Masaki Hattori
commited on
Commit
·
d76bd93
1
Parent(s):
0bc20cb
Model save
Browse files- README.md +83 -0
- trainer_state.json +618 -0
README.md
ADDED
|
@@ -0,0 +1,83 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
base_model: rinna/japanese-hubert-base
|
| 4 |
+
tags:
|
| 5 |
+
- generated_from_trainer
|
| 6 |
+
metrics:
|
| 7 |
+
- wer
|
| 8 |
+
model-index:
|
| 9 |
+
- name: hubert-rinnna-jp-jdrtsp-fw07sp-clean
|
| 10 |
+
results: []
|
| 11 |
+
---
|
| 12 |
+
|
| 13 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 14 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 15 |
+
|
| 16 |
+
# hubert-rinnna-jp-jdrtsp-fw07sp-clean
|
| 17 |
+
|
| 18 |
+
This model is a fine-tuned version of [rinna/japanese-hubert-base](https://huggingface.co/rinna/japanese-hubert-base) on the None dataset.
|
| 19 |
+
It achieves the following results on the evaluation set:
|
| 20 |
+
- Loss: 0.2393
|
| 21 |
+
- Wer: 0.2187
|
| 22 |
+
- Cer: 0.1210
|
| 23 |
+
|
| 24 |
+
## Model description
|
| 25 |
+
|
| 26 |
+
More information needed
|
| 27 |
+
|
| 28 |
+
## Intended uses & limitations
|
| 29 |
+
|
| 30 |
+
More information needed
|
| 31 |
+
|
| 32 |
+
## Training and evaluation data
|
| 33 |
+
|
| 34 |
+
More information needed
|
| 35 |
+
|
| 36 |
+
## Training procedure
|
| 37 |
+
|
| 38 |
+
### Training hyperparameters
|
| 39 |
+
|
| 40 |
+
The following hyperparameters were used during training:
|
| 41 |
+
- learning_rate: 2e-05
|
| 42 |
+
- train_batch_size: 32
|
| 43 |
+
- eval_batch_size: 16
|
| 44 |
+
- seed: 42
|
| 45 |
+
- gradient_accumulation_steps: 2
|
| 46 |
+
- total_train_batch_size: 64
|
| 47 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 48 |
+
- lr_scheduler_type: linear
|
| 49 |
+
- lr_scheduler_warmup_steps: 1000
|
| 50 |
+
- num_epochs: 20
|
| 51 |
+
|
| 52 |
+
### Training results
|
| 53 |
+
|
| 54 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|
| 55 |
+
|:-------------:|:-----:|:----:|:---------------:|:------:|:------:|
|
| 56 |
+
| 8.6058 | 1.0 | 164 | 6.2883 | 0.9743 | 0.9861 |
|
| 57 |
+
| 4.6616 | 2.0 | 328 | 4.0451 | 0.9743 | 0.9861 |
|
| 58 |
+
| 3.8526 | 3.0 | 492 | 3.5417 | 0.9743 | 0.9861 |
|
| 59 |
+
| 3.2384 | 4.0 | 656 | 3.0505 | 0.9743 | 0.9861 |
|
| 60 |
+
| 2.7948 | 5.0 | 820 | 2.6706 | 0.9743 | 0.9861 |
|
| 61 |
+
| 2.549 | 6.0 | 984 | 2.4268 | 0.9743 | 0.9861 |
|
| 62 |
+
| 2.1808 | 7.0 | 1148 | 1.8554 | 0.9743 | 0.9861 |
|
| 63 |
+
| 1.6069 | 8.0 | 1312 | 1.2551 | 0.6822 | 0.6231 |
|
| 64 |
+
| 1.1916 | 9.0 | 1476 | 0.7985 | 0.3679 | 0.2242 |
|
| 65 |
+
| 0.9977 | 10.0 | 1640 | 0.6234 | 0.3118 | 0.1827 |
|
| 66 |
+
| 0.836 | 11.0 | 1804 | 0.5103 | 0.2801 | 0.1643 |
|
| 67 |
+
| 0.7515 | 12.0 | 1968 | 0.4305 | 0.2663 | 0.1549 |
|
| 68 |
+
| 0.7045 | 13.0 | 2132 | 0.3688 | 0.2489 | 0.1413 |
|
| 69 |
+
| 0.6533 | 14.0 | 2296 | 0.3258 | 0.2399 | 0.1340 |
|
| 70 |
+
| 0.5906 | 15.0 | 2460 | 0.2941 | 0.2318 | 0.1288 |
|
| 71 |
+
| 0.5746 | 16.0 | 2624 | 0.2748 | 0.2300 | 0.1278 |
|
| 72 |
+
| 0.5169 | 17.0 | 2788 | 0.2573 | 0.2240 | 0.1242 |
|
| 73 |
+
| 0.5511 | 18.0 | 2952 | 0.2479 | 0.2211 | 0.1228 |
