File size: 4,489 Bytes
f02ecff 892e025 f02ecff |
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 |
---
language:
- ko
license: apache-2.0
tags:
- hf-asr-leaderboard
- generated_from_trainer
base_model: openai/whisper-small
datasets:
- korean_samll_dataset4
model-index:
- name: korean-small_t35
results: []
---
<!-- 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. -->
# korean-small_t35
This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the korean_samll_dataset4 dataset.
It achieves the following results on the evaluation set:
- Loss: 0.1385
- Cer: 5.2553
## 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: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 3.0
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Cer |
|:-------------:|:-----:|:-----:|:---------------:|:------:|
| 0.2501 | 0.06 | 200 | 0.2322 | 8.7579 |
| 0.217 | 0.12 | 400 | 0.2118 | 8.6072 |
| 0.1947 | 0.18 | 600 | 0.2011 | 7.4035 |
| 0.1938 | 0.24 | 800 | 0.1941 | 7.3665 |
| 0.1878 | 0.3 | 1000 | 0.1826 | 7.0860 |
| 0.192 | 0.36 | 1200 | 0.1786 | 6.8894 |
| 0.1768 | 0.42 | 1400 | 0.1739 | 6.5072 |
| 0.1777 | 0.48 | 1600 | 0.1708 | 6.4205 |
| 0.1714 | 0.54 | 1800 | 0.1675 | 6.6288 |
| 0.171 | 0.6 | 2000 | 0.1637 | 6.3026 |
| 0.1678 | 0.66 | 2200 | 0.1638 | 6.4964 |
| 0.1606 | 0.72 | 2400 | 0.1604 | 6.4205 |
| 0.1541 | 0.78 | 2600 | 0.1580 | 6.1524 |
| 0.1578 | 0.84 | 2800 | 0.1550 | 5.8736 |
| 0.1524 | 0.9 | 3000 | 0.1535 | 5.9458 |
| 0.153 | 0.96 | 3200 | 0.1512 | 5.8205 |
| 0.112 | 1.02 | 3400 | 0.1492 | 5.7590 |
| 0.0833 | 1.08 | 3600 | 0.1491 | 5.7022 |
| 0.0928 | 1.14 | 3800 | 0.1495 | 5.6578 |
| 0.1005 | 1.2 | 4000 | 0.1480 | 6.0906 |
| 0.0918 | 1.26 | 4200 | 0.1475 | 5.8175 |
| 0.0929 | 1.32 | 4400 | 0.1470 | 5.7632 |
| 0.091 | 1.38 | 4600 | 0.1460 | 5.6557 |
| 0.0858 | 1.44 | 4800 | 0.1445 | 5.6947 |
| 0.0889 | 1.5 | 5000 | 0.1435 | 5.6632 |
| 0.0903 | 1.56 | 5200 | 0.1442 | 5.6412 |
| 0.0894 | 1.61 | 5400 | 0.1426 | 5.5711 |
| 0.0842 | 1.67 | 5600 | 0.1426 | 5.4424 |
| 0.0926 | 1.73 | 5800 | 0.1419 | 5.4171 |
| 0.0801 | 1.79 | 6000 | 0.1400 | 5.3960 |
| 0.0843 | 1.85 | 6200 | 0.1397 | 5.5648 |
| 0.0909 | 1.91 | 6400 | 0.1386 | 5.4677 |
| 0.0816 | 1.97 | 6600 | 0.1384 | 5.6586 |
| 0.0484 | 2.03 | 6800 | 0.1421 | 5.4541 |
| 0.0506 | 2.09 | 7000 | 0.1408 | 5.4424 |
| 0.0475 | 2.15 | 7200 | 0.1410 | 5.5565 |
| 0.0477 | 2.21 | 7400 | 0.1406 | 5.5453 |
| 0.0465 | 2.27 | 7600 | 0.1407 | 5.3383 |
| 0.0487 | 2.33 | 7800 | 0.1404 | 5.4192 |
| 0.0438 | 2.39 | 8000 | 0.1400 | 5.4088 |
| 0.0432 | 2.45 | 8200 | 0.1404 | 5.4022 |
| 0.0457 | 2.51 | 8400 | 0.1410 | 5.3852 |
| 0.0468 | 2.57 | 8600 | 0.1398 | 5.2881 |
| 0.0456 | 2.63 | 8800 | 0.1390 | 5.2789 |
| 0.0426 | 2.69 | 9000 | 0.1390 | 5.3619 |
| 0.0437 | 2.75 | 9200 | 0.1385 | 5.2553 |
| 0.0467 | 2.81 | 9400 | 0.1386 | 5.3404 |
| 0.044 | 2.87 | 9600 | 0.1383 | 5.3180 |
| 0.0445 | 2.93 | 9800 | 0.1382 | 5.3072 |
| 0.0453 | 2.99 | 10000 | 0.1380 | 5.3267 |
### Framework versions
- Transformers 4.39.0.dev0
- Pytorch 2.2.1+cu121
- Datasets 2.18.0
- Tokenizers 0.15.2
|