kingabzpro's picture
Update README.md
319c507
|
raw
history blame
2.91 kB
metadata
language:
  - ga-IE
license: apache-2.0
tags:
  - automatic-speech-recognition
  - robust-speech-event
datasets:
  - mozilla-foundation/common_voice_8_0
metrics:
  - wer
  - cer
model-index:
  - name: wav2vec2-large-xls-r-1b-Irish-Abid
    results:
      - task:
          type: automatic-speech-recognition
          name: Speech Recognition
        dataset:
          type: mozilla-foundation/common_voice_8_0
          name: Common Voice ga-IE
          args: ga-IE
        metrics:
          - type: wer
            value: 42.36
            name: Test WER
          - type: cer
            value: 17.68
            name: Test CER

wav2vec2-large-xls-r-1b-Irish

This model is a fine-tuned version of facebook/wav2vec2-xls-r-1b on the common_voice dataset. It achieves the following results on the evaluation set:

  • Loss: 1.3599
  • Wer: 0.4236
  • Cer: 0.1768

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: 32
  • eval_batch_size: 8
  • seed: 42
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 128
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 200
  • num_epochs: 100
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer Cer
6.3955 12.48 100 2.9897 1.0 1.0
2.3811 24.97 200 1.2304 0.7140 0.3106
1.0476 37.48 300 1.0661 0.5597 0.2407
0.7014 49.97 400 1.1788 0.4799 0.1947
0.4409 62.48 500 1.2649 0.4658 0.1997
0.4839 74.97 600 1.3259 0.4450 0.1868
0.3643 87.48 700 1.3506 0.4312 0.1760
0.3468 99.97 800 1.3599 0.4236 0.1768

Framework versions

  • Transformers 4.17.0.dev0
  • Pytorch 1.10.2+cu102
  • Datasets 1.18.2.dev0
  • Tokenizers 0.11.0