|
| 74 |
+
| 0.5318 | 19.0 | 3116 | 0.2410 | 0.2186 | 0.1210 |
|
| 75 |
+
| 0.5174 | 20.0 | 3280 | 0.2393 | 0.2187 | 0.1210 |
|
| 76 |
+
|
| 77 |
+
|
| 78 |
+
### Framework versions
|
| 79 |
+
|
| 80 |
+
- Transformers 4.34.0.dev0
|
| 81 |
+
- Pytorch 2.0.1+cu118
|
| 82 |
+
- Datasets 2.14.5
|
| 83 |
+
- Tokenizers 0.13.3
|
trainer_state.json
ADDED
|
@@ -0,0 +1,618 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 20.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 3280,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.3,
|
| 13 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 14 |
+
"loss": 9.6805,
|
| 15 |
+
"step": 50
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.61,
|
| 19 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 20 |
+
"loss": 9.3474,
|
| 21 |
+
"step": 100
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.91,
|
| 25 |
+
"learning_rate": 3e-06,
|
| 26 |
+
"loss": 8.6058,
|
| 27 |
+
"step": 150
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 1.0,
|
| 31 |
+
"eval_cer": 0.986131055853677,
|
| 32 |
+
"eval_loss": 6.288315296173096,
|
| 33 |
+
"eval_runtime": 21.6292,
|
| 34 |
+
"eval_samples_per_second": 121.318,
|
| 35 |
+
"eval_steps_per_second": 7.582,
|
| 36 |
+
"eval_wer": 0.9742561721882254,
|
| 37 |
+
"step": 164
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 1.22,
|
| 41 |
+
"learning_rate": 4.000000000000001e-06,
|
| 42 |
+
"loss": 6.5168,
|
| 43 |
+
"step": 200
|
| 44 |
+
},
|
| 45 |
+
{
|
| 46 |
+
"epoch": 1.52,
|
| 47 |
+
"learning_rate": 5e-06,
|
| 48 |
+
"loss": 5.178,
|
| 49 |
+
"step": 250
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"epoch": 1.83,
|
| 53 |
+
"learning_rate": 6e-06,
|
| 54 |
+
"loss": 4.6616,
|
| 55 |
+
"step": 300
|
| 56 |
+
},
|
| 57 |
+
{
|
| 58 |
+
"epoch": 2.0,
|
| 59 |
+
"eval_cer": 0.986131055853677,
|
| 60 |
+
"eval_loss": 4.045103549957275,
|
| 61 |
+
"eval_runtime": 21.856,
|
| 62 |
+
"eval_samples_per_second": 120.059,
|
| 63 |
+
"eval_steps_per_second": 7.504,
|
| 64 |
+
"eval_wer": 0.9742561721882254,
|
| 65 |
+
"step": 328
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 2.13,
|
| 69 |
+
"learning_rate": 7e-06,
|
| 70 |
+
"loss": 4.2963,
|
| 71 |
+
"step": 350
|
| 72 |
+
},
|
| 73 |
+
{
|
| 74 |
+
"epoch": 2.44,
|
| 75 |
+
"learning_rate": 8.000000000000001e-06,
|
| 76 |
+
"loss": 4.0823,
|
| 77 |
+
"step": 400
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 2.74,
|
| 81 |
+
"learning_rate": 9e-06,
|
| 82 |
+
"loss": 3.8526,
|
| 83 |
+
"step": 450
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 3.0,
|
| 87 |
+
"eval_cer": 0.986131055853677,
|
| 88 |
+
"eval_loss": 3.5416758060455322,
|
| 89 |
+
"eval_runtime": 21.8041,
|
| 90 |
+
"eval_samples_per_second": 120.344,
|
| 91 |
+
"eval_steps_per_second": 7.522,
|
| 92 |
+
"eval_wer": 0.9742561721882254,
|
| 93 |
+
"step": 492
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 3.05,
|
| 97 |
+
"learning_rate": 1e-05,
|
| 98 |
+
"loss": 3.7408,
|
| 99 |
+
"step": 500
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 3.35,
|
| 103 |
+
