Model save
Browse files- README.md +69 -0
- trainer_state.json +684 -0
README.md
ADDED
|
@@ -0,0 +1,69 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
base_model: jonatasgrosman/wav2vec2-large-xlsr-53-japanese
|
| 4 |
+
tags:
|
| 5 |
+
- generated_from_trainer
|
| 6 |
+
metrics:
|
| 7 |
+
- wer
|
| 8 |
+
model-index:
|
| 9 |
+
- name: Wav2vec_1
|
| 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 |
+
# Wav2vec_1
|
| 17 |
+
|
| 18 |
+
This model is a fine-tuned version of [jonatasgrosman/wav2vec2-large-xlsr-53-japanese](https://huggingface.co/jonatasgrosman/wav2vec2-large-xlsr-53-japanese) on the None dataset.
|
| 19 |
+
It achieves the following results on the evaluation set:
|
| 20 |
+
- Loss: 0.0459
|
| 21 |
+
- Wer: 0.2213
|
| 22 |
+
- Cer: 0.1608
|
| 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: 6e-05
|
| 42 |
+
- train_batch_size: 32
|
| 43 |
+
- eval_batch_size: 32
|
| 44 |
+
- seed: 4
|
| 45 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 46 |
+
- lr_scheduler_type: linear
|
| 47 |
+
- lr_scheduler_warmup_steps: 1000
|
| 48 |
+
- num_epochs: 8
|
| 49 |
+
|
| 50 |
+
### Training results
|
| 51 |
+
|
| 52 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|
| 53 |
+
|:-------------:|:-----:|:----:|:---------------:|:------:|:------:|
|
| 54 |
+
| 3.4904 | 1.0 | 120 | 3.4430 | 0.9970 | 0.9991 |
|
| 55 |
+
| 1.1939 | 2.0 | 240 | 1.0064 | 0.8270 | 0.6265 |
|
| 56 |
+
| 0.7726 | 3.0 | 360 | 0.6257 | 0.8198 | 0.5705 |
|
| 57 |
+
| 0.5502 | 4.0 | 480 | 0.4148 | 0.5910 | 0.3415 |
|
| 58 |
+
| 0.4152 | 5.0 | 600 | 0.2439 | 0.4167 | 0.2182 |
|
| 59 |
+
| 0.3159 | 6.0 | 720 | 0.1359 | 0.3084 | 0.1762 |
|
| 60 |
+
| 0.2425 | 7.0 | 840 | 0.0737 | 0.2523 | 0.1509 |
|
| 61 |
+
| 0.1921 | 8.0 | 960 | 0.0459 | 0.2213 | 0.1608 |
|
| 62 |
+
|
| 63 |
+
|
| 64 |
+
### Framework versions
|
| 65 |
+
|
| 66 |
+
- Transformers 4.35.2
|
| 67 |
+
- Pytorch 2.1.0+cu121
|
| 68 |
+
- Datasets 2.14.6
|
| 69 |
+
- Tokenizers 0.15.0
|
trainer_state.json
ADDED
|
@@ -0,0 +1,684 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 8.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 960,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.08,
|
| 13 |
+
"learning_rate": 6.000000000000001e-07,
|
| 14 |
+
"loss": 16.3378,
|
| 15 |
+
"step": 10
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.17,
|
| 19 |
+
"learning_rate": 1.2000000000000002e-06,
|
| 20 |
+
"loss": 16.7146,
|
| 21 |
+
"step": 20
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.25,
|
| 25 |
+
"learning_rate": 1.8e-06,
|
| 26 |
+
"loss": 16.2989,
|
| 27 |
+
"step": 30
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.33,
|
| 31 |
+
"learning_rate": 2.4000000000000003e-06,
|
| 32 |
+
"loss": 15.6823,
|
| 33 |
+
"step": 40
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.42,
|
| 37 |
+
"learning_rate": 3e-06,
|
| 38 |
+
"loss": 15.2213,
|
| 39 |
+
"step": 50
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.5,
|
| 43 |
+
"learning_rate": 3.6e-06,
|
| 44 |
+
"loss": 13.8045,
|
| 45 |
+
"step": 60
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.58,
|
| 49 |
+
"learning_rate": 4.2000000000000004e-06,
|
| 50 |
+
"loss": 12.3665,
|
| 51 |
+
"step": 70
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.67,
|
| 55 |
+
"learning_rate": 4.800000000000001e-06,
|
| 56 |
+
"loss": 9.9596,
|
| 57 |
+
"step": 80
