metadata
license: cc-by-nc-4.0
base_model: facebook/mms-1b-all
tags:
- generated_from_trainer
datasets:
- common_voice_11_0
metrics:
- wer
model-index:
- name: wav2vec2-kyrgyz-colab
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: common_voice_11_0
type: common_voice_11_0
config: ky
split: test
args: ky
metrics:
- name: Wer
type: wer
value: 0.26476260446321975
wav2vec2-kyrgyz-colab
This model is a fine-tuned version of facebook/mms-1b-all on the common_voice_11_0 dataset. It achieves the following results on the evaluation set:
- Loss: 0.2952
- Wer: 0.2648
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: 1e-05
- train_batch_size: 2
- 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: 200
- num_epochs: 5
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 9.9224 | 0.29 | 500 | 3.9047 | 1.0023 |
| 3.3028 | 0.59 | 1000 | 2.7653 | 0.9997 |
| 2.4414 | 0.88 | 1500 | 1.1551 | 0.6330 |
| 1.5571 | 1.18 | 2000 | 0.5679 | 0.3712 |
| 1.2479 | 1.47 | 2500 | 0.4202 | 0.3291 |
| 1.0857 | 1.76 | 3000 | 0.3655 | 0.3105 |
| 1.0304 | 2.06 | 3500 | 0.3392 | 0.3007 |
| 0.9989 | 2.35 | 4000 | 0.3245 | 0.2909 |
| 0.9711 | 2.65 | 4500 | 0.3157 | 0.2825 |
| 0.9371 | 2.94 | 5000 | 0.3093 | 0.2764 |
| 0.9423 | 3.24 | 5500 | 0.3047 | 0.2732 |
| 0.9226 | 3.53 | 6000 | 0.3017 | 0.2715 |
| 0.9365 | 3.82 | 6500 | 0.2990 | 0.2689 |
| 0.8969 | 4.12 | 7000 | 0.2971 | 0.2668 |
| 0.9101 | 4.41 | 7500 | 0.2961 | 0.2644 |
| 0.9044 | 4.71 | 8000 | 0.2954 | 0.2642 |
| 0.902 | 5.0 | 8500 | 0.2952 | 0.2648 |
Framework versions
- Transformers 4.37.0.dev0
- Pytorch 2.1.2+cu121
- Datasets 2.16.1
- Tokenizers 0.15.0