"learning_rate": 1.1000000000000001e-05,
|
| 104 |
+
"loss": 3.5235,
|
| 105 |
+
"step": 550
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 3.66,
|
| 109 |
+
"learning_rate": 1.2e-05,
|
| 110 |
+
"loss": 3.3617,
|
| 111 |
+
"step": 600
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 3.96,
|
| 115 |
+
"learning_rate": 1.3000000000000001e-05,
|
| 116 |
+
"loss": 3.2384,
|
| 117 |
+
"step": 650
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 4.0,
|
| 121 |
+
"eval_cer": 0.986131055853677,
|
| 122 |
+
"eval_loss": 3.0504775047302246,
|
| 123 |
+
"eval_runtime": 22.0029,
|
| 124 |
+
"eval_samples_per_second": 119.257,
|
| 125 |
+
"eval_steps_per_second": 7.454,
|
| 126 |
+
"eval_wer": 0.9742561721882254,
|
| 127 |
+
"step": 656
|
| 128 |
+
},
|
| 129 |
+
{
|
| 130 |
+
"epoch": 4.27,
|
| 131 |
+
"learning_rate": 1.4e-05,
|
| 132 |
+
"loss": 3.0665,
|
| 133 |
+
"step": 700
|
| 134 |
+
},
|
| 135 |
+
{
|
| 136 |
+
"epoch": 4.57,
|
| 137 |
+
"learning_rate": 1.5000000000000002e-05,
|
| 138 |
+
"loss": 2.9046,
|
| 139 |
+
"step": 750
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 4.88,
|
| 143 |
+
"learning_rate": 1.6000000000000003e-05,
|
| 144 |
+
"loss": 2.7948,
|
| 145 |
+
"step": 800
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 5.0,
|
| 149 |
+
"eval_cer": 0.986131055853677,
|
| 150 |
+
"eval_loss": 2.670631170272827,
|
| 151 |
+
"eval_runtime": 21.7032,
|
| 152 |
+
"eval_samples_per_second": 120.904,
|
| 153 |
+
"eval_steps_per_second": 7.556,
|
| 154 |
+
"eval_wer": 0.9742561721882254,
|
| 155 |
+
"step": 820
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 5.18,
|
| 159 |
+
"learning_rate": 1.7e-05,
|
| 160 |
+
"loss": 2.6959,
|
| 161 |
+
"step": 850
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 5.49,
|
| 165 |
+
"learning_rate": 1.8e-05,
|
| 166 |
+
"loss": 2.6047,
|
| 167 |
+
"step": 900
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 5.79,
|
| 171 |
+
"learning_rate": 1.9e-05,
|
| 172 |
+
"loss": 2.549,
|
| 173 |
+
"step": 950
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 6.0,
|
| 177 |
+
"eval_cer": 0.986131055853677,
|
| 178 |
+
"eval_loss": 2.426814317703247,
|
| 179 |
+
"eval_runtime": 21.9773,
|
| 180 |
+
"eval_samples_per_second": 119.396,
|
| 181 |
+
"eval_steps_per_second": 7.462,
|
| 182 |
+
"eval_wer": 0.9742561721882254,
|
| 183 |
+
"step": 984
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 6.1,
|
| 187 |
+
"learning_rate": 2e-05,
|
| 188 |
+
"loss": 2.4746,
|
| 189 |
+
"step": 1000
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 6.4,
|
| 193 |
+
"learning_rate": 1.9561403508771933e-05,
|
| 194 |
+
"loss": 2.353,
|
| 195 |
+
"step": 1050
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 6.71,
|
| 199 |
+
"learning_rate": 1.912280701754386e-05,
|
| 200 |
+
"loss": 2.1808,
|
| 201 |
+
"step": 1100
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 7.0,
|
| 205 |
+
"eval_cer": 0.986131055853677,
|
| 206 |
+
"eval_loss": 1.8554401397705078,
|
| 207 |
+
"eval_runtime": 21.7818,
|
| 208 |
+
"eval_samples_per_second": 120.467,
|
| 209 |
+
