<!-- 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. -->
This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the MOZILLA-FOUNDATION/COMMON_VOICE_8_0 - HI dataset. It achieves the following results on the evaluation set:
- Loss: 0.5258
- Wer: 1.0073
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: 7.5e-05
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 8
- total_train_batch_size: 128
- total_eval_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 2000
- num_epochs: 100.0
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 4.917 | 16.13 | 500 | 4.8963 | 1.0 |
| 3.3585 | 32.25 | 1000 | 3.3069 | 1.0000 |
| 1.5873 | 48.38 | 1500 | 0.8274 | 1.0061 |
| 1.2654 | 64.51 | 2000 | 0.6250 | 1.0076 |
| 1.0917 | 80.64 | 2500 | 0.5460 | 1.0056 |
| 1.0001 | 96.76 | 3000 | 0.5304 | 1.0083 |
Framework versions
- Transformers 4.17.0.dev0
- Pytorch 1.10.2+cu113
- Datasets 1.18.4.dev0
- Tokenizers 0.11.0
