w2v-bert-2.0-ur / README.md
mwz's picture
Upload tokenizer
ef600d2 verified
|
raw
history blame
1.98 kB
metadata
tags:
  - generated_from_trainer
datasets:
  - common_voice_16_0
metrics:
  - wer
base_model: ylacombe/w2v-bert-2.0
model-index:
  - name: w2v-bert-2.0-ur
    results:
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          name: common_voice_16_0
          type: common_voice_16_0
          config: ur
          split: test
          args: ur
        metrics:
          - type: wer
            value: 0.2984838198687486
            name: Wer

w2v-bert-2.0-ur

This model is a fine-tuned version of ylacombe/w2v-bert-2.0 on the common_voice_16_0 dataset. It achieves the following results on the evaluation set:

  • Loss: inf
  • Wer: 0.2985

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: 5e-05
  • train_batch_size: 8
  • eval_batch_size: 8
  • seed: 42
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 16
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 500
  • num_epochs: 10
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
0.2789 2.4 300 inf 0.3200
0.2724 4.8 600 inf 0.3320
0.1912 7.2 900 inf 0.2935
0.0931 9.6 1200 inf 0.2985

Framework versions

  • Transformers 4.37.0.dev0
  • Pytorch 2.1.0+cu121
  • Datasets 2.16.1
  • Tokenizers 0.15.1