metadata
license: cc-by-nc-4.0
tags:
- generated_from_trainer
datasets:
- common_voice_6_1
metrics:
- wer
model-index:
- name: wav2vec2-large-mms-1b-odia
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: common_voice_6_1
type: common_voice_6_1
config: or
split: test
args: or
metrics:
- name: Wer
type: wer
value: 1.0526315789473684
wav2vec2-large-mms-1b-odia
This model is a fine-tuned version of facebook/mms-1b-all on the common_voice_6_1 dataset. It achieves the following results on the evaluation set:
- Loss: 0.2591
- Wer: 1.0526
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: 0.001
- train_batch_size: 12
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 100
- num_epochs: 4
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 20.1037 | 0.23 | 10 | 20.9125 | 1.0 |
| 17.8006 | 0.45 | 20 | 15.9823 | 1.0 |
| 12.0829 | 0.68 | 30 | 9.3068 | 1.0 |
| 5.2122 | 0.91 | 40 | 3.6577 | 1.0012 |
| 2.8945 | 1.14 | 50 | 1.9252 | 1.2448 |
| 1.2442 | 1.36 | 60 | 0.7219 | 1.0220 |
| 0.5149 | 1.59 | 70 | 0.3858 | 1.0122 |
| 0.3685 | 1.82 | 80 | 0.3202 | 1.0147 |
| 0.3529 | 2.05 | 90 | 0.3093 | 1.0147 |
| 0.2863 | 2.27 | 100 | 0.3130 | 1.0135 |
| 0.2643 | 2.5 | 110 | 0.3145 | 1.0098 |
| 0.2518 | 2.73 | 120 | 0.2861 | 1.0588 |
| 0.2783 | 2.95 | 130 | 0.2668 | 1.0649 |
| 0.2586 | 3.18 | 140 | 0.2714 | 1.0355 |
| 0.243 | 3.41 | 150 | 0.2631 | 1.0453 |
| 0.2261 | 3.64 | 160 | 0.2642 | 1.0367 |
| 0.2365 | 3.86 | 170 | 0.2591 | 1.0526 |
Framework versions
- Transformers 4.31.0.dev0
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3