<!-- 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_nl-run2-asr_af-run5
This model is a fine-tuned version of lucas-meyer/xls-r-fleurs_nl-run2 on asr_af. It achieves the following results:
- Val WER: 0.4140
- Test WER: 0.4251
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 (Train) |
|---|---|---|---|---|
| 11.2739 | 0.29 | 50 | 5.2927 | 1.0 |
| 3.8398 | 0.59 | 100 | 3.3122 | 1.0 |
| 3.1152 | 0.88 | 150 | 3.0188 | 1.0 |
| 2.9857 | 1.17 | 200 | 2.9748 | 0.9998 |
| 2.8793 | 1.47 | 250 | 2.7253 | 1.0 |
| 2.0828 | 1.76 | 300 | 1.5219 | 0.9291 |
| 1.1713 | 2.05 | 350 | 0.8089 | 0.6017 |
| 0.74 | 2.35 | 400 | 0.6447 | 0.5613 |
| 0.6223 | 2.64 | 450 | 0.5806 | 0.4899 |
| 0.5661 | 2.93 | 500 | 0.5890 | 0.4928 |
| 0.4624 | 3.23 | 550 | 0.5796 | 0.4767 |
| 0.4107 | 3.52 | 600 | 0.5077 | 0.4624 |
| 0.3755 | 3.81 | 650 | 0.4489 | 0.4109 |
| 0.3255 | 4.11 | 700 | 0.4474 | 0.3887 |
| 0.2728 | 4.4 | 750 | 0.4477 | 0.3958 |
| 0.2756 | 4.69 | 800 | 0.4477 | 0.3841 |
| 0.282 | 4.99 | 850 | 0.4243 | 0.3914 |
| 0.2362 | 5.28 | 900 | 0.4756 | 0.4081 |
| 0.2262 | 5.57 | 950 | 0.4554 | 0.3824 |
| 0.2315 | 5.87 | 1000 | 0.3963 | 0.3721 |
| 0.2016 | 6.16 | 1050 | 0.4290 | 0.3734 |
| 0.1772 | 6.45 | 1100 | 0.4419 | 0.3649 |
| 0.1835 | 6.74 | 1150 | 0.4339 | 0.3560 |
Framework versions
- Transformers 4.28.0
- Pytorch 2.0.1+cu117
- Datasets 2.14.4
- Tokenizers 0.13.3
