wav2vec2-base-timit-demo-colab

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: 0.4707
• Wer: 0.3411

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: 32
• 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
3.4575	4.0	500	1.3367	0.9724
0.594	8.0	1000	0.4365	0.4390
0.2195	12.0	1500	0.4438	0.3955
0.1246	16.0	2000	0.4741	0.3707
0.082	20.0	2500	0.4766	0.3564
0.0605	24.0	3000	0.4657	0.3475
0.0458	28.0	3500	0.4707	0.3411

Framework versions

• Transformers 4.11.3
• Pytorch 1.10.0+cu111
• Datasets 1.18.3
• Tokenizers 0.10.3