File size: 6,854 Bytes
65977bf 1bfa1f3 65977bf 1bfa1f3 65977bf |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 |
---
license: apache-2.0
base_model: facebook/wav2vec2-xls-r-300m
tags:
- generated_from_trainer
datasets:
- audiofolder
metrics:
- wer
model-index:
- name: wav2vec2-xls-r-300m-ja-syllable-cv-14
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: audiofolder
type: audiofolder
config: default
split: train[:20%]
args: default
metrics:
- name: Wer
type: wer
value: 0.04376879385232209
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# wav2vec2-xls-r-300m-ja-syllable-cv-14
This model is a fine-tuned version of [facebook/wav2vec2-xls-r-300m](https://huggingface.co./facebook/wav2vec2-xls-r-300m) on Japanese using the train, dev, and validation splits of Common Voice 14.0.
It achieves the following results on the evaluation set:
- Loss: 0.2005
- Wer: 0.0438
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
Training: Common Voice 14.0 ja train, dev, validated
Test: Common Voice 14.0 ja test[:20%]
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0001
- train_batch_size: 2
- eval_batch_size: 1
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 8
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 30
### Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|:-------------:|:-----:|:------:|:---------------:|:------:|
| 3.9103 | 0.37 | 2000 | 0.6143 | 0.1658 |
| 0.5883 | 0.75 | 4000 | 0.4720 | 0.1340 |
| 0.4759 | 1.12 | 6000 | 0.4080 | 0.1193 |
| 0.4115 | 1.49 | 8000 | 0.3758 | 0.1173 |
| 0.3833 | 1.87 | 10000 | 0.3591 | 0.1134 |
| 0.3351 | 2.24 | 12000 | 0.3440 | 0.1011 |
| 0.3129 | 2.61 | 14000 | 0.3550 | 0.1001 |
| 0.3016 | 2.99 | 16000 | 0.3041 | 0.0949 |
| 0.262 | 3.36 | 18000 | 0.2885 | 0.0853 |
| 0.2571 | 3.73 | 20000 | 0.2825 | 0.0874 |
| 0.2382 | 4.1 | 22000 | 0.2816 | 0.0848 |
| 0.2171 | 4.48 | 24000 | 0.2732 | 0.0770 |
| 0.2116 | 4.85 | 26000 | 0.2665 | 0.0773 |
| 0.1964 | 5.22 | 28000 | 0.2703 | 0.0819 |
| 0.1905 | 5.6 | 30000 | 0.2748 | 0.0822 |
| 0.1855 | 5.97 | 32000 | 0.2572 | 0.0757 |
| 0.1653 | 6.34 | 34000 | 0.2964 | 0.0803 |
| 0.1684 | 6.72 | 36000 | 0.2744 | 0.0745 |
| 0.1661 | 7.09 | 38000 | 0.2640 | 0.0790 |
| 0.1504 | 7.46 | 40000 | 0.2803 | 0.0785 |
| 0.1555 | 7.84 | 42000 | 0.2459 | 0.0703 |
| 0.1408 | 8.21 | 44000 | 0.2666 | 0.0736 |
| 0.1343 | 8.58 | 46000 | 0.2546 | 0.0711 |
| 0.1358 | 8.96 | 48000 | 0.2582 | 0.0691 |
| 0.1256 | 9.33 | 50000 | 0.2616 | 0.0709 |
| 0.1245 | 9.7 | 52000 | 0.2621 | 0.0712 |
| 0.1195 | 10.07 | 54000 | 0.2819 | 0.0692 |
