facebook_wav2vec2-xls-r-300m_meet_tr_p_10_30h
This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.5611
- Cer: 0.1334
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: 0.0001
- train_batch_size: 7
- eval_batch_size: 56
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 1000
- num_epochs: 3
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Cer |
---|---|---|---|---|
6.2651 | 0.21 | 500 | 5.9159 | 1.0 |
2.8965 | 0.43 | 1000 | 2.0493 | 0.4184 |
1.3122 | 0.64 | 1500 | 1.0627 | 0.2433 |
1.0682 | 0.85 | 2000 | 0.8674 | 0.2083 |
0.9281 | 1.07 | 2500 | 0.7413 | 0.1749 |
0.8103 | 1.28 | 3000 | 0.7032 | 0.1652 |
0.7823 | 1.5 | 3500 | 0.6806 | 0.1616 |
0.7429 | 1.71 | 4000 | 0.6430 | 0.1547 |
0.7375 | 1.92 | 4500 | 0.6253 | 0.1533 |
0.6299 | 2.14 | 5000 | 0.6375 | 0.1423 |
0.5801 | 2.35 | 5500 | 0.6086 | 0.1398 |
0.5735 | 2.56 | 6000 | 0.5808 | 0.1394 |
0.5448 | 2.78 | 6500 | 0.5736 | 0.1351 |
0.5555 | 2.99 | 7000 | 0.5611 | 0.1334 |
Framework versions
- Transformers 4.17.0
- Pytorch 1.10.0+cu113
- Datasets 1.18.3
- Tokenizers 0.15.2
- Downloads last month
- 1