End of training
Browse files- README.md +10 -9
- all_results.json +12 -0
- eval_results.json +8 -0
- runs/Apr23_01-39-44_ip-172-31-23-125/events.out.tfevents.1713854533.ip-172-31-23-125.5477.1 +3 -0
- train_results.json +7 -0
- trainer_state.json +1275 -0
README.md
CHANGED
|
@@ -2,38 +2,39 @@
|
|
| 2 |
license: apache-2.0
|
| 3 |
base_model: arun100/whisper-base-uk-1
|
| 4 |
tags:
|
|
|
|
| 5 |
- generated_from_trainer
|
| 6 |
datasets:
|
| 7 |
-
- fleurs
|
| 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: fleurs
|
| 18 |
-
type: fleurs
|
| 19 |
config: uk_ua
|
| 20 |
split: test
|
| 21 |
args: uk_ua
|
| 22 |
metrics:
|
| 23 |
- name: Wer
|
| 24 |
type: wer
|
| 25 |
-
value: 33.
|
| 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 [arun100/whisper-base-uk-1](https://huggingface.co/arun100/whisper-base-uk-1) on the fleurs dataset.
|
| 34 |
It achieves the following results on the evaluation set:
|
| 35 |
-
- Loss: 0.
|
| 36 |
-
- Wer: 33.
|
| 37 |
|
| 38 |
## Model description
|
| 39 |
|
|
|
|
| 2 |
license: apache-2.0
|
| 3 |
base_model: arun100/whisper-base-uk-1
|
| 4 |
tags:
|
| 5 |
+
- whisper-event
|
| 6 |
- generated_from_trainer
|
| 7 |
datasets:
|
| 8 |
+
- google/fleurs
|
| 9 |
metrics:
|
| 10 |
- wer
|
| 11 |
model-index:
|
| 12 |
+
- name: Whisper Base Ukrainian
|
| 13 |
results:
|
| 14 |
- task:
|
| 15 |
name: Automatic Speech Recognition
|
| 16 |
type: automatic-speech-recognition
|
| 17 |
dataset:
|
| 18 |
+
name: google/fleurs uk_ua
|
| 19 |
+
type: google/fleurs
|
| 20 |
config: uk_ua
|
| 21 |
split: test
|
| 22 |
args: uk_ua
|
| 23 |
metrics:
|
| 24 |
- name: Wer
|
| 25 |
type: wer
|
| 26 |
+
value: 33.562978427279056
|
| 27 |
---
|
| 28 |
|
| 29 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 30 |
should probably proofread and complete it, then remove this comment. -->
|
| 31 |
|
| 32 |
+
# Whisper Base Ukrainian
|
| 33 |
|
| 34 |
+
This model is a fine-tuned version of [arun100/whisper-base-uk-1](https://huggingface.co/arun100/whisper-base-uk-1) on the google/fleurs uk_ua dataset.
|
| 35 |
It achieves the following results on the evaluation set:
|
| 36 |
+
- Loss: 0.4710
|
| 37 |
+
- Wer: 33.5630
|
| 38 |
|
| 39 |
## Model description
|
| 40 |
|
all_results.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 476.0,
|
| 3 |
+
"eval_loss": 0.47103559970855713,
|
| 4 |
+
"eval_runtime": 97.3107,
|
| 5 |
+
"eval_samples_per_second": 7.707,
|
| 6 |
+
"eval_steps_per_second": 0.247,
|
| 7 |
+
"eval_wer": 33.562978427279056,
|
| 8 |
+
"train_loss": 0.17557046995162964,
|
| 9 |
+
"train_runtime": 18026.516,
|
| 10 |
+
"train_samples_per_second": 17.752,
|
| 11 |
+
"train_steps_per_second": 0.277
|
| 12 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 476.0,
|
| 3 |
+
"eval_loss": 0.47103559970855713,
|
| 4 |
+
"eval_runtime": 97.3107,
|
| 5 |
+
"eval_samples_per_second": 7.707,
|
| 6 |
+
"eval_steps_per_second": 0.247,
|
| 7 |
+
"eval_wer": 33.562978427279056
|
| 8 |
+
}
|
runs/Apr23_01-39-44_ip-172-31-23-125/events.out.tfevents.1713854533.ip-172-31-23-125.5477.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cb4c9c087fa4a90c9eceaa77b765d887aca349307c6122db4042c7ad137d74c0
|
| 3 |
+
size 406
|
train_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 476.0,
|
| 3 |
+
"train_loss": 0.17557046995162964,
|
| 4 |
+
"train_runtime": 18026.516,
|
| 5 |
+
"train_samples_per_second": 17.752,
|
| 6 |
+
"train_steps_per_second": 0.277
|
| 7 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1275 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 33.562978427279056,
|
| 3 |
+
"best_model_checkpoint": "./checkpoint-1000",
|
| 4 |
+
"epoch": 476.0004,
|
| 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": 2.0,
|
| 13 |
+
"learning_rate": 1.1499999999999999e-08,
|
| 14 |
+
"loss": 0.729,
|
| 15 |
+
"step": 25
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 4.0,
|
| 19 |
+
"learning_rate": 2.4e-08,
|
| 20 |
+
"loss": 0.7262,
|
| 21 |
+
"step": 50
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 7.0,
|
| 25 |
+
"learning_rate": 3.6499999999999996e-08,
|
| 26 |
+
"loss": 0.7071,
|
| 27 |
+
"step": 75
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 9.0,
|
| 31 |
+
"learning_rate": 4.9e-08,
|
| 32 |
+
"loss": 0.7016,
|
| 33 |
+
"step": 100
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 11.0,
|
| 37 |
+
"learning_rate": 6.15e-08,
|
| 38 |
+
"loss": 0.6788,
|
| 39 |
+
"step": 125
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 14.0,
|
| 43 |
+
"learning_rate": 7.399999999999999e-08,
|
| 44 |
+
"loss": 0.654,
|
| 45 |
+
"step": 150
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 16.0,
|
| 49 |
+
"learning_rate": 8.649999999999999e-08,
|
| 50 |
+
"loss": 0.6294,
|
| 51 |
+
"step": 175
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 19.0,
|
| 55 |
+
"learning_rate": 9.9e-08,
|
| 56 |
+
"loss": 0.5968,
