whisper-base-finetuned-bmd-V1-fp16-20241111_180200-LOSO-section-out1

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: 3.6195
• Accuracy: 0.3793
• F1: 0.3333

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: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
No log	0.8571	3	1.1100	0.2414	0.1806
No log	2.0	7	1.2438	0.1724	0.1061
1.0564	2.8571	10	1.2948	0.2069	0.1287
1.0564	4.0	14	1.4553	0.3103	0.2008
1.0564	4.8571	17	1.6435	0.2759	0.1798
0.601	6.0	21	1.4658	0.3793	0.3412
0.601	6.8571	24	1.7203	0.3103	0.2597
0.601	8.0	28	1.8225	0.4138	0.4122
0.227	8.8571	31	1.7533	0.4483	0.4552
0.227	10.0	35	2.6984	0.3448	0.2730
0.227	10.8571	38	3.1588	0.2759	0.1850
0.0553	12.0	42	2.9259	0.4138	0.3870
0.0553	12.8571	45	2.9862	0.4483	0.4370
0.0553	14.0	49	3.2382	0.4483	0.4370
0.0072	14.8571	52	3.4094	0.4138	0.3870
0.0072	16.0	56	3.5602	0.4138	0.3870
0.0072	16.8571	59	3.6131	0.3793	0.3333
0.0034	17.1429	60	3.6195	0.3793	0.3333

Framework versions

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