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.75,
|
| 61 |
+
"learning_rate": 5.4e-06,
|
| 62 |
+
"loss": 7.1678,
|
| 63 |
+
"step": 90
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.83,
|
| 67 |
+
"learning_rate": 6e-06,
|
| 68 |
+
"loss": 5.6617,
|
| 69 |
+
"step": 100
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.92,
|
| 73 |
+
"learning_rate": 6.6e-06,
|
| 74 |
+
"loss": 4.5458,
|
| 75 |
+
"step": 110
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 1.0,
|
| 79 |
+
"learning_rate": 7.2e-06,
|
| 80 |
+
"loss": 3.4904,
|
| 81 |
+
"step": 120
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 1.0,
|
| 85 |
+
"eval_cer": 0.999086184248101,
|
| 86 |
+
"eval_loss": 3.4430177211761475,
|
| 87 |
+
"eval_runtime": 13.8851,
|
| 88 |
+
"eval_samples_per_second": 69.139,
|
| 89 |
+
"eval_steps_per_second": 2.161,
|
| 90 |
+
"eval_wer": 0.996996996996997,
|
| 91 |
+
"step": 120
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"epoch": 1.08,
|
| 95 |
+
"learning_rate": 7.8e-06,
|
| 96 |
+
"loss": 3.2384,
|
| 97 |
+
"step": 130
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"epoch": 1.17,
|
| 101 |
+
"learning_rate": 8.400000000000001e-06,
|
| 102 |
+
"loss": 2.8932,
|
| 103 |
+
"step": 140
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 1.25,
|
| 107 |
+
"learning_rate": 9e-06,
|
| 108 |
+
"loss": 2.7276,
|
| 109 |
+
"step": 150
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 1.33,
|
| 113 |
+
"learning_rate": 9.600000000000001e-06,
|
| 114 |
+
"loss": 2.6587,
|
| 115 |
+
"step": 160
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 1.42,
|
| 119 |
+
"learning_rate": 1.02e-05,
|
| 120 |
+
"loss": 2.3907,
|
| 121 |
+
"step": 170
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 1.5,
|
| 125 |
+
"learning_rate": 1.08e-05,
|
| 126 |
+
"loss": 2.22,
|
| 127 |
+
"step": 180
|
| 128 |
+
},
|
| 129 |
+
{
|
| 130 |
+
"epoch": 1.58,
|
| 131 |
+
"learning_rate": 1.1400000000000001e-05,
|
| 132 |
+
"loss": 2.0003,
|
| 133 |
+
"step": 190
|
| 134 |
+
},
|
| 135 |
+
{
|
| 136 |
+
"epoch": 1.67,
|
| 137 |
+
"learning_rate": 1.2e-05,
|
| 138 |
+
"loss": 1.8618,
|
| 139 |
+
"step": 200
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 1.75,
|
| 143 |
+
"learning_rate": 1.26e-05,
|
| 144 |
+
"loss": 1.6286,
|
| 145 |
+
"step": 210
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 1.83,
|
| 149 |
+
"learning_rate": 1.32e-05,
|
| 150 |
+
"loss": 1.4244,
|
| 151 |
+
"step": 220
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 1.92,
|
| 155 |
+
"learning_rate": 1.3800000000000002e-05,
|
| 156 |
+
"loss": 1.3615,
|
| 157 |
+
"step": 230
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 2.0,
|
| 161 |
+
"learning_rate": 1.44e-05,
|
| 162 |
+
"loss": 1.1939,
|
| 163 |
+
"step": 240
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 2.0,
|
| 167 |
+
"eval_cer": 0.6264778114112742,
|
| 168 |
+
"eval_loss": 1.006431221961975,
|
| 169 |
+
"eval_runtime": 15.0117,
|
| 170 |
+
"eval_samples_per_second": 63.95,
|
| 171 |
+
"eval_steps_per_second": 1.998,
|
| 172 |
+
"eval_wer": 0.8269519519519519,
|
| 173 |
+
"step": 240
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 2.08,
|
| 177 |
+
"learning_rate": 1.5e-05,
|
| 178 |
+
"loss": 1.3824,
|
| 179 |
+
"step": 250
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 2.17,
|
| 183 |
+
"learning_rate": 1.56e-05,
|
| 184 |
+
"loss": 1.0939,
|
| 185 |