|
| 57 |
+
"step": 200
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 21.0,
|
| 61 |
+
"learning_rate": 1.115e-07,
|
| 62 |
+
"loss": 0.5725,
|
| 63 |
+
"step": 225
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 23.0,
|
| 67 |
+
"learning_rate": 1.24e-07,
|
| 68 |
+
"loss": 0.5483,
|
| 69 |
+
"step": 250
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 26.0,
|
| 73 |
+
"learning_rate": 1.365e-07,
|
| 74 |
+
"loss": 0.522,
|
| 75 |
+
"step": 275
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 28.0,
|
| 79 |
+
"learning_rate": 1.49e-07,
|
| 80 |
+
"loss": 0.5107,
|
| 81 |
+
"step": 300
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 30.0,
|
| 85 |
+
"learning_rate": 1.615e-07,
|
| 86 |
+
"loss": 0.4823,
|
| 87 |
+
"step": 325
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 33.0,
|
| 91 |
+
"learning_rate": 1.7399999999999997e-07,
|
| 92 |
+
"loss": 0.4793,
|
| 93 |
+
"step": 350
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 35.0,
|
| 97 |
+
"learning_rate": 1.8649999999999998e-07,
|
| 98 |
+
"loss": 0.4597,
|
| 99 |
+
"step": 375
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 38.0,
|
| 103 |
+
"learning_rate": 1.99e-07,
|
| 104 |
+
"loss": 0.4533,
|
| 105 |
+
"step": 400
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 40.0,
|
| 109 |
+
"learning_rate": 2.1149999999999998e-07,
|
| 110 |
+
"loss": 0.4347,
|
| 111 |
+
"step": 425
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 42.0,
|
| 115 |
+
"learning_rate": 2.24e-07,
|
| 116 |
+
"loss": 0.4293,
|
| 117 |
+
"step": 450
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 45.0,
|
| 121 |
+
"learning_rate": 2.3649999999999998e-07,
|
| 122 |
+
"loss": 0.4133,
|
| 123 |
+
"step": 475
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 47.0,
|
| 127 |
+
"learning_rate": 2.4899999999999997e-07,
|
| 128 |
+
"loss": 0.406,
|
| 129 |
+
"step": 500
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 49.0,
|
| 133 |
+
"learning_rate": 2.487222222222222e-07,
|
| 134 |
+
"loss": 0.3914,
|
| 135 |
+
"step": 525
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 52.0,
|
| 139 |
+
"learning_rate": 2.473333333333333e-07,
|
| 140 |
+
"loss": 0.3806,
|
| 141 |
+
"step": 550
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 54.0,
|
| 145 |
+
"learning_rate": 2.4594444444444445e-07,
|
| 146 |
+
"loss": 0.3753,
|
| 147 |
+
"step": 575
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 57.0,
|
| 151 |
+
"learning_rate": 2.4455555555555555e-07,
|
| 152 |
+
"loss": 0.3656,
|
| 153 |
+
"step": 600
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 59.0,
|
| 157 |
+
"learning_rate": 2.4316666666666665e-07,
|
| 158 |
+
"loss": 0.359,
|
| 159 |
+
"step": 625
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 61.0,
|
| 163 |
+
"learning_rate": 2.4177777777777775e-07,
|
| 164 |
+
"loss": 0.3518,
|
| 165 |
+
"step": 650
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 64.0,
|
| 169 |
+
"learning_rate": 2.403888888888889e-07,
|
| 170 |
+
"loss": 0.3407,
|
| 171 |
+
"step": 675
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 66.0,
|
| 175 |
+
"learning_rate": 2.3899999999999996e-07,
|
| 176 |
+
"loss": 0.3351,
|
| 177 |
+
"step": 700
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 69.0,
|
| 181 |
+
"learning_rate": 2.3761111111111108e-07,
|
| 182 |
+
"loss": 0.3315,
|
| 183 |
+
"step": 725
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 71.0,
|
| 187 |
+
"learning_rate": 2.362222222222222e-07,
|
| 188 |
+
"loss": 0.3223,
|
| 189 |
+
"step": 750
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 73.0,
|
| 193 |
+
"learning_rate": 2.3483333333333334e-07,
|
| 194 |
+
"loss": 0.3137,
|
| 195 |
+
"step": 775
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 76.0,
|
| 199 |
+
"learning_rate": 2.3344444444444444e-07,
|
| 200 |
+
"loss": 0.3102,
|
| 201 |
+
"step": 800
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 78.0,
|
| 205 |
+
"learning_rate": 2.3205555555555554e-07,
|
| 206 |
+
"loss": 0.3054,
|
| 207 |
+
"step": 825
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 80.0,
|
| 211 |
+
"learning_rate": 2.3066666666666664e-07,
|
| 212 |
+
"loss": 0.2968,
|
| 213 |
+
"step": 850
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 83.0,
|
| 217 |
+
"learning_rate": 2.2927777777777777e-07,
|
| 218 |
+
"loss": 0.289,
|
| 219 |
+
"step": 875
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 85.0,
|
| 223 |
+
"learning_rate": 2.2788888888888887e-07,
|
| 224 |
+
"loss": 0.2893,
|
| 225 |
+
"step": 900
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 88.0,
|
| 229 |
+
"learning_rate": 2.265e-07,
|
| 230 |
+
"loss": 0.2802,
|
| 231 |
+
"step": 925
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 90.0,
|
| 235 |
+
"learning_rate": 2.2511111111111112e-07,
|
| 236 |
+
"loss": 0.273,
|
| 237 |
+
"step": 950