"eval_steps_per_second": 7.529,
|
| 210 |
+
"eval_wer": 0.9742561721882254,
|
| 211 |
+
"step": 1148
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 7.01,
|
| 215 |
+
"learning_rate": 1.868421052631579e-05,
|
| 216 |
+
"loss": 2.028,
|
| 217 |
+
"step": 1150
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"epoch": 7.32,
|
| 221 |
+
"learning_rate": 1.824561403508772e-05,
|
| 222 |
+
"loss": 1.8932,
|
| 223 |
+
"step": 1200
|
| 224 |
+
},
|
| 225 |
+
{
|
| 226 |
+
"epoch": 7.62,
|
| 227 |
+
"learning_rate": 1.780701754385965e-05,
|
| 228 |
+
"loss": 1.7557,
|
| 229 |
+
"step": 1250
|
| 230 |
+
},
|
| 231 |
+
{
|
| 232 |
+
"epoch": 7.93,
|
| 233 |
+
"learning_rate": 1.736842105263158e-05,
|
| 234 |
+
"loss": 1.6069,
|
| 235 |
+
"step": 1300
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 8.0,
|
| 239 |
+
"eval_cer": 0.6230821555437986,
|
| 240 |
+
"eval_loss": 1.2551288604736328,
|
| 241 |
+
"eval_runtime": 21.5405,
|
| 242 |
+
"eval_samples_per_second": 121.817,
|
| 243 |
+
"eval_steps_per_second": 7.614,
|
| 244 |
+
"eval_wer": 0.6822114370120278,
|
| 245 |
+
"step": 1312
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 8.23,
|
| 249 |
+
"learning_rate": 1.692982456140351e-05,
|
| 250 |
+
"loss": 1.4826,
|
| 251 |
+
"step": 1350
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 8.54,
|
| 255 |
+
"learning_rate": 1.649122807017544e-05,
|
| 256 |
+
"loss": 1.3003,
|
| 257 |
+
"step": 1400
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 8.84,
|
| 261 |
+
"learning_rate": 1.605263157894737e-05,
|
| 262 |
+
"loss": 1.1916,
|
| 263 |
+
"step": 1450
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 9.0,
|
| 267 |
+
"eval_cer": 0.22420829869246467,
|
| 268 |
+
"eval_loss": 0.7985360026359558,
|
| 269 |
+
"eval_runtime": 22.1233,
|
| 270 |
+
"eval_samples_per_second": 118.608,
|
| 271 |
+
"eval_steps_per_second": 7.413,
|
| 272 |
+
"eval_wer": 0.3679397903917845,
|
| 273 |
+
"step": 1476
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 9.15,
|
| 277 |
+
"learning_rate": 1.56140350877193e-05,
|
| 278 |
+
"loss": 1.1239,
|
| 279 |
+
"step": 1500
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 9.45,
|
| 283 |
+
"learning_rate": 1.517543859649123e-05,
|
| 284 |
+
"loss": 1.0425,
|
| 285 |
+
"step": 1550
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 9.76,
|
| 289 |
+
"learning_rate": 1.4736842105263159e-05,
|
| 290 |
+
"loss": 0.9977,
|
| 291 |
+
"step": 1600
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 10.0,
|
| 295 |
+
"eval_cer": 0.18267704633058726,
|
| 296 |
+
"eval_loss": 0.6234104633331299,
|
| 297 |
+
"eval_runtime": 21.4356,
|
| 298 |
+
"eval_samples_per_second": 122.413,
|
| 299 |
+
"eval_steps_per_second": 7.651,
|
| 300 |
+
"eval_wer": 0.31180980516283324,
|
| 301 |
+
"step": 1640
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 10.06,
|
| 305 |
+
"learning_rate": 1.429824561403509e-05,
|
| 306 |
+
"loss": 0.9625,
|
| 307 |
+
"step": 1650
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 10.37,
|
| 311 |
+
"learning_rate": 1.385964912280702e-05,
|
| 312 |
+
"loss": 0.9316,
