File size: 1,628 Bytes
4abc1d2 f9d18d1 4abc1d2 ad16c50 908a208 ad16c50 4abc1d2 fe8245b 4abc1d2 908a208 4abc1d2 f9d18d1 4abc1d2 f9d18d1 ad16c50 f9d18d1 4abc1d2 |
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 |
---
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: wav2vec2-large-asr-th-2
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. -->
# wav2vec2-large-asr-th-2
This model was trained from scratch on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.2590
- Wer: 0.4210
- Cer: 0.1146
## 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: 0.0002
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 3
- total_train_batch_size: 48
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 600
- training_steps: 4000
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|:-------------:|:-----:|:----:|:---------------:|:------:|:------:|
| 0.6189 | 0.24 | 1000 | 0.3571 | 0.5060 | 0.1449 |
| 0.5435 | 0.47 | 2000 | 0.3255 | 0.4712 | 0.1328 |
| 0.5612 | 0.71 | 3000 | 0.2870 | 0.4464 | 0.1233 |
| 0.6737 | 0.95 | 4000 | 0.2590 | 0.4210 | 0.1146 |
### Framework versions
- Transformers 4.27.3
- Pytorch 1.13.1+cu116
- Datasets 2.10.1
- Tokenizers 0.13.2
|