w2v2-libri-10min

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: 2.0727
• Wer: 0.5920

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: 0.0003
• 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: 500
• training_steps: 2500

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
4.8063	62.5	250	2.9177	1.0
0.8552	125.0	500	1.6335	0.6445
0.0664	187.5	750	1.8028	0.6362
0.0317	250.0	1000	1.7384	0.6335
0.0228	312.5	1250	1.9538	0.6183
0.0143	375.0	1500	1.9389	0.6210
0.0095	437.5	1750	2.1478	0.6238
0.0056	500.0	2000	2.0574	0.5989
0.004	562.5	2250	2.1041	0.6044
0.0026	625.0	2500	2.0727	0.5920

Framework versions

• Transformers 4.30.1
• Pytorch 2.0.1+cu118
• Datasets 1.18.3
• Tokenizers 0.13.3