|
| 313 |
+
"step": 1700
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 10.67,
|
| 317 |
+
"learning_rate": 1.3421052631578948e-05,
|
| 318 |
+
"loss": 0.8949,
|
| 319 |
+
"step": 1750
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 10.98,
|
| 323 |
+
"learning_rate": 1.2982456140350879e-05,
|
| 324 |
+
"loss": 0.836,
|
| 325 |
+
"step": 1800
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 11.0,
|
| 329 |
+
"eval_cer": 0.16431108759730934,
|
| 330 |
+
"eval_loss": 0.510288417339325,
|
| 331 |
+
"eval_runtime": 21.4424,
|
| 332 |
+
"eval_samples_per_second": 122.374,
|
| 333 |
+
"eval_steps_per_second": 7.648,
|
| 334 |
+
"eval_wer": 0.2800872195259197,
|
| 335 |
+
"step": 1804
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 11.28,
|
| 339 |
+
"learning_rate": 1.2543859649122808e-05,
|
| 340 |
+
"loss": 0.821,
|
| 341 |
+
"step": 1850
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 11.59,
|
| 345 |
+
"learning_rate": 1.2105263157894737e-05,
|
| 346 |
+
"loss": 0.7918,
|
| 347 |
+
"step": 1900
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 11.89,
|
| 351 |
+
"learning_rate": 1.1666666666666668e-05,
|
| 352 |
+
"loss": 0.7515,
|
| 353 |
+
"step": 1950
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 12.0,
|
| 357 |
+
"eval_cer": 0.1549391580379412,
|
| 358 |
+
"eval_loss": 0.43052738904953003,
|
| 359 |
+
"eval_runtime": 21.9755,
|
| 360 |
+
"eval_samples_per_second": 119.406,
|
| 361 |
+
"eval_steps_per_second": 7.463,
|
| 362 |
+
"eval_wer": 0.26630090736442286,
|
| 363 |
+
"step": 1968
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 12.2,
|
| 367 |
+
"learning_rate": 1.1228070175438597e-05,
|
| 368 |
+
"loss": 0.7517,
|
| 369 |
+
"step": 2000
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 12.5,
|
| 373 |
+
"learning_rate": 1.0789473684210528e-05,
|
| 374 |
+
"loss": 0.7086,
|
| 375 |
+
"step": 2050
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 12.8,
|
| 379 |
+
"learning_rate": 1.0350877192982459e-05,
|
| 380 |
+
"loss": 0.7045,
|
| 381 |
+
"step": 2100
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 13.0,
|
| 385 |
+
"eval_cer": 0.14133474416143904,
|
| 386 |
+
"eval_loss": 0.3688310384750366,
|
| 387 |
+
"eval_runtime": 21.9834,
|
| 388 |
+
"eval_samples_per_second": 119.363,
|
| 389 |
+
"eval_steps_per_second": 7.46,
|
| 390 |
+
"eval_wer": 0.24892734050784274,
|
| 391 |
+
"step": 2132
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 13.11,
|
| 395 |
+
"learning_rate": 9.912280701754386e-06,
|
| 396 |
+
"loss": 0.6799,
|
| 397 |
+
"step": 2150
|
| 398 |
+
},
|
| 399 |
+
{
|
| 400 |
+
"epoch": 13.41,
|
| 401 |
+
"learning_rate": 9.473684210526315e-06,
|
| 402 |
+
"loss": 0.6578,
|
| 403 |
+
"step": 2200
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 13.72,
|
| 407 |
+
"learning_rate": 9.035087719298246e-06,
|
| 408 |
+
"loss": 0.6533,
|
| 409 |
+
"step": 2250
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 14.0,
|
| 413 |
+
"eval_cer": 0.1339656866450004,
|
| 414 |
+
"eval_loss": 0.3257971704006195,
|
| 415 |
+
"eval_runtime": 21.7706,
|
