AshtonLKY/Whisper_ASR_ATC_v2

Whisper_ASR_ATC

This model is a fine-tuned version of openai/whisper-small on the AshtonLKY/augmented_audio dataset. It achieves the following results on the evaluation set:

• Loss: 0.0171
• Wer: 10.2591

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: 10000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.2282	0.3	1000	0.2253	49.5224
0.1461	0.6	2000	0.1456	42.3271
0.1052	0.89	3000	0.1061	10.8325
0.0698	1.19	4000	0.0708	13.8258
0.043	1.49	5000	0.0537	11.0072
0.0407	1.79	6000	0.0383	10.9401
0.019	2.08	7000	0.0349	15.2078
0.0323	2.38	8000	0.0268	11.4068
0.0164	2.68	9000	0.0236	12.3902
0.0153	2.98	10000	0.0171	10.2591

Framework versions

• Transformers 4.36.2
• Pytorch 2.1.0+cu121
• Datasets 2.16.1
• Tokenizers 0.15.0