wav2vec2-large-teacher-base-student-en-asr-timit

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

• Loss: 73.5882
• Wer: 0.3422

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: 0.0001
• train_batch_size: 64
• eval_batch_size: 64
• 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: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
920.6083	3.17	200	1256.0675	1.0
660.5993	6.35	400	717.6098	0.9238
336.5288	9.52	600	202.0025	0.5306
131.3178	12.7	800	108.0701	0.4335
73.4232	15.87	1000	90.2797	0.3728
54.9439	19.05	1200	76.9043	0.3636
44.6595	22.22	1400	79.2443	0.3550
38.6381	25.4	1600	73.6277	0.3493
35.074	28.57	1800	73.5882	0.3422

Framework versions

• Transformers 4.25.1
• Pytorch 1.13.0+cu116
• Datasets 1.18.3
• Tokenizers 0.13.2