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 92.0,
|
| 241 |
+
"learning_rate": 2.237222222222222e-07,
|
| 242 |
+
"loss": 0.2741,
|
| 243 |
+
"step": 975
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 95.0,
|
| 247 |
+
"learning_rate": 2.2233333333333333e-07,
|
| 248 |
+
"loss": 0.2683,
|
| 249 |
+
"step": 1000
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 95.0,
|
| 253 |
+
"eval_loss": 0.47103559970855713,
|
| 254 |
+
"eval_runtime": 117.5444,
|
| 255 |
+
"eval_samples_per_second": 6.381,
|
| 256 |
+
"eval_steps_per_second": 0.204,
|
| 257 |
+
"eval_wer": 33.562978427279056,
|
| 258 |
+
"step": 1000
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 97.0,
|
| 262 |
+
"learning_rate": 2.2094444444444443e-07,
|
| 263 |
+
"loss": 0.2598,
|
| 264 |
+
"step": 1025
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 99.0,
|
| 268 |
+
"learning_rate": 2.1955555555555555e-07,
|
| 269 |
+
"loss": 0.2568,
|
| 270 |
+
"step": 1050
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 102.0,
|
| 274 |
+
"learning_rate": 2.1816666666666665e-07,
|
| 275 |
+
"loss": 0.2498,
|
| 276 |
+
"step": 1075
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 104.0,
|
| 280 |
+
"learning_rate": 2.1677777777777778e-07,
|
| 281 |
+
"loss": 0.2462,
|
| 282 |
+
"step": 1100
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 107.0,
|
| 286 |
+
"learning_rate": 2.1538888888888888e-07,
|
| 287 |
+
"loss": 0.2401,
|
| 288 |
+
"step": 1125
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 109.0,
|
| 292 |
+
"learning_rate": 2.1399999999999998e-07,
|
| 293 |
+
"loss": 0.2404,
|
| 294 |
+
"step": 1150
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 111.0,
|
| 298 |
+
"learning_rate": 2.126111111111111e-07,
|
| 299 |
+
"loss": 0.2348,
|
| 300 |
+
"step": 1175
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 114.0,
|
| 304 |
+
"learning_rate": 2.112222222222222e-07,
|
| 305 |
+
"loss": 0.2308,
|
| 306 |
+
"step": 1200
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 116.0,
|
| 310 |
+
"learning_rate": 2.0983333333333334e-07,
|
| 311 |
+
"loss": 0.2281,
|
| 312 |
+
"step": 1225
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 119.0,
|
| 316 |
+
"learning_rate": 2.0844444444444441e-07,
|
| 317 |
+
"loss": 0.2232,
|
| 318 |
+
"step": 1250
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 121.0,
|
| 322 |
+
"learning_rate": 2.0705555555555554e-07,
|
| 323 |
+
"loss": 0.2216,
|
| 324 |
+
"step": 1275
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 123.0,
|
| 328 |
+
"learning_rate": 2.0566666666666664e-07,
|
| 329 |
+
"loss": 0.2144,
|
| 330 |
+
"step": 1300
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 126.0,
|
| 334 |
+
"learning_rate": 2.0427777777777777e-07,
|
| 335 |
+
"loss": 0.2125,
|
| 336 |
+
"step": 1325
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 128.0,
|
| 340 |
+
"learning_rate": 2.028888888888889e-07,
|
| 341 |
+
"loss": 0.2103,
|
| 342 |
+
"step": 1350
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 130.0,
|
| 346 |
+
"learning_rate": 2.015e-07,
|
| 347 |
+
"loss": 0.2039,
|
| 348 |
+
"step": 1375
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 133.0,
|
| 352 |
+
"learning_rate": 2.001111111111111e-07,
|
| 353 |
+
"loss": 0.2002,
|
| 354 |
+
"step": 1400
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 135.0,
|
| 358 |
+
"learning_rate": 1.987222222222222e-07,
|
| 359 |
+
"loss": 0.2015,
|
| 360 |
+
"step": 1425
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 138.0,
|
| 364 |
+
"learning_rate": 1.9733333333333333e-07,
|
| 365 |
+
"loss": 0.1954,
|
| 366 |
+
"step": 1450
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 140.0,
|
| 370 |
+
"learning_rate": 1.9594444444444443e-07,
|
| 371 |
+
"loss": 0.1929,
|
| 372 |
+
"step": 1475
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 142.0,
|
| 376 |
+
"learning_rate": 1.9455555555555556e-07,
|
| 377 |
+
"loss": 0.1893,
|
| 378 |
+
"step": 1500
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 145.0,
|
| 382 |
+
"learning_rate": 1.9316666666666666e-07,
|
| 383 |
+
"loss": 0.1873,
|
| 384 |
+
"step": 1525
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 147.0,
|
| 388 |
+
"learning_rate": 1.9177777777777776e-07,
|
| 389 |
+
"loss": 0.1841,
|
| 390 |
+
"step": 1550
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 149.0,
|
| 394 |
+
"learning_rate": 1.9038888888888888e-07,
|
| 395 |
+
"loss": 0.1817,
|
| 396 |
+
"step": 1575
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 152.0,
|
| 400 |
+
"learning_rate": 1.8899999999999999e-07,
|
| 401 |
+
"loss": 0.179,
|
| 402 |
+
"step": 1600
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 154.0,
|
| 406 |
+
"learning_rate": 1.876111111111111e-07,
|
| 407 |
+
"loss": 0.1749,
|
| 408 |
+
"step": 1625
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 157.0,
|
| 412 |
+
