finetune_wav2vec2_960h_six

This model is a fine-tuned version of facebook/wav2vec2-base-960h on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.7147
• Wer: 31.6162
• Cer: 15.8883

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: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 2000
• training_steps: 4500

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
0.9942	18.5185	1000	0.7147	31.6162	15.8883
0.5163	37.0370	2000	0.7961	31.2892	15.4668
0.3093	55.5556	3000	0.9768	30.9894	15.2757
0.2142	74.0741	4000	1.0128	29.0270	14.6912

Framework versions

• Transformers 4.40.2
• Pytorch 1.12.1+cu116
• Datasets 2.19.1
• Tokenizers 0.19.1