate": 8.977777777777777e-08,
|
| 1148 |
+
"loss": 0.1985,
|
| 1149 |
+
"step": 4600
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 32.57,
|
| 1153 |
+
"learning_rate": 8.422222222222223e-08,
|
| 1154 |
+
"loss": 0.1988,
|
| 1155 |
+
"step": 4625
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 32.75,
|
| 1159 |
+
"learning_rate": 7.866666666666666e-08,
|
| 1160 |
+
"loss": 0.2003,
|
| 1161 |
+
"step": 4650
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 32.92,
|
| 1165 |
+
"learning_rate": 7.311111111111111e-08,
|
| 1166 |
+
"loss": 0.1939,
|
| 1167 |
+
"step": 4675
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 33.1,
|
| 1171 |
+
"learning_rate": 6.755555555555554e-08,
|
| 1172 |
+
"loss": 0.2,
|
| 1173 |
+
"step": 4700
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 33.27,
|
| 1177 |
+
"learning_rate": 6.2e-08,
|
| 1178 |
+
"loss": 0.194,
|
| 1179 |
+
"step": 4725
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 33.45,
|
| 1183 |
+
"learning_rate": 5.644444444444444e-08,
|
| 1184 |
+
"loss": 0.1938,
|
| 1185 |
+
"step": 4750
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 33.63,
|
| 1189 |
+
"learning_rate": 5.0888888888888886e-08,
|
| 1190 |
+
"loss": 0.1974,
|
| 1191 |
+
"step": 4775
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 33.8,
|
| 1195 |
+
"learning_rate": 4.5333333333333336e-08,
|
| 1196 |
+
"loss": 0.1901,
|
| 1197 |
+
"step": 4800
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 33.98,
|
| 1201 |
+
"learning_rate": 3.977777777777778e-08,
|
| 1202 |
+
"loss": 0.2007,
|
| 1203 |
+
"step": 4825
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 34.15,
|
| 1207 |
+
"learning_rate": 3.422222222222222e-08,
|
| 1208 |
+
"loss": 0.1894,
|
| 1209 |
+
"step": 4850
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 34.33,
|
| 1213 |
+
"learning_rate": 2.8666666666666665e-08,
|
| 1214 |
+
"loss": 0.1919,
|
| 1215 |
+
"step": 4875
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 34.51,
|
| 1219 |
+
"learning_rate": 2.311111111111111e-08,
|
| 1220 |
+
"loss": 0.203,
|
| 1221 |
+
"step": 4900
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 34.68,
|
| 1225 |
+
"learning_rate": 1.7555555555555555e-08,
|
| 1226 |
+
"loss": 0.194,
|
| 1227 |
+
"step": 4925
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 34.86,
|
| 1231 |
+
"learning_rate": 1.2e-08,
|
| 1232 |
+
"loss": 0.2002,
|
| 1233 |
+
"step": 4950
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 35.04,
|
| 1237 |
+
"learning_rate": 6.444444444444444e-09,
|
| 1238 |
+
"loss": 0.1946,
|
| 1239 |
+
"step": 4975
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 35.21,
|
| 1243 |
+
"learning_rate": 8.888888888888889e-10,
|
| 1244 |
+
"loss": 0.1956,
|
| 1245 |
+
"step": 5000
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 35.21,
|
| 1249 |
+
"eval_loss": 0.38145187497138977,
|
| 1250 |
+
"eval_runtime": 141.6355,
|
| 1251 |
+
"eval_samples_per_second": 64.052,
|
| 1252 |
+
"eval_steps_per_second": 1.003,
|
| 1253 |
+
"eval_wer": 19.403279334932474,
|
| 1254 |
+
"step": 5000
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 35.21,
|
| 1258 |
+
"step": 5000,
|
| 1259 |
+
"total_flos": 4.14377543467008e+19,
|
| 1260 |
+
"train_loss": 0.30019377393722535,
|
| 1261 |
+
"train_runtime": 10549.3759,
|
| 1262 |
+
"train_samples_per_second": 60.667,
|
| 1263 |
+
"train_steps_per_second": 0.474
|
| 1264 |
+
}
|
| 1265 |
+
],
|
| 1266 |
+
"logging_steps": 25,
|
| 1267 |
+
"max_steps": 5000,
|
| 1268 |
+
"num_input_tokens_seen": 0,
|
| 1269 |
+
"num_train_epochs": 36,
|
| 1270 |
+
"save_steps": 1000,
|
| 1271 |
+
"total_flos": 4.14377543467008e+19,
|
| 1272 |
+
"train_batch_size": 128,
|
| 1273 |
+
"trial_name": null,
|
| 1274 |
+
"trial_params": null
|
| 1275 |
+
}
|