"learning_rate": 1.8622222222222221e-07,
|
| 413 |
+
"loss": 0.1722,
|
| 414 |
+
"step": 1650
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 159.0,
|
| 418 |
+
"learning_rate": 1.8483333333333331e-07,
|
| 419 |
+
"loss": 0.1696,
|
| 420 |
+
"step": 1675
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 161.0,
|
| 424 |
+
"learning_rate": 1.8344444444444444e-07,
|
| 425 |
+
"loss": 0.169,
|
| 426 |
+
"step": 1700
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 164.0,
|
| 430 |
+
"learning_rate": 1.8205555555555554e-07,
|
| 431 |
+
"loss": 0.1656,
|
| 432 |
+
"step": 1725
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 166.0,
|
| 436 |
+
"learning_rate": 1.8066666666666667e-07,
|
| 437 |
+
"loss": 0.1605,
|
| 438 |
+
"step": 1750
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 169.0,
|
| 442 |
+
"learning_rate": 1.7927777777777777e-07,
|
| 443 |
+
"loss": 0.1627,
|
| 444 |
+
"step": 1775
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 171.0,
|
| 448 |
+
"learning_rate": 1.778888888888889e-07,
|
| 449 |
+
"loss": 0.1574,
|
| 450 |
+
"step": 1800
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 173.0,
|
| 454 |
+
"learning_rate": 1.7649999999999997e-07,
|
| 455 |
+
"loss": 0.1552,
|
| 456 |
+
"step": 1825
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 176.0,
|
| 460 |
+
"learning_rate": 1.751111111111111e-07,
|
| 461 |
+
"loss": 0.1532,
|
| 462 |
+
"step": 1850
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"epoch": 178.0,
|
| 466 |
+
"learning_rate": 1.7372222222222223e-07,
|
| 467 |
+
"loss": 0.1517,
|
| 468 |
+
"step": 1875
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 180.0,
|
| 472 |
+
"learning_rate": 1.7233333333333333e-07,
|
| 473 |
+
"loss": 0.1483,
|
| 474 |
+
"step": 1900
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 183.0,
|
| 478 |
+
"learning_rate": 1.7094444444444446e-07,
|
| 479 |
+
"loss": 0.147,
|
| 480 |
+
"step": 1925
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 185.0,
|
| 484 |
+
"learning_rate": 1.6955555555555553e-07,
|
| 485 |
+
"loss": 0.1458,
|
| 486 |
+
"step": 1950
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 188.0,
|
| 490 |
+
"learning_rate": 1.6816666666666666e-07,
|
| 491 |
+
"loss": 0.142,
|
| 492 |
+
"step": 1975
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 190.0,
|
| 496 |
+
"learning_rate": 1.6677777777777776e-07,
|
| 497 |
+
"loss": 0.142,
|
| 498 |
+
"step": 2000
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 190.0,
|
| 502 |
+
"eval_loss": 0.4714232385158539,
|
| 503 |
+
"eval_runtime": 99.0664,
|
| 504 |
+
"eval_samples_per_second": 7.571,
|
| 505 |
+
"eval_steps_per_second": 0.242,
|
| 506 |
+
"eval_wer": 33.83437717466945,
|
| 507 |
+
"step": 2000
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 192.0,
|
| 511 |
+
"learning_rate": 1.6538888888888889e-07,
|
| 512 |
+
"loss": 0.1394,
|
| 513 |
+
"step": 2025
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 195.0,
|
| 517 |
+
"learning_rate": 1.64e-07,
|
| 518 |
+
"loss": 0.1373,
|
| 519 |
+
"step": 2050
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 197.0,
|
| 523 |
+
"learning_rate": 1.6261111111111111e-07,
|
| 524 |
+
"loss": 0.1349,
|
| 525 |
+
"step": 2075
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 199.0,
|
| 529 |
+
"learning_rate": 1.6122222222222221e-07,
|
| 530 |
+
"loss": 0.1337,
|
| 531 |
+
"step": 2100
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 202.0,
|
| 535 |
+
"learning_rate": 1.5983333333333332e-07,
|
| 536 |
+
"loss": 0.1336,
|
| 537 |
+
"step": 2125
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 204.0,
|
| 541 |
+
"learning_rate": 1.5844444444444444e-07,
|
| 542 |
+
"loss": 0.1306,
|
| 543 |
+
"step": 2150
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 207.0,
|
| 547 |
+
"learning_rate": 1.5705555555555554e-07,
|
| 548 |
+
"loss": 0.1286,
|
| 549 |
+
"step": 2175
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 209.0,
|
| 553 |
+
"learning_rate": 1.5566666666666667e-07,
|
| 554 |
+
"loss": 0.1279,
|
| 555 |
+
"step": 2200
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 211.0,
|
| 559 |
+
"learning_rate": 1.5427777777777777e-07,
|
| 560 |
+
"loss": 0.124,
|
| 561 |
+
"step": 2225
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 214.0,
|
| 565 |
+
"learning_rate": 1.5288888888888887e-07,
|
| 566 |
+
"loss": 0.1244,
|
| 567 |
+
"step": 2250
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 216.0,
|
| 571 |
+
"learning_rate": 1.515e-07,
|
| 572 |
+
"loss": 0.1208,
|
| 573 |
+
"step": 2275
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 219.0,
|
| 577 |
+
"learning_rate": 1.501111111111111e-07,
|
| 578 |
+
"loss": 0.1203,
|
| 579 |
+
"step": 2300
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 221.0,
|
| 583 |
+
"learning_rate": 1.4872222222222223e-07,
|
| 584 |
+
"loss": 0.1183,
|
| 585 |
+
"step": 2325
