hausa-asr / README.md
Cdial's picture
Upload README.md
2bfb84f
|
raw
history blame
3.4 kB
metadata
language:
  - ha
license: apache-2.0
tags:
  - automatic-speech-recognition
  - mozilla-foundation/common_voice_8_0
  - generated_from_trainer
  - ha
  - robust-speech-event
  - model_for_talk
datasets:
  - mozilla-foundation/common_voice_8_0
model-index:
  - name: Akashpb13/Hausa_xlsr
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: Common Voice 8
          type: mozilla-foundation/common_voice_8_0
          args: ha
        metrics:
          - name: Test WER
            type: wer
            value: 0.20614541257934219
          - name: Test CER
            type: cer
            value: 0.04358048053214061
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: Robust Speech Event - Dev Data
          type: speech-recognition-community-v2/dev_data
          args: ha
        metrics:
          - name: Test WER
            type: wer
            value: 0.20614541257934219
          - name: Test CER
            type: cer
            value: 0.04358048053214061

Akashpb13/Hausa_xlsr

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m It achieves the following results on the evaluation set (which is 10 percent of train data set merged with invalidated data, reported, other, and dev datasets):

  • Loss: 0.275118
  • Wer: 0.329955

Model description

"facebook/wav2vec2-xls-r-300m" was finetuned.

Intended uses & limitations

More information needed

Training and evaluation data

Training data - Common voice Hausa train.tsv, dev.tsv, invalidated.tsv, reported.tsv and other.tsv Only those points were considered where upvotes were greater than downvotes and duplicates were removed after concatenation of all the datasets given in common voice 7.0

Training procedure

For creating the training dataset, all possible datasets were appended and 90-10 split was used.

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 0.000096
  • train_batch_size: 16
  • eval_batch_size: 16
  • seed: 13
  • gradient_accumulation_steps: 2
  • lr_scheduler_type: cosine_with_restarts
  • lr_scheduler_warmup_steps: 500
  • num_epochs: 50
  • mixed_precision_training: Native AMP

Training results

Step Training Loss Validation Loss Wer
500 5.175900 2.750914 1.000000
1000 1.028700 0.338649 0.497999
1500 0.332200 0.246896 0.402241
2000 0.227300 0.239640 0.395839
2500 0.175000 0.239577 0.373966
3000 0.140400 0.243272 0.356095
3500 0.119200 0.263761 0.365164
4000 0.099300 0.265954 0.353428
4500 0.084400 0.276367 0.349693
5000 0.073700 0.282631 0.343825
5500 0.068000 0.282344 0.341158
6000 0.064500 0.281591 0.342491

Framework versions

  • Transformers 4.16.0.dev0
  • Pytorch 1.10.0+cu102
  • Datasets 1.18.3
  • Tokenizers 0.10.3

Evaluation Commands

  1. To evaluate on mozilla-foundation/common_voice_8_0 with split test
python eval.py --model_id Akashpb13/Hausa_xlsr --dataset mozilla-foundation/common_voice_8_0 --config ha --split test