| 416 |
+
"eval_samples_per_second": 120.529,
|
| 417 |
+
"eval_steps_per_second": 7.533,
|
| 418 |
+
"eval_wer": 0.2398536962791025,
|
| 419 |
+
"step": 2296
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 14.02,
|
| 423 |
+
"learning_rate": 8.596491228070176e-06,
|
| 424 |
+
"loss": 0.6267,
|
| 425 |
+
"step": 2300
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 14.33,
|
| 429 |
+
"learning_rate": 8.157894736842106e-06,
|
| 430 |
+
"loss": 0.6625,
|
| 431 |
+
"step": 2350
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 14.63,
|
| 435 |
+
"learning_rate": 7.719298245614036e-06,
|
| 436 |
+
"loss": 0.6162,
|
| 437 |
+
"step": 2400
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 14.94,
|
| 441 |
+
"learning_rate": 7.280701754385966e-06,
|
| 442 |
+
"loss": 0.5906,
|
| 443 |
+
"step": 2450
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 15.0,
|
| 447 |
+
"eval_cer": 0.1287884513642204,
|
| 448 |
+
"eval_loss": 0.29405683279037476,
|
| 449 |
+
"eval_runtime": 21.5389,
|
| 450 |
+
"eval_samples_per_second": 121.826,
|
| 451 |
+
"eval_steps_per_second": 7.614,
|
| 452 |
+
"eval_wer": 0.23183512696068087,
|
| 453 |
+
"step": 2460
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 15.24,
|
| 457 |
+
"learning_rate": 6.842105263157896e-06,
|
| 458 |
+
"loss": 0.6021,
|
| 459 |
+
"step": 2500
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 15.55,
|
| 463 |
+
"learning_rate": 6.403508771929825e-06,
|
| 464 |
+
"loss": 0.5722,
|
| 465 |
+
"step": 2550
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 15.85,
|
| 469 |
+
"learning_rate": 5.964912280701755e-06,
|
| 470 |
+
"loss": 0.5746,
|
| 471 |
+
"step": 2600
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 16.0,
|
| 475 |
+
"eval_cer": 0.12780591036202857,
|
| 476 |
+
"eval_loss": 0.2748357355594635,
|
| 477 |
+
"eval_runtime": 21.719,
|
| 478 |
+
"eval_samples_per_second": 120.816,
|
| 479 |
+
"eval_steps_per_second": 7.551,
|
| 480 |
+
"eval_wer": 0.23000633044946192,
|
| 481 |
+
"step": 2624
|
| 482 |
+
},
|
| 483 |
+
{
|
| 484 |
+
"epoch": 16.16,
|
| 485 |
+
"learning_rate": 5.526315789473685e-06,
|
| 486 |
+
"loss": 0.559,
|
| 487 |
+
"step": 2650
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 16.46,
|
| 491 |
+
"learning_rate": 5.087719298245615e-06,
|
| 492 |
+
"loss": 0.5676,
|
| 493 |
+
"step": 2700
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 16.77,
|
| 497 |
+
"learning_rate": 4.649122807017544e-06,
|
| 498 |
+
"loss": 0.5169,
|
| 499 |
+
"step": 2750
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 17.0,
|
| 503 |
+
"eval_cer": 0.12417806666162799,
|
| 504 |
+
"eval_loss": 0.25731268525123596,
|
| 505 |
+
"eval_runtime": 21.7044,
|
| 506 |
+
"eval_samples_per_second": 120.897,
|
| 507 |
+
"eval_steps_per_second": 7.556,
|
| 508 |
+
"eval_wer": 0.224027572624323,
|
| 509 |
+
"step": 2788
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 17.07,
|
| 513 |
+
"learning_rate": 4.210526315789474e-06,
|
| 514 |
+
"loss": 0.5603,
|
| 515 |
+
"step": 2800
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 17.38,
|