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 223.0,
|
| 589 |
+
"learning_rate": 1.4733333333333333e-07,
|
| 590 |
+
"loss": 0.1177,
|
| 591 |
+
"step": 2350
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 226.0,
|
| 595 |
+
"learning_rate": 1.4594444444444443e-07,
|
| 596 |
+
"loss": 0.1175,
|
| 597 |
+
"step": 2375
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 228.0,
|
| 601 |
+
"learning_rate": 1.4455555555555553e-07,
|
| 602 |
+
"loss": 0.1135,
|
| 603 |
+
"step": 2400
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 230.0,
|
| 607 |
+
"learning_rate": 1.4316666666666666e-07,
|
| 608 |
+
"loss": 0.1138,
|
| 609 |
+
"step": 2425
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 233.0,
|
| 613 |
+
"learning_rate": 1.4177777777777779e-07,
|
| 614 |
+
"loss": 0.1115,
|
| 615 |
+
"step": 2450
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 235.0,
|
| 619 |
+
"learning_rate": 1.4038888888888889e-07,
|
| 620 |
+
"loss": 0.1111,
|
| 621 |
+
"step": 2475
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 238.0,
|
| 625 |
+
"learning_rate": 1.3900000000000001e-07,
|
| 626 |
+
"loss": 0.1094,
|
| 627 |
+
"step": 2500
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 240.0,
|
| 631 |
+
"learning_rate": 1.376111111111111e-07,
|
| 632 |
+
"loss": 0.1076,
|
| 633 |
+
"step": 2525
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 242.0,
|
| 637 |
+
"learning_rate": 1.3622222222222222e-07,
|
| 638 |
+
"loss": 0.1066,
|
| 639 |
+
"step": 2550
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 245.0,
|
| 643 |
+
"learning_rate": 1.3483333333333332e-07,
|
| 644 |
+
"loss": 0.1047,
|
| 645 |
+
"step": 2575
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 247.0,
|
| 649 |
+
"learning_rate": 1.3344444444444444e-07,
|
| 650 |
+
"loss": 0.1041,
|
| 651 |
+
"step": 2600
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 249.0,
|
| 655 |
+
"learning_rate": 1.3205555555555557e-07,
|
| 656 |
+
"loss": 0.1037,
|
| 657 |
+
"step": 2625
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 252.0,
|
| 661 |
+
"learning_rate": 1.3066666666666665e-07,
|
| 662 |
+
"loss": 0.1009,
|
| 663 |
+
"step": 2650
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 254.0,
|
| 667 |
+
"learning_rate": 1.2927777777777777e-07,
|
| 668 |
+
"loss": 0.1007,
|
| 669 |
+
"step": 2675
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 257.0,
|
| 673 |
+
"learning_rate": 1.2788888888888887e-07,
|
| 674 |
+
"loss": 0.0996,
|
| 675 |
+
"step": 2700
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 259.0,
|
| 679 |
+
"learning_rate": 1.265e-07,
|
| 680 |
+
"loss": 0.0979,
|
| 681 |
+
"step": 2725
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 261.0,
|
| 685 |
+
"learning_rate": 1.251111111111111e-07,
|
| 686 |
+
"loss": 0.0969,
|
| 687 |
+
"step": 2750
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 264.0,
|
| 691 |
+
"learning_rate": 1.237222222222222e-07,
|
| 692 |
+
"loss": 0.0966,
|
| 693 |
+
"step": 2775
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 266.0,
|
| 697 |
+
"learning_rate": 1.2233333333333333e-07,
|
| 698 |
+
"loss": 0.096,
|
| 699 |
+
"step": 2800
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 269.0,
|
| 703 |
+
"learning_rate": 1.2094444444444443e-07,
|
| 704 |
+
"loss": 0.0939,
|
| 705 |
+
"step": 2825
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 271.0,
|
| 709 |
+
"learning_rate": 1.1955555555555556e-07,
|
| 710 |
+
"loss": 0.0922,
|
| 711 |
+
"step": 2850
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 273.0,
|
| 715 |
+
"learning_rate": 1.1816666666666666e-07,
|
| 716 |
+
"loss": 0.0932,
|
| 717 |
+
"step": 2875
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 276.0,
|
| 721 |
+
"learning_rate": 1.1677777777777776e-07,
|
| 722 |
+
"loss": 0.0914,
|
| 723 |
+
"step": 2900
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 278.0,
|
| 727 |
+
"learning_rate": 1.1538888888888889e-07,
|
| 728 |
+
"loss": 0.0894,
|
| 729 |
+
"step": 2925
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 280.0,
|
| 733 |
+
"learning_rate": 1.14e-07,
|
| 734 |
+
"loss": 0.09,
|
| 735 |
+
"step": 2950
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 283.0,
|
| 739 |
+
"learning_rate": 1.126111111111111e-07,
|
| 740 |
+
"loss": 0.0887,
|
| 741 |
+
"step": 2975
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 285.0,
|
| 745 |
+
"learning_rate": 1.1122222222222222e-07,
|
| 746 |
+
"loss": 0.0871,
|
| 747 |
+
"step": 3000
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 285.0,
|
| 751 |
+
"eval_loss": 0.4781545102596283,
|
| 752 |
+
"eval_runtime": 100.2584,
|
| 753 |
+
"eval_samples_per_second": 7.481,
|
| 754 |
+
"eval_steps_per_second": 0.239,
|
| 755 |
+
"eval_wer": 33.95963813500348,
|
| 756 |
+
"step": 3000
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 288.0,
|
| 760 |
+
