cali-whisper-tiny.en-drop-007-production

This model is a fine-tuned version of 1Developer/cali-whisper-tiny.en-drop-006-production on the audiofolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.0000
• Wer Ortho: 0.0
• Wer: 0.0

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: 16
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant_with_warmup
• lr_scheduler_warmup_steps: 50
• training_steps: 500
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Ortho	Wer
No log	4.0	20	0.0231	0.0	0.0
0.7287	8.0	40	0.0001	0.0	0.0
0.0002	12.0	60	0.0000	0.0	0.0
0.0000	16.0	80	0.0000	0.0	0.0
0.0000	20.0	100	0.0000	0.0	0.0
0.0000	24.0	120	0.0000	0.0	0.0
0.0000	28.0	140	0.0000	0.0	0.0
0.0000	32.0	160	0.0000	0.0	0.0
0.0000	36.0	180	0.0000	0.0	0.0
0.0000	40.0	200	0.0000	0.0	0.0
0.0000	44.0	220	0.0000	0.0	0.0
0.0000	48.0	240	0.0000	0.0	0.0
0.0000	52.0	260	0.0000	0.0	0.0
0.0000	56.0	280	0.0000	0.0	0.0
0.0000	60.0	300	0.0000	0.0	0.0
0.0000	64.0	320	0.0000	0.0	0.0
0.0000	68.0	340	0.0000	0.0	0.0
0.0000	72.0	360	0.0000	0.0	0.0
0.0000	76.0	380	0.0000	0.0	0.0
0.0000	80.0	400	0.0000	0.0	0.0
0.0000	84.0	420	0.0000	0.0	0.0
0.0000	88.0	440	0.0000	0.0	0.0
0.0000	92.0	460	0.0000	0.0	0.0
0.0000	96.0	480	0.0000	0.0	0.0
0.0000	100.0	500	0.0000	0.0	0.0

Framework versions

• Transformers 5.13.1
• Pytorch 2.11.0+cu128
• Datasets 5.0.1
• Tokenizers 0.22.2