+
"step": 260
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 2.25,
|
| 189 |
+
"learning_rate": 1.62e-05,
|
| 190 |
+
"loss": 1.0596,
|
| 191 |
+
"step": 270
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 2.33,
|
| 195 |
+
"learning_rate": 1.6800000000000002e-05,
|
| 196 |
+
"loss": 1.0087,
|
| 197 |
+
"step": 280
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 2.42,
|
| 201 |
+
"learning_rate": 1.74e-05,
|
| 202 |
+
"loss": 0.9328,
|
| 203 |
+
"step": 290
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 2.5,
|
| 207 |
+
"learning_rate": 1.8e-05,
|
| 208 |
+
"loss": 0.9045,
|
| 209 |
+
"step": 300
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 2.58,
|
| 213 |
+
"learning_rate": 1.86e-05,
|
| 214 |
+
"loss": 0.8645,
|
| 215 |
+
"step": 310
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 2.67,
|
| 219 |
+
"learning_rate": 1.9200000000000003e-05,
|
| 220 |
+
"loss": 0.8474,
|
| 221 |
+
"step": 320
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 2.75,
|
| 225 |
+
"learning_rate": 1.98e-05,
|
| 226 |
+
"loss": 0.7985,
|
| 227 |
+
"step": 330
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 2.83,
|
| 231 |
+
"learning_rate": 2.04e-05,
|
| 232 |
+
"loss": 0.7874,
|
| 233 |
+
"step": 340
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 2.92,
|
| 237 |
+
"learning_rate": 2.1e-05,
|
| 238 |
+
"loss": 0.8,
|
| 239 |
+
"step": 350
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 3.0,
|
| 243 |
+
"learning_rate": 2.16e-05,
|
| 244 |
+
"loss": 0.7726,
|
| 245 |
+
"step": 360
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 3.0,
|
| 249 |
+
"eval_cer": 0.5705065966074591,
|
| 250 |
+
"eval_loss": 0.6256773471832275,
|
| 251 |
+
"eval_runtime": 13.9464,
|
| 252 |
+
"eval_samples_per_second": 68.835,
|
| 253 |
+
"eval_steps_per_second": 2.151,
|
| 254 |
+
"eval_wer": 0.8198198198198198,
|
| 255 |
+
"step": 360
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 3.08,
|
| 259 |
+
"learning_rate": 2.22e-05,
|
| 260 |
+
"loss": 0.7963,
|
| 261 |
+
"step": 370
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 3.17,
|
| 265 |
+
"learning_rate": 2.2800000000000002e-05,
|
| 266 |
+
"loss": 0.7342,
|
| 267 |
+
"step": 380
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 3.25,
|
| 271 |
+
"learning_rate": 2.3400000000000003e-05,
|
| 272 |
+
"loss": 0.7324,
|
| 273 |
+
"step": 390
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 3.33,
|
| 277 |
+
"learning_rate": 2.4e-05,
|
| 278 |
+
"loss": 0.6865,
|
| 279 |
+
"step": 400
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 3.42,
|
| 283 |
+
"learning_rate": 2.4599999999999998e-05,
|
| 284 |
+
"loss": 0.6731,
|
| 285 |
+
"step": 410
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 3.5,
|
| 289 |
+
"learning_rate": 2.52e-05,
|
| 290 |
+
"loss": 0.6683,
|
| 291 |
+
"step": 420
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 3.58,
|
| 295 |
+
"learning_rate": 2.58e-05,
|
| 296 |
+
"loss": 0.6583,
|
| 297 |
+
"step": 430
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 3.67,
|
| 301 |
+
"learning_rate": 2.64e-05,
|
| 302 |
+
"loss": 0.6218,
|
| 303 |
+
"step": 440
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 3.75,
|
| 307 |
+
"learning_rate": 2.7000000000000002e-05,
|
| 308 |
+
"loss": 0.5825,
|
| 309 |
+
"step": 450
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 3.83,
|
| 313 |
+