"learning_rate": 1.0983333333333333e-07,
|
| 761 |
+
"loss": 0.0874,
|
| 762 |
+
"step": 3025
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 290.0,
|
| 766 |
+
"learning_rate": 1.0844444444444443e-07,
|
| 767 |
+
"loss": 0.0862,
|
| 768 |
+
"step": 3050
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 292.0,
|
| 772 |
+
"learning_rate": 1.0705555555555555e-07,
|
| 773 |
+
"loss": 0.0849,
|
| 774 |
+
"step": 3075
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 295.0,
|
| 778 |
+
"learning_rate": 1.0566666666666667e-07,
|
| 779 |
+
"loss": 0.0841,
|
| 780 |
+
"step": 3100
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 297.0,
|
| 784 |
+
"learning_rate": 1.0427777777777777e-07,
|
| 785 |
+
"loss": 0.0836,
|
| 786 |
+
"step": 3125
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 299.0,
|
| 790 |
+
"learning_rate": 1.0288888888888889e-07,
|
| 791 |
+
"loss": 0.0824,
|
| 792 |
+
"step": 3150
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 302.0,
|
| 796 |
+
"learning_rate": 1.015e-07,
|
| 797 |
+
"loss": 0.0825,
|
| 798 |
+
"step": 3175
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 304.0,
|
| 802 |
+
"learning_rate": 1.001111111111111e-07,
|
| 803 |
+
"loss": 0.0811,
|
| 804 |
+
"step": 3200
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 307.0,
|
| 808 |
+
"learning_rate": 9.872222222222222e-08,
|
| 809 |
+
"loss": 0.0804,
|
| 810 |
+
"step": 3225
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 309.0,
|
| 814 |
+
"learning_rate": 9.733333333333332e-08,
|
| 815 |
+
"loss": 0.0804,
|
| 816 |
+
"step": 3250
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 311.0,
|
| 820 |
+
"learning_rate": 9.594444444444445e-08,
|
| 821 |
+
"loss": 0.0799,
|
| 822 |
+
"step": 3275
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 314.0,
|
| 826 |
+
"learning_rate": 9.455555555555556e-08,
|
| 827 |
+
"loss": 0.0784,
|
| 828 |
+
"step": 3300
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 316.0,
|
| 832 |
+
"learning_rate": 9.316666666666666e-08,
|
| 833 |
+
"loss": 0.0778,
|
| 834 |
+
"step": 3325
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 319.0,
|
| 838 |
+
"learning_rate": 9.177777777777777e-08,
|
| 839 |
+
"loss": 0.0762,
|
| 840 |
+
"step": 3350
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 321.0,
|
| 844 |
+
"learning_rate": 9.038888888888889e-08,
|
| 845 |
+
"loss": 0.077,
|
| 846 |
+
"step": 3375
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 323.0,
|
| 850 |
+
"learning_rate": 8.899999999999999e-08,
|
| 851 |
+
"loss": 0.0761,
|
| 852 |
+
"step": 3400
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 326.0,
|
| 856 |
+
"learning_rate": 8.76111111111111e-08,
|
| 857 |
+
"loss": 0.0755,
|
| 858 |
+
"step": 3425
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 328.0,
|
| 862 |
+
"learning_rate": 8.622222222222222e-08,
|
| 863 |
+
"loss": 0.0753,
|
| 864 |
+
"step": 3450
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 330.0,
|
| 868 |
+
"learning_rate": 8.483333333333333e-08,
|
| 869 |
+
"loss": 0.074,
|
| 870 |
+
"step": 3475
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 333.0,
|
| 874 |
+
"learning_rate": 8.344444444444445e-08,
|
| 875 |
+
"loss": 0.0736,
|
| 876 |
+
"step": 3500
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 335.0,
|
| 880 |
+
"learning_rate": 8.205555555555556e-08,
|
| 881 |
+
"loss": 0.0731,
|
| 882 |
+
"step": 3525
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 338.0,
|
| 886 |
+
"learning_rate": 8.066666666666666e-08,
|
| 887 |
+
"loss": 0.0728,
|
| 888 |
+
"step": 3550
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 340.0,
|
| 892 |
+
"learning_rate": 7.927777777777778e-08,
|
| 893 |
+
"loss": 0.0719,
|
| 894 |
+
"step": 3575
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 342.0,
|
| 898 |
+
"learning_rate": 7.788888888888888e-08,
|
| 899 |
+
"loss": 0.0721,
|
| 900 |
+
"step": 3600
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 345.0,
|
| 904 |
+
"learning_rate": 7.649999999999999e-08,
|
| 905 |
+
"loss": 0.0719,
|
| 906 |
+
"step": 3625
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 347.0,
|
| 910 |
+
"learning_rate": 7.511111111111112e-08,
|
| 911 |
+
"loss": 0.0701,
|
| 912 |
+
"step": 3650
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 349.0,
|
| 916 |
+
"learning_rate": 7.372222222222222e-08,
|
| 917 |
+
"loss": 0.0705,
|
| 918 |
+
"step": 3675
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 352.0,
|
| 922 |
+
"learning_rate": 7.233333333333333e-08,
|
| 923 |
+
"loss": 0.0703,
|
| 924 |
+
"step": 3700
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 354.0,
|
| 928 |
+
"learning_rate": 7.094444444444445e-08,
|
| 929 |
+
"loss": 0.0689,
|
| 930 |
+
"step": 3725
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 357.0,
|
| 934 |
+
"learning_rate": 6.955555555555555e-08,
|
| 935 |
+
