wav2vec2-large-xls-r-300m-assamese_speech_to_IPA_new2_test
This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the common_voice_11_0 dataset. It achieves the following results on the evaluation set:
- Loss: 0.8522
- Wer: 0.9675
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: 0.0003
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 50
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
4.5759 | 9.8765 | 400 | 0.8360 | 0.9968 |
0.2808 | 19.7531 | 800 | 0.8647 | 0.9838 |
0.1122 | 29.6296 | 1200 | 0.8335 | 0.9740 |
0.0706 | 39.5062 | 1600 | 0.8604 | 0.9708 |
0.0505 | 49.3827 | 2000 | 0.8522 | 0.9675 |
Framework versions
- Transformers 4.41.1
- Pytorch 2.3.0+cu121
- Datasets 2.19.2
- Tokenizers 0.19.1
- Downloads last month
- 33