"learning_rate": 2.7600000000000003e-05,
|
| 314 |
+
"loss": 0.5552,
|
| 315 |
+
"step": 460
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 3.92,
|
| 319 |
+
"learning_rate": 2.8199999999999998e-05,
|
| 320 |
+
"loss": 0.6122,
|
| 321 |
+
"step": 470
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 4.0,
|
| 325 |
+
"learning_rate": 2.88e-05,
|
| 326 |
+
"loss": 0.5502,
|
| 327 |
+
"step": 480
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 4.0,
|
| 331 |
+
"eval_cer": 0.34153863727225997,
|
| 332 |
+
"eval_loss": 0.41475021839141846,
|
| 333 |
+
"eval_runtime": 13.9543,
|
| 334 |
+
"eval_samples_per_second": 68.796,
|
| 335 |
+
"eval_steps_per_second": 2.15,
|
| 336 |
+
"eval_wer": 0.5910285285285285,
|
| 337 |
+
"step": 480
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 4.08,
|
| 341 |
+
"learning_rate": 2.94e-05,
|
| 342 |
+
"loss": 0.5787,
|
| 343 |
+
"step": 490
|
| 344 |
+
},
|
| 345 |
+
{
|
| 346 |
+
"epoch": 4.17,
|
| 347 |
+
"learning_rate": 3e-05,
|
| 348 |
+
"loss": 0.4964,
|
| 349 |
+
"step": 500
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 4.25,
|
| 353 |
+
"learning_rate": 3.06e-05,
|
| 354 |
+
"loss": 0.5245,
|
| 355 |
+
"step": 510
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 4.33,
|
| 359 |
+
"learning_rate": 3.12e-05,
|
| 360 |
+
"loss": 0.4688,
|
| 361 |
+
"step": 520
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 4.42,
|
| 365 |
+
"learning_rate": 3.18e-05,
|
| 366 |
+
"loss": 0.5043,
|
| 367 |
+
"step": 530
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 4.5,
|
| 371 |
+
"learning_rate": 3.24e-05,
|
| 372 |
+
"loss": 0.4769,
|
| 373 |
+
"step": 540
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 4.58,
|
| 377 |
+
"learning_rate": 3.3e-05,
|
| 378 |
+
"loss": 0.4966,
|
| 379 |
+
"step": 550
|
| 380 |
+
},
|
| 381 |
+
{
|
| 382 |
+
"epoch": 4.67,
|
| 383 |
+
"learning_rate": 3.3600000000000004e-05,
|
| 384 |
+
"loss": 0.4772,
|
| 385 |
+
"step": 560
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 4.75,
|
| 389 |
+
"learning_rate": 3.42e-05,
|
| 390 |
+
"loss": 0.4364,
|
| 391 |
+
"step": 570
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 4.83,
|
| 395 |
+
"learning_rate": 3.48e-05,
|
| 396 |
+
"loss": 0.417,
|
| 397 |
+
"step": 580
|
| 398 |
+
},
|
| 399 |
+
{
|
| 400 |
+
"epoch": 4.92,
|
| 401 |
+
"learning_rate": 3.54e-05,
|
| 402 |
+
"loss": 0.4407,
|
| 403 |
+
"step": 590
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 5.0,
|
| 407 |
+
"learning_rate": 3.6e-05,
|
| 408 |
+
"loss": 0.4152,
|
| 409 |
+
"step": 600
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 5.0,
|
| 413 |
+
"eval_cer": 0.21817351076589184,
|
| 414 |
+
"eval_loss": 0.24392470717430115,
|
| 415 |
+
"eval_runtime": 13.9841,
|
| 416 |
+
"eval_samples_per_second": 68.65,
|
| 417 |
+
"eval_steps_per_second": 2.145,
|
| 418 |
+
"eval_wer": 0.4166666666666667,
|
| 419 |
+
"step": 600
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 5.08,
|
| 423 |
+
"learning_rate": 3.66e-05,
|
| 424 |
+
"loss": 0.364,
|
| 425 |
+
"step": 610
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 5.17,
|
| 429 |
+
"learning_rate": 3.72e-05,
|
| 430 |
+
"loss": 0.4115,
|
| 431 |
+
"step": 620
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 5.25,
|
| 435 |
+
"learning_rate": 3.7800000000000004e-05,
|
| 436 |
+
"loss": 0.3769,
|
| 437 |
+
"step": 630