"loss": 0.0694,
|
| 936 |
+
"step": 3750
|
| 937 |
+
},
|
| 938 |
+
{
|
| 939 |
+
"epoch": 359.0,
|
| 940 |
+
"learning_rate": 6.816666666666666e-08,
|
| 941 |
+
"loss": 0.0685,
|
| 942 |
+
"step": 3775
|
| 943 |
+
},
|
| 944 |
+
{
|
| 945 |
+
"epoch": 361.0,
|
| 946 |
+
"learning_rate": 6.677777777777778e-08,
|
| 947 |
+
"loss": 0.0681,
|
| 948 |
+
"step": 3800
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 364.0,
|
| 952 |
+
"learning_rate": 6.538888888888888e-08,
|
| 953 |
+
"loss": 0.0674,
|
| 954 |
+
"step": 3825
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 366.0,
|
| 958 |
+
"learning_rate": 6.4e-08,
|
| 959 |
+
"loss": 0.0671,
|
| 960 |
+
"step": 3850
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 369.0,
|
| 964 |
+
"learning_rate": 6.261111111111112e-08,
|
| 965 |
+
"loss": 0.0677,
|
| 966 |
+
"step": 3875
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 371.0,
|
| 970 |
+
"learning_rate": 6.122222222222222e-08,
|
| 971 |
+
"loss": 0.0666,
|
| 972 |
+
"step": 3900
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"epoch": 373.0,
|
| 976 |
+
"learning_rate": 5.983333333333333e-08,
|
| 977 |
+
"loss": 0.0658,
|
| 978 |
+
"step": 3925
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 376.0,
|
| 982 |
+
"learning_rate": 5.844444444444444e-08,
|
| 983 |
+
"loss": 0.0656,
|
| 984 |
+
"step": 3950
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 378.0,
|
| 988 |
+
"learning_rate": 5.705555555555555e-08,
|
| 989 |
+
"loss": 0.0652,
|
| 990 |
+
"step": 3975
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 380.0,
|
| 994 |
+
"learning_rate": 5.566666666666667e-08,
|
| 995 |
+
"loss": 0.0656,
|
| 996 |
+
"step": 4000
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 380.0,
|
| 1000 |
+
"eval_loss": 0.4829937517642975,
|
| 1001 |
+
"eval_runtime": 96.1124,
|
| 1002 |
+
"eval_samples_per_second": 7.803,
|
| 1003 |
+
"eval_steps_per_second": 0.25,
|
| 1004 |
+
"eval_wer": 33.72303409881698,
|
| 1005 |
+
"step": 4000
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 383.0,
|
| 1009 |
+
"learning_rate": 5.4277777777777776e-08,
|
| 1010 |
+
"loss": 0.0645,
|
| 1011 |
+
"step": 4025
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 385.0,
|
| 1015 |
+
"learning_rate": 5.2888888888888884e-08,
|
| 1016 |
+
"loss": 0.0647,
|
| 1017 |
+
"step": 4050
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 388.0,
|
| 1021 |
+
"learning_rate": 5.15e-08,
|
| 1022 |
+
"loss": 0.0651,
|
| 1023 |
+
"step": 4075
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 390.0,
|
| 1027 |
+
"learning_rate": 5.011111111111111e-08,
|
| 1028 |
+
"loss": 0.0642,
|
| 1029 |
+
"step": 4100
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 392.0,
|
| 1033 |
+
"learning_rate": 4.872222222222222e-08,
|
| 1034 |
+
"loss": 0.0641,
|
| 1035 |
+
"step": 4125
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 395.0,
|
| 1039 |
+
"learning_rate": 4.733333333333333e-08,
|
| 1040 |
+
"loss": 0.0634,
|
| 1041 |
+
"step": 4150
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 397.0,
|
| 1045 |
+
"learning_rate": 4.594444444444445e-08,
|
| 1046 |
+
"loss": 0.063,
|
| 1047 |
+
"step": 4175
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 399.0,
|
| 1051 |
+
"learning_rate": 4.4555555555555555e-08,
|
| 1052 |
+
"loss": 0.0628,
|
| 1053 |
+
"step": 4200
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 402.0,
|
| 1057 |
+
"learning_rate": 4.316666666666666e-08,
|
| 1058 |
+
"loss": 0.0631,
|
| 1059 |
+
"step": 4225
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 404.0,
|
| 1063 |
+
"learning_rate": 4.177777777777777e-08,
|
| 1064 |
+
"loss": 0.0622,
|
| 1065 |
+
"step": 4250
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 407.0,
|
| 1069 |
+
"learning_rate": 4.038888888888889e-08,
|
| 1070 |
+
"loss": 0.0617,
|
| 1071 |
+
"step": 4275
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 409.0,
|
| 1075 |
+
"learning_rate": 3.9e-08,
|
| 1076 |
+
"loss": 0.0619,
|
| 1077 |
+
"step": 4300
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 411.0,
|
| 1081 |
+
"learning_rate": 3.7611111111111106e-08,
|
| 1082 |
+
"loss": 0.0622,
|
| 1083 |
+
"step": 4325
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 414.0,
|
| 1087 |
+
"learning_rate": 3.622222222222223e-08,
|
| 1088 |
+
"loss": 0.0617,
|
| 1089 |
+
"step": 4350
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 416.0,
|
| 1093 |
+
"learning_rate": 3.4833333333333334e-08,
|
| 1094 |
+
"loss": 0.0615,
|
| 1095 |
+
"step": 4375
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 419.0,
|
| 1099 |
+
"learning_rate": 3.344444444444444e-08,
|
| 1100 |
+
"loss": 0.0611,
|
| 1101 |
+
"step": 4400
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 421.0,
|
| 1105 |
+
"learning_rate": 3.205555555555555e-08,
|
| 1106 |
+
"loss": 0.0615,
|
| 1107 |
+
"step": 4425
