wav2vec2-base-timit-demo-colab10

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: 0.4460
• Wer: 0.3425

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: 16
• eval_batch_size: 8
• 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
4.9891	3.52	500	3.1554	1.0
1.71	7.04	1000	0.7122	0.5811
0.6164	10.56	1500	0.5149	0.4880
0.4188	14.08	2000	0.4726	0.4344
0.3038	17.61	2500	0.4765	0.4092
0.2312	21.13	3000	0.4387	0.3765
0.1867	24.65	3500	0.4411	0.3583
0.1582	28.17	4000	0.4460	0.3425

Framework versions

• Transformers 4.11.3
• Pytorch 1.11.0+cu113
• Datasets 1.18.3
• Tokenizers 0.10.3