DrishtiSharma's picture
update model card README.md
ebba215
|
raw
history blame
2.31 kB
metadata
license: apache-2.0
tags:
  - generated_from_trainer
datasets:
  - common_voice
model-index:
  - name: ''
    results: []

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the common_voice dataset. It achieves the following results on the evaluation set:

  • Loss: 0.2852
  • Wer: 0.2396

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: 7e-05
  • train_batch_size: 32
  • eval_batch_size: 32
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 1000
  • num_epochs: 100.0
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
6.9294 6.1 500 2.9712 1.0
2.8305 12.2 1000 1.7073 0.9479
1.4795 18.29 1500 0.5756 0.6397
1.3433 24.39 2000 0.4968 0.5424
1.1766 30.49 2500 0.4185 0.4743
1.0017 36.59 3000 0.3303 0.3578
0.9358 42.68 3500 0.3003 0.3051
0.8358 48.78 4000 0.3045 0.2884
0.7647 54.88 4500 0.2866 0.2677
0.7482 60.98 5000 0.2829 0.2585
0.6943 67.07 5500 0.2782 0.2478
0.6586 73.17 6000 0.2911 0.2537
0.6425 79.27 6500 0.2817 0.2462
0.6067 85.37 7000 0.2910 0.2436
0.5974 91.46 7500 0.2875 0.2430
0.5812 97.56 8000 0.2852 0.2396

Framework versions

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