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 423.0,
|
| 1111 |
+
"learning_rate": 3.0666666666666663e-08,
|
| 1112 |
+
"loss": 0.0609,
|
| 1113 |
+
"step": 4450
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 426.0,
|
| 1117 |
+
"learning_rate": 2.9277777777777777e-08,
|
| 1118 |
+
"loss": 0.061,
|
| 1119 |
+
"step": 4475
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 428.0,
|
| 1123 |
+
"learning_rate": 2.7888888888888888e-08,
|
| 1124 |
+
"loss": 0.0602,
|
| 1125 |
+
"step": 4500
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 430.0,
|
| 1129 |
+
"learning_rate": 2.65e-08,
|
| 1130 |
+
"loss": 0.0606,
|
| 1131 |
+
"step": 4525
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 433.0,
|
| 1135 |
+
"learning_rate": 2.511111111111111e-08,
|
| 1136 |
+
"loss": 0.0595,
|
| 1137 |
+
"step": 4550
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 435.0,
|
| 1141 |
+
"learning_rate": 2.372222222222222e-08,
|
| 1142 |
+
"loss": 0.0608,
|
| 1143 |
+
"step": 4575
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 438.0,
|
| 1147 |
+
"learning_rate": 2.233333333333333e-08,
|
| 1148 |
+
"loss": 0.0599,
|
| 1149 |
+
"step": 4600
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 440.0,
|
| 1153 |
+
"learning_rate": 2.0944444444444446e-08,
|
| 1154 |
+
"loss": 0.0602,
|
| 1155 |
+
"step": 4625
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 442.0,
|
| 1159 |
+
"learning_rate": 1.9555555555555553e-08,
|
| 1160 |
+
"loss": 0.0593,
|
| 1161 |
+
"step": 4650
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 445.0,
|
| 1165 |
+
"learning_rate": 1.8166666666666667e-08,
|
| 1166 |
+
"loss": 0.0597,
|
| 1167 |
+
"step": 4675
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 447.0,
|
| 1171 |
+
"learning_rate": 1.6777777777777775e-08,
|
| 1172 |
+
"loss": 0.0592,
|
| 1173 |
+
"step": 4700
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 449.0,
|
| 1177 |
+
"learning_rate": 1.538888888888889e-08,
|
| 1178 |
+
"loss": 0.059,
|
| 1179 |
+
"step": 4725
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 452.0,
|
| 1183 |
+
"learning_rate": 1.4e-08,
|
| 1184 |
+
"loss": 0.0591,
|
| 1185 |
+
"step": 4750
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 454.0,
|
| 1189 |
+
"learning_rate": 1.261111111111111e-08,
|
| 1190 |
+
"loss": 0.0598,
|
| 1191 |
+
"step": 4775
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 457.0,
|
| 1195 |
+
"learning_rate": 1.1222222222222221e-08,
|
| 1196 |
+
"loss": 0.0597,
|
| 1197 |
+
"step": 4800
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 459.0,
|
| 1201 |
+
"learning_rate": 9.833333333333332e-09,
|
| 1202 |
+
"loss": 0.0588,
|
| 1203 |
+
"step": 4825
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 461.0,
|
| 1207 |
+
"learning_rate": 8.444444444444443e-09,
|
| 1208 |
+
"loss": 0.0598,
|
| 1209 |
+
"step": 4850
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 464.0,
|
| 1213 |
+
"learning_rate": 7.055555555555555e-09,
|
| 1214 |
+
"loss": 0.0588,
|
| 1215 |
+
"step": 4875
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 466.0,
|
| 1219 |
+
"learning_rate": 5.666666666666667e-09,
|
| 1220 |
+
"loss": 0.0596,
|
| 1221 |
+
"step": 4900
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 469.0,
|
| 1225 |
+
"learning_rate": 4.277777777777778e-09,
|
| 1226 |
+
"loss": 0.0579,
|
| 1227 |
+
"step": 4925
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 471.0,
|
| 1231 |
+
"learning_rate": 2.8888888888888886e-09,
|
| 1232 |
+
"loss": 0.0596,
|
| 1233 |
+
"step": 4950
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 473.0,
|
| 1237 |
+
"learning_rate": 1.5e-09,
|
| 1238 |
+
"loss": 0.0588,
|
| 1239 |
+
"step": 4975
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 476.0,
|
| 1243 |
+
"learning_rate": 1.1111111111111111e-10,
|
| 1244 |
+
"loss": 0.0595,
|
| 1245 |
+
"step": 5000
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 476.0,
|
| 1249 |
+
"eval_loss": 0.4846940040588379,
|
| 1250 |
+
"eval_runtime": 110.8316,
|
| 1251 |
+
"eval_samples_per_second": 6.767,
|
| 1252 |
+
"eval_steps_per_second": 0.217,
|
| 1253 |
+
"eval_wer": 33.7160751565762,
|
| 1254 |
+
"step": 5000
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 476.0,
|
| 1258 |
+
"step": 5000,
|
| 1259 |
+
"total_flos": 2.004511153324032e+19,
|
| 1260 |
+
"train_loss": 0.17557046995162964,
|
| 1261 |
+
"train_runtime": 18026.516,
|
| 1262 |
+
"train_samples_per_second": 17.752,
|
| 1263 |
+
"train_steps_per_second": 0.277
|
| 1264 |
+
}
|
| 1265 |
+
],
|
| 1266 |
+
"logging_steps": 25,
|
| 1267 |
+
"max_steps": 5000,
|
| 1268 |
+
"num_input_tokens_seen": 0,
|
| 1269 |
+
"num_train_epochs": 9223372036854775807,
|
| 1270 |
+
"save_steps": 1000,
|
| 1271 |
+
"total_flos": 2.004511153324032e+19,
|
| 1272 |
+
"train_batch_size": 32,
|
| 1273 |
+
"trial_name": null,
|
| 1274 |
+
"trial_params": null
|
| 1275 |
+
}
|