| 0.1122 | 10.45 | 56000 | 0.2666 | 0.0699 |
| 0.1094 | 10.82 | 58000 | 0.2504 | 0.0666 |
| 0.1062 | 11.19 | 60000 | 0.2610 | 0.0666 |
| 0.1023 | 11.57 | 62000 | 0.2586 | 0.0656 |
| 0.1036 | 11.94 | 64000 | 0.2463 | 0.0646 |
| 0.096 | 12.31 | 66000 | 0.2677 | 0.0676 |
| 0.0942 | 12.69 | 68000 | 0.2284 | 0.0607 |
| 0.0939 | 13.06 | 70000 | 0.2663 | 0.0658 |
| 0.0857 | 13.43 | 72000 | 0.2583 | 0.0653 |
| 0.0889 | 13.81 | 74000 | 0.2215 | 0.0616 |
| 0.0832 | 14.18 | 76000 | 0.2502 | 0.0631 |
| 0.0813 | 14.55 | 78000 | 0.2472 | 0.0638 |
| 0.0796 | 14.93 | 80000 | 0.2218 | 0.0600 |
| 0.0774 | 15.3 | 82000 | 0.2376 | 0.0600 |
| 0.0754 | 15.67 | 84000 | 0.2361 | 0.0588 |
| 0.0745 | 16.04 | 86000 | 0.2578 | 0.0618 |
| 0.0722 | 16.42 | 88000 | 0.2468 | 0.0604 |
| 0.0709 | 16.79 | 90000 | 0.2268 | 0.0597 |
| 0.0688 | 17.16 | 92000 | 0.2270 | 0.0555 |
| 0.0665 | 17.54 | 94000 | 0.2320 | 0.0565 |
| 0.0651 | 17.91 | 96000 | 0.2408 | 0.0600 |
| 0.062 | 18.28 | 98000 | 0.2286 | 0.0550 |
| 0.0609 | 18.66 | 100000 | 0.2314 | 0.0558 |
| 0.0598 | 19.03 | 102000 | 0.2275 | 0.0547 |
| 0.057 | 19.4 | 104000 | 0.2359 | 0.0547 |
| 0.0559 | 19.78 | 106000 | 0.2501 | 0.0565 |
| 0.0557 | 20.15 | 108000 | 0.2186 | 0.0530 |
| 0.0519 | 20.52 | 110000 | 0.2281 | 0.0520 |
| 0.0532 | 20.9 | 112000 | 0.2342 | 0.0525 |
| 0.0521 | 21.27 | 114000 | 0.2265 | 0.0527 |
| 0.0513 | 21.64 | 116000 | 0.2263 | 0.0528 |
| 0.0485 | 22.01 | 118000 | 0.2343 | 0.0535 |
| 0.0454 | 22.39 | 120000 | 0.2393 | 0.0517 |
| 0.0454 | 22.76 | 122000 | 0.2314 | 0.0520 |
| 0.0448 | 23.13 | 124000 | 0.2395 | 0.0493 |
| 0.0444 | 23.51 | 126000 | 0.2299 | 0.0509 |
| 0.0434 | 23.88 | 128000 | 0.2300 | 0.0499 |
| 0.0402 | 24.25 | 130000 | 0.2314 | 0.0498 |
| 0.0395 | 24.63 | 132000 | 0.2259 | 0.0478 |
| 0.0383 | 25.0 | 134000 | 0.2202 | 0.0481 |
| 0.0374 | 25.37 | 136000 | 0.2158 | 0.0484 |
| 0.0375 | 25.75 | 138000 | 0.2165 | 0.0471 |
| 0.0366 | 26.12 | 140000 | 0.2142 | 0.0469 |
| 0.0347 | 26.49 | 142000 | 0.2139 | 0.0468 |
| 0.0337 | 26.87 | 144000 | 0.2152 | 0.0477 |
| 0.0343 | 27.24 | 146000 | 0.2059 | 0.0463 |
| 0.0328 | 27.61 | 148000 | 0.2108 | 0.0469 |
| 0.0324 | 27.99 | 150000 | 0.2061 | 0.0453 |
| 0.0302 | 28.36 | 152000 | 0.2026 | 0.0450 |
| 0.0316 | 28.73 | 154000 | 0.2057 | 0.0450 |
| 0.0298 | 29.1 | 156000 | 0.2005 | 0.0439 |
| 0.0301 | 29.48 | 158000 | 0.1983 | 0.0440 |
| 0.0296 | 29.85 | 160000 | 0.2005 | 0.0438 |
### Framework versions
- Transformers 4.31.0
- Pytorch 2.0.1+cu117
- Datasets 2.14.3
- Tokenizers 0.13.3
|