File size: 4,561 Bytes
96ed779 |
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 |
---
language:
- ko
license: apache-2.0
base_model: openai/whisper-small
tags:
- hf-asr-leaderboard
- generated_from_trainer
datasets:
- arrow
metrics:
- wer
model-index:
- name: whisper-kor3
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: whisper-kor3
type: arrow
config: default
split: train
args: 'config: ko, split: valid'
metrics:
- name: Wer
type: wer
value: 24.690290982425815
---
<!-- 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. -->
# whisper-kor3
This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the whisper-kor3 dataset.
It achieves the following results on the evaluation set:
- Loss: 0.4157
- Wer: 24.6903
- Cer: 11.3851
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 4000
### Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|:-------------:|:-----:|:----:|:---------------:|:-------:|:-------:|
| 1.2195 | 0.05 | 100 | 1.0198 | 34.4857 | 16.2544 |
| 0.7295 | 0.09 | 200 | 0.7220 | 32.6995 | 14.9684 |
| 0.5236 | 0.14 | 300 | 0.5703 | 31.4463 | 14.2549 |
| 0.4976 | 0.18 | 400 | 0.5461 | 31.8640 | 14.6274 |
| 0.479 | 0.23 | 500 | 0.5296 | 30.4091 | 14.0902 |
| 0.4544 | 0.28 | 600 | 0.5219 | 31.7920 | 16.3916 |
| 0.4672 | 0.32 | 700 | 0.5100 | 30.4955 | 13.9138 |
| 0.4305 | 0.37 | 800 | 0.5043 | 30.1354 | 14.5960 |
| 0.4561 | 0.42 | 900 | 0.4941 | 28.8101 | 13.2513 |
| 0.398 | 0.46 | 1000 | 0.4846 | 31.3166 | 14.2980 |
| 0.4338 | 0.51 | 1100 | 0.4780 | 28.0755 | 12.8945 |
| 0.4121 | 0.55 | 1200 | 0.4728 | 27.4128 | 12.5417 |
| 0.4217 | 0.6 | 1300 | 0.4693 | 28.2772 | 14.4392 |
| 0.3881 | 0.65 | 1400 | 0.4639 | 27.6577 | 13.0082 |
| 0.4035 | 0.69 | 1500 | 0.4593 | 26.9231 | 12.4436 |
| 0.4146 | 0.74 | 1600 | 0.4555 | 28.4212 | 13.7609 |
| 0.3837 | 0.78 | 1700 | 0.4511 | 28.8822 | 13.7845 |
| 0.3969 | 0.83 | 1800 | 0.4485 | 29.2135 | 14.2235 |
| 0.4368 | 0.88 | 1900 | 0.4414 | 26.5918 | 12.1457 |
| 0.3679 | 0.92 | 2000 | 0.4376 | 26.4477 | 12.1770 |
| 0.4496 | 0.97 | 2100 | 0.4335 | 30.1354 | 14.9018 |
| 0.3049 | 1.02 | 2200 | 0.4314 | 26.1164 | 12.9180 |
| 0.2213 | 1.06 | 2300 | 0.4325 | 25.9147 | 11.8046 |
| 0.2732 | 1.11 | 2400 | 0.4303 | 26.0012 | 11.8987 |
| 0.2568 | 1.15 | 2500 | 0.4293 | 25.9291 | 11.7576 |
| 0.2456 | 1.2 | 2600 | 0.4289 | 25.6986 | 11.7066 |
| 0.2702 | 1.25 | 2700 | 0.4262 | 25.8283 | 11.8203 |
| 0.2744 | 1.29 | 2800 | 0.4235 | 25.8139 | 11.8124 |
| 0.2742 | 1.34 | 2900 | 0.4254 | 25.6266 | 11.6360 |
| 0.2798 | 1.39 | 3000 | 0.4238 | 25.5546 | 11.6399 |
| 0.2593 | 1.43 | 3100 | 0.4219 | 26.1020 | 12.4632 |
| 0.2619 | 1.48 | 3200 | 0.4208 | 25.3241 | 11.4714 |
| 0.2633 | 1.52 | 3300 | 0.4210 | 26.6350 | 12.9964 |
| 0.2603 | 1.57 | 3400 | 0.4189 | 25.2809 | 11.4243 |
| 0.2992 | 1.62 | 3500 | 0.4189 | 25.2377 | 11.3969 |
| 0.2453 | 1.66 | 3600 | 0.4176 | 25.2377 | 11.5145 |
| 0.2475 | 1.71 | 3700 | 0.4172 | 24.8487 | 11.3969 |
| 0.2545 | 1.75 | 3800 | 0.4164 | 25.0216 | 11.4596 |
| 0.272 | 1.8 | 3900 | 0.4160 | 24.6471 | 11.2714 |
| 0.2339 | 1.85 | 4000 | 0.4157 | 24.6903 | 11.3851 |
### Framework versions
- Transformers 4.33.2
- Pytorch 2.0.1+cu117
- Datasets 2.14.5
- Tokenizers 0.13.3
|