Edit model card

my_awesome_asr_mind_model

This model is a fine-tuned version of facebook/wav2vec2-base on the minds14 dataset. It achieves the following results on the evaluation set:

  • Loss: 3.0295
  • Wer: 1.0

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: 1e-05
  • train_batch_size: 2
  • eval_batch_size: 8
  • seed: 42
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 4
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 500
  • training_steps: 2000

Training results

Training Loss Epoch Step Validation Loss Wer
36.0013 0.71 50 54.8783 1.0231
29.2024 1.43 100 43.3910 1.0
18.6027 2.14 150 24.3702 1.0
6.5918 2.86 200 7.0226 1.0
4.0512 3.57 250 4.3558 1.0
3.6216 4.29 300 4.0701 1.0
3.4344 5.0 350 3.8139 1.0
3.5985 5.71 400 3.6550 1.0
3.5004 6.43 450 3.5076 1.0
3.3916 7.14 500 3.4591 1.0
3.1966 7.86 550 3.3332 1.0
3.2384 8.57 600 3.2828 1.0
3.1981 9.29 650 3.2563 1.0
3.1743 10.0 700 3.2011 1.0
3.1251 10.71 750 3.1600 1.0
3.0371 11.43 800 3.1436 1.0
3.0702 12.14 850 3.1633 1.0
3.0748 12.86 900 3.1194 1.0
3.0459 13.57 950 3.1797 1.0
3.0496 14.29 1000 3.1073 1.0
3.0744 15.0 1050 3.1033 1.0
3.0342 15.71 1100 3.0702 1.0
3.0469 16.43 1150 3.0680 1.0
3.0234 17.14 1200 3.0650 1.0
3.0739 17.86 1250 3.0586 1.0
2.9964 18.57 1300 3.0542 1.0
3.0906 19.29 1350 3.0519 1.0
2.9823 20.0 1400 3.0456 1.0
3.038 20.71 1450 3.0399 1.0
2.9952 21.43 1500 3.0357 1.0
3.0092 22.14 1550 3.0571 1.0
2.9838 22.86 1600 3.0354 1.0
3.0611 23.57 1650 3.0435 1.0
2.9924 24.29 1700 3.0368 1.0
2.9854 25.0 1750 3.0580 1.0
3.0193 25.71 1800 3.0347 1.0
2.9694 26.43 1850 3.0335 1.0
3.0039 27.14 1900 3.0318 1.0
2.9789 27.86 1950 3.0322 1.0
2.9828 28.57 2000 3.0295 1.0

Framework versions

  • Transformers 4.30.0
  • Pytorch 2.2.1+cu121
  • Datasets 2.18.0
  • Tokenizers 0.13.3
Downloads last month
0
Inference API
or
This model can be loaded on Inference API (serverless).

Evaluation results