huyue012's picture
update model card README.md
4211786
|
raw
history blame
1.97 kB
metadata
license: apache-2.0
tags:
  - generated_from_trainer
model-index:
  - name: wav2vec2-base-cynthia-tedlium-2500-v2
    results: []

wav2vec2-base-cynthia-tedlium-2500-v2

This model is a fine-tuned version of facebook/wav2vec2-base-960h on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.6376
  • Wer: 0.2137

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: 0.0005
  • train_batch_size: 16
  • eval_batch_size: 8
  • seed: 42
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 32
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 500
  • num_epochs: 50
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
0.7233 5.26 400 5.6115 0.4909
0.7224 10.52 800 1.1835 0.3629
0.5193 15.78 1200 0.5786 0.2994
0.4114 21.05 1600 0.6535 0.2823
0.3279 26.31 2000 0.6783 0.2709
0.2711 31.58 2400 0.6570 0.2490
0.2776 36.84 2800 0.6724 0.2358
0.1805 42.1 3200 0.6212 0.2241
0.1535 47.37 3600 0.6376 0.2137

Framework versions

  • Transformers 4.11.3
  • Pytorch 1.10.0+cu113
  • Datasets 1.13.3
  • Tokenizers 0.10.3