whisper-base-finetuned-bmd-20241112_114002

This model is a fine-tuned version of openai/whisper-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 2.9882
• Accuracy: 0.2941
• F1: 0.2676
• Precision: 0.5462
• Recall: 0.2941
• Sensitivity: 0.2941

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: 0.0001
• train_batch_size: 16
• eval_batch_size: 16
• seed: 1968
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• 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_ratio: 0.1
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall	Sensitivity
No log	0.8571	3	1.1337	0.2647	0.1108	0.0701	0.2647	0.2647
No log	2.0	7	1.1425	0.2647	0.1108	0.0701	0.2647	0.2647
1.0516	2.8571	10	1.1001	0.5	0.4068	0.3585	0.5	0.5
1.0516	4.0	14	1.1083	0.5294	0.4725	0.5349	0.5294	0.5294
1.0516	4.8571	17	1.4131	0.3235	0.2838	0.5783	0.3235	0.3235
0.6189	6.0	21	1.1835	0.5	0.5036	0.5107	0.5	0.5
0.6189	6.8571	24	1.5920	0.3235	0.3248	0.5662	0.3235	0.3235
0.6189	8.0	28	2.0293	0.3529	0.25	0.2047	0.3529	0.3529
0.1708	8.8571	31	2.1477	0.3529	0.3436	0.5845	0.3529	0.3529
0.1708	10.0	35	2.5696	0.3235	0.2841	0.5607	0.3235	0.3235
0.1708	10.8571	38	2.9175	0.3529	0.2994	0.5809	0.3529	0.3529
0.0173	12.0	42	2.9863	0.2941	0.2676	0.5462	0.2941	0.2941
0.0173	12.8571	45	2.9882	0.2941	0.2676	0.5462	0.2941	0.2941

Framework versions

• Transformers 4.46.2
• Pytorch 2.5.0+cu121
• Datasets 3.1.0
• Tokenizers 0.20.3