whisper-a-clp

This model is a fine-tuned version of openai/whisper-small on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0267
• Wer: 11.7400

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 132
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
3.4117	2.5063	100	0.6501	50.3145
0.1828	5.0	200	0.0349	25.3669
0.0163	7.5063	300	0.0345	19.4969
0.0111	10.0	400	0.0335	25.3669
0.0076	12.5063	500	0.0279	14.2558
0.0074	15.0	600	0.0262	12.3690
0.0064	17.5063	700	0.0260	12.1593
0.0051	20.0	800	0.0262	12.1593
0.0041	22.5063	900	0.0280	9.2243
0.0037	25.0	1000	0.0275	11.5304
0.0029	27.5063	1100	0.0267	11.7400

Framework versions

• Transformers 4.47.0.dev0
• Pytorch 2.4.0
• Datasets 3.0.1
• Tokenizers 0.20.0