| 519 |
+
"learning_rate": 3.7719298245614037e-06,
|
| 520 |
+
"loss": 0.5362,
|
| 521 |
+
"step": 2850
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 17.68,
|
| 525 |
+
"learning_rate": 3.3333333333333333e-06,
|
| 526 |
+
"loss": 0.5187,
|
| 527 |
+
"step": 2900
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 17.99,
|
| 531 |
+
"learning_rate": 2.8947368421052634e-06,
|
| 532 |
+
"loss": 0.5511,
|
| 533 |
+
"step": 2950
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 18.0,
|
| 537 |
+
"eval_cer": 0.12277983523543194,
|
| 538 |
+
"eval_loss": 0.24789051711559296,
|
| 539 |
+
"eval_runtime": 21.566,
|
| 540 |
+
"eval_samples_per_second": 121.673,
|
| 541 |
+
"eval_steps_per_second": 7.605,
|
| 542 |
+
"eval_wer": 0.22107336287543083,
|
| 543 |
+
"step": 2952
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 18.29,
|
| 547 |
+
"learning_rate": 2.456140350877193e-06,
|
| 548 |
+
"loss": 0.541,
|
| 549 |
+
"step": 3000
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 18.6,
|
| 553 |
+
"learning_rate": 2.017543859649123e-06,
|
| 554 |
+
"loss": 0.5095,
|
| 555 |
+
"step": 3050
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 18.9,
|
| 559 |
+
"learning_rate": 1.5789473684210526e-06,
|
| 560 |
+
"loss": 0.5318,
|
| 561 |
+
"step": 3100
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 19.0,
|
| 565 |
+
"eval_cer": 0.12100370342377749,
|
| 566 |
+
"eval_loss": 0.2409585565328598,
|
| 567 |
+
"eval_runtime": 21.8589,
|
| 568 |
+
"eval_samples_per_second": 120.043,
|
| 569 |
+
"eval_steps_per_second": 7.503,
|
| 570 |
+
"eval_wer": 0.21861152141802068,
|
| 571 |
+
"step": 3116
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 19.21,
|
| 575 |
+
"learning_rate": 1.1403508771929824e-06,
|
| 576 |
+
"loss": 0.5156,
|
| 577 |
+
"step": 3150
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 19.51,
|
| 581 |
+
"learning_rate": 7.017543859649123e-07,
|
| 582 |
+
"loss": 0.4971,
|
| 583 |
+
"step": 3200
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 19.82,
|
| 587 |
+
"learning_rate": 2.6315789473684213e-07,
|
| 588 |
+
"loss": 0.5174,
|
| 589 |
+
"step": 3250
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 20.0,
|
| 593 |
+
"eval_cer": 0.12100370342377749,
|
| 594 |
+
"eval_loss": 0.23930250108242035,
|
| 595 |
+
"eval_runtime": 21.6599,
|
| 596 |
+
"eval_samples_per_second": 121.146,
|
| 597 |
+
"eval_steps_per_second": 7.572,
|
| 598 |
+
"eval_wer": 0.21868185974537527,
|
| 599 |
+
"step": 3280
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 20.0,
|
| 603 |
+
"step": 3280,
|
| 604 |
+
"total_flos": 1.8383341927882616e+18,
|
| 605 |
+
"train_loss": 1.9778692466456715,
|
| 606 |
+
"train_runtime": 7257.7038,
|
| 607 |
+
"train_samples_per_second": 28.924,
|
| 608 |
+
"train_steps_per_second": 0.452
|
| 609 |
+
}
|
| 610 |
+
],
|
| 611 |
+
"logging_steps": 50,
|
| 612 |
+
"max_steps": 3280,
|
| 613 |
+
"num_train_epochs": 20,
|
| 614 |
+
"save_steps": 500,
|
| 615 |
+
"total_flos": 1.8383341927882616e+18,
|
| 616 |
+
"trial_name": null,
|
| 617 |
+
"trial_params": null
|
| 618 |
+
}
|