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 5.33,
|
| 441 |
+
"learning_rate": 3.8400000000000005e-05,
|
| 442 |
+
"loss": 0.3635,
|
| 443 |
+
"step": 640
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 5.42,
|
| 447 |
+
"learning_rate": 3.9e-05,
|
| 448 |
+
"loss": 0.3743,
|
| 449 |
+
"step": 650
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 5.5,
|
| 453 |
+
"learning_rate": 3.96e-05,
|
| 454 |
+
"loss": 0.3188,
|
| 455 |
+
"step": 660
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 5.58,
|
| 459 |
+
"learning_rate": 4.02e-05,
|
| 460 |
+
"loss": 0.3608,
|
| 461 |
+
"step": 670
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 5.67,
|
| 465 |
+
"learning_rate": 4.08e-05,
|
| 466 |
+
"loss": 0.3285,
|
| 467 |
+
"step": 680
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 5.75,
|
| 471 |
+
"learning_rate": 4.14e-05,
|
| 472 |
+
"loss": 0.2964,
|
| 473 |
+
"step": 690
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 5.83,
|
| 477 |
+
"learning_rate": 4.2e-05,
|
| 478 |
+
"loss": 0.2799,
|
| 479 |
+
"step": 700
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 5.92,
|
| 483 |
+
"learning_rate": 4.26e-05,
|
| 484 |
+
"loss": 0.3272,
|
| 485 |
+
"step": 710
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 6.0,
|
| 489 |
+
"learning_rate": 4.32e-05,
|
| 490 |
+
"loss": 0.3159,
|
| 491 |
+
"step": 720
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 6.0,
|
| 495 |
+
"eval_cer": 0.17619509966303043,
|
| 496 |
+
"eval_loss": 0.13585154712200165,
|
| 497 |
+
"eval_runtime": 13.8057,
|
| 498 |
+
"eval_samples_per_second": 69.536,
|
| 499 |
+
"eval_steps_per_second": 2.173,
|
| 500 |
+
"eval_wer": 0.3083708708708709,
|
| 501 |
+
"step": 720
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 6.08,
|
| 505 |
+
"learning_rate": 4.38e-05,
|
| 506 |
+
"loss": 0.3256,
|
| 507 |
+
"step": 730
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 6.17,
|
| 511 |
+
"learning_rate": 4.44e-05,
|
| 512 |
+
"loss": 0.2651,
|
| 513 |
+
"step": 740
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 6.25,
|
| 517 |
+
"learning_rate": 4.5e-05,
|
| 518 |
+
"loss": 0.2502,
|
| 519 |
+
"step": 750
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 6.33,
|
| 523 |
+
"learning_rate": 4.5600000000000004e-05,
|
| 524 |
+
"loss": 0.2632,
|
| 525 |
+
"step": 760
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 6.42,
|
| 529 |
+
"learning_rate": 4.6200000000000005e-05,
|
| 530 |
+
"loss": 0.2412,
|
| 531 |
+
"step": 770
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 6.5,
|
| 535 |
+
"learning_rate": 4.6800000000000006e-05,
|
| 536 |
+
"loss": 0.2871,
|
| 537 |
+
"step": 780
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 6.58,
|
| 541 |
+
"learning_rate": 4.74e-05,
|
| 542 |
+
"loss": 0.2409,
|
| 543 |
+
"step": 790
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 6.67,
|
| 547 |
+
"learning_rate": 4.8e-05,
|
| 548 |
+
"loss": 0.2091,
|
| 549 |
+
"step": 800
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 6.75,
|
| 553 |
+
"learning_rate": 4.86e-05,
|
| 554 |
+
"loss": 0.2677,
|
| 555 |
+
"step": 810
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 6.83,
|
| 559 |
+
"learning_rate": 4.9199999999999997e-05,
|
| 560 |
+
"loss": 0.2109,
|
| 561 |
+
"step": 820
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 6.92,
|
| 565 |
+
"learning_rate": 4.98e-05,
|
| 566 |
+
