hanasim's picture
Model save
3805c5d verified
|
raw
history blame
2.28 kB
metadata
license: cc-by-nc-4.0
base_model: facebook/mms-1b-all
tags:
  - generated_from_trainer
datasets:
  - common_voice_16_0
metrics:
  - wer
model-index:
  - name: breeze-listen-w2v2-ml
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: common_voice_16_0
          type: common_voice_16_0
          config: ml
          split: test
          args: ml
        metrics:
          - name: Wer
            type: wer
            value: 0.5345542501727713

breeze-listen-w2v2-ml

This model is a fine-tuned version of facebook/mms-1b-all on the common_voice_16_0 dataset. It achieves the following results on the evaluation set:

  • Loss: 0.2698
  • Wer: 0.5346

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: 4
  • eval_batch_size: 8
  • seed: 42
  • distributed_type: multi-GPU
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 100
  • num_epochs: 4.0
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
No log 0.41 200 5.4728 1.0757
No log 0.81 400 5.1274 1.0038
6.5037 1.22 600 0.6167 0.8131
6.5037 1.63 800 0.3284 0.5829
1.0482 2.03 1000 0.3169 0.5667
1.0482 2.44 1200 0.2876 0.5425
1.0482 2.85 1400 0.2847 0.5522
0.4314 3.25 1600 0.2746 0.5394
0.4314 3.66 1800 0.2698 0.5346

Framework versions

  • Transformers 4.38.0.dev0
  • Pytorch 2.1.2+cu121
  • Datasets 2.16.1
  • Tokenizers 0.15.1