<!-- 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. -->
seq-xls-r-fleurs_zu-run3-asr_xh-run5
This model is a fine-tuned version of lucas-meyer/xls-r-fleurs_zu-run3 on asr_xh. It achieves the following results on the evaluation set:
- Val WER: 0.5578
- Test WER: 0.5766
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0003
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 30
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 14.2303 | 0.32 | 50 | 6.1214 | 1.0 |
| 4.1753 | 0.64 | 100 | 3.3769 | 1.0 |
| 3.263 | 0.96 | 150 | 3.0843 | 1.0 |
| 3.0162 | 1.27 | 200 | 2.7783 | 1.0 |
| 1.9813 | 1.59 | 250 | 0.8590 | 0.8218 |
| 0.841 | 1.91 | 300 | 0.5767 | 0.6932 |
| 0.7204 | 2.23 | 350 | 0.5838 | 0.7140 |
| 0.5802 | 2.55 | 400 | 0.5019 | 0.6793 |
| 0.5307 | 2.87 | 450 | 0.4409 | 0.6203 |
| 0.4678 | 3.18 | 500 | 0.4813 | 0.6064 |
| 0.4309 | 3.5 | 550 | 0.5182 | 0.6375 |
| 0.44 | 3.82 | 600 | 0.3939 | 0.5594 |
| 0.3664 | 4.14 | 650 | 0.4210 | 0.5585 |
| 0.3169 | 4.46 | 700 | 0.4745 | 0.5873 |
| 0.3344 | 4.78 | 750 | 0.4249 | 0.5526 |
Framework versions
- Transformers 4.28.0
- Pytorch 2.0.1+cu117
- Datasets 2.14.4
- Tokenizers 0.13.3
