AlbertoFor's picture
update model card README.md
efc31eb
metadata
language:
  - it
license: apache-2.0
tags:
  - automatic-speech-recognition
  - common_voice
  - generated_from_trainer
datasets:
  - common_voice
metrics:
  - wer
model-index:
  - name: wav2vec2-common_voice-it-demo
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: COMMON_VOICE - IT
          type: common_voice
          config: it
          split: test[80%:]
          args: 'Config: it, Training split: train[85%:], Eval split: test[80%:]'
        metrics:
          - name: Wer
            type: wer
            value: 0.23675718221172767

wav2vec2-common_voice-it-demo

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the COMMON_VOICE - IT dataset. It achieves the following results on the evaluation set:

  • Loss: 0.3484
  • Wer: 0.2368

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: 0.0003
  • train_batch_size: 4
  • eval_batch_size: 8
  • seed: 42
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 8
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 500
  • num_epochs: 10.0
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
No log 0.37 400 0.9124 0.7336
3.904 0.74 800 0.4753 0.5022
0.4384 1.1 1200 0.3941 0.3731
0.2985 1.47 1600 0.4007 0.3830
0.2719 1.84 2000 0.3576 0.3597
0.2719 2.21 2400 0.3571 0.3286
0.2158 2.57 2800 0.3465 0.3198
0.2054 2.94 3200 0.3162 0.2982
0.1783 3.31 3600 0.3295 0.3089
0.1495 3.68 4000 0.3248 0.3034
0.1495 4.04 4400 0.3101 0.3028
0.1397 4.41 4800 0.3588 0.3006
0.123 4.78 5200 0.3451 0.3041
0.115 5.15 5600 0.3333 0.2921
0.0947 5.51 6000 0.3331 0.2858
0.0947 5.88 6400 0.3536 0.2950
0.0952 6.25 6800 0.3344 0.2786
0.0778 6.62 7200 0.3363 0.2699
0.0744 6.99 7600 0.3246 0.2655
0.0648 7.35 8000 0.3390 0.2627
0.0648 7.72 8400 0.3405 0.2630
0.0591 8.09 8800 0.3367 0.2534
0.0527 8.46 9200 0.3448 0.2509
0.0461 8.82 9600 0.3379 0.2425
0.0408 9.19 10000 0.3491 0.2409
0.0408 9.56 10400 0.3456 0.2377
0.0393 9.93 10800 0.3488 0.2370

Framework versions

  • Transformers 4.29.2
  • Pytorch 2.0.1+cu117
  • Datasets 2.12.0
  • Tokenizers 0.11.0