"loss": 0.1886,
|
| 567 |
+
"step": 830
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 7.0,
|
| 571 |
+
"learning_rate": 5.04e-05,
|
| 572 |
+
"loss": 0.2425,
|
| 573 |
+
"step": 840
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 7.0,
|
| 577 |
+
"eval_cer": 0.15089382603232623,
|
| 578 |
+
"eval_loss": 0.07371211796998978,
|
| 579 |
+
"eval_runtime": 13.9128,
|
| 580 |
+
"eval_samples_per_second": 69.001,
|
| 581 |
+
"eval_steps_per_second": 2.156,
|
| 582 |
+
"eval_wer": 0.25225225225225223,
|
| 583 |
+
"step": 840
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 7.08,
|
| 587 |
+
"learning_rate": 5.1e-05,
|
| 588 |
+
"loss": 0.2069,
|
| 589 |
+
"step": 850
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 7.17,
|
| 593 |
+
"learning_rate": 5.16e-05,
|
| 594 |
+
"loss": 0.1888,
|
| 595 |
+
"step": 860
|
| 596 |
+
},
|
| 597 |
+
{
|
| 598 |
+
"epoch": 7.25,
|
| 599 |
+
"learning_rate": 5.22e-05,
|
| 600 |
+
"loss": 0.1926,
|
| 601 |
+
"step": 870
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 7.33,
|
| 605 |
+
"learning_rate": 5.28e-05,
|
| 606 |
+
"loss": 0.1794,
|
| 607 |
+
"step": 880
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 7.42,
|
| 611 |
+
"learning_rate": 5.3400000000000004e-05,
|
| 612 |
+
"loss": 0.2078,
|
| 613 |
+
"step": 890
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"epoch": 7.5,
|
| 617 |
+
"learning_rate": 5.4000000000000005e-05,
|
| 618 |
+
"loss": 0.1646,
|
| 619 |
+
"step": 900
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 7.58,
|
| 623 |
+
"learning_rate": 5.4600000000000006e-05,
|
| 624 |
+
"loss": 0.1417,
|
| 625 |
+
"step": 910
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 7.67,
|
| 629 |
+
"learning_rate": 5.520000000000001e-05,
|
| 630 |
+
"loss": 0.1454,
|
| 631 |
+
"step": 920
|
| 632 |
+
},
|
| 633 |
+
{
|
| 634 |
+
"epoch": 7.75,
|
| 635 |
+
"learning_rate": 5.58e-05,
|
| 636 |
+
"loss": 0.173,
|
| 637 |
+
"step": 930
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"epoch": 7.83,
|
| 641 |
+
"learning_rate": 5.6399999999999995e-05,
|
| 642 |
+
"loss": 0.1449,
|
| 643 |
+
"step": 940
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 7.92,
|
| 647 |
+
"learning_rate": 5.6999999999999996e-05,
|
| 648 |
+
"loss": 0.1359,
|
| 649 |
+
"step": 950
|
| 650 |
+
},
|
| 651 |
+
{
|
| 652 |
+
"epoch": 8.0,
|
| 653 |
+
"learning_rate": 5.76e-05,
|
| 654 |
+
"loss": 0.1921,
|
| 655 |
+
"step": 960
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 8.0,
|
| 659 |
+
"eval_cer": 0.16083157233422812,
|
| 660 |
+
"eval_loss": 0.045883145183324814,
|
| 661 |
+
"eval_runtime": 13.9631,
|
| 662 |
+
"eval_samples_per_second": 68.753,
|
| 663 |
+
"eval_steps_per_second": 2.149,
|
| 664 |
+
"eval_wer": 0.22128378378378377,
|
| 665 |
+
"step": 960
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 8.0,
|
| 669 |
+
"step": 960,
|
| 670 |
+
"total_flos": 2.451994748193635e+18,
|
| 671 |
+
"train_loss": 2.0509253946443398,
|
| 672 |
+
"train_runtime": 1630.9039,
|
| 673 |
+
"train_samples_per_second": 18.836,
|
| 674 |
+
"train_steps_per_second": 0.589
|
| 675 |
+
}
|
| 676 |
+
],
|
| 677 |
+
"logging_steps": 10,
|
| 678 |
+
"max_steps": 960,
|
| 679 |
+
"num_train_epochs": 8,
|
| 680 |
+
"save_steps": 500,
|
| 681 |
+
"total_flos": 2.451994748193635e+18,
|
| 682 |
+
"trial_name": null,
|
| 683 |
+
"trial_params": null
|
| 684 |
+
}
|