whisper-base-mixedv1.5

This model was trained from scratch on the audiofolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.3495
• Cer: 9.6739

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: 1e-05
• 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: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer
0.0005	66.67	1000	0.3284	11.0326
0.0002	133.33	2000	0.3376	9.8370
0.0001	200.0	3000	0.3470	9.6739
0.0001	266.67	4000	0.3495	9.6739

Framework versions

• Transformers 4.36.0.dev0
• Pytorch 2.1.1
• Datasets 2.15.0
• Tokenizers 0.15.0