gender_classification_MHDP_asr_dataset_V1

This model is a fine-tuned version of JaesungHuh/voice-gender-classifier on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.1446
• Accuracy: 0.9569
• F1: 0.9606
• Roc Auc: 0.9920
• Confusion Matrix: [[110, 11], [0, 134]]

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: 4
• eval_batch_size: 4
• 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: cosine
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Roc Auc	Confusion Matrix
0.5215	1.0	175	0.5170	0.8070	0.78	0.9839	[[53, 0], [22, 39]]
0.3483	2.0	350	0.4779	0.8158	0.7921	0.9542	[[53, 0], [21, 40]]
0.477	3.0	525	0.2958	0.9035	0.9009	1.0	[[53, 0], [11, 50]]
0.327	4.0	700	0.2351	0.9386	0.9391	1.0	[[53, 0], [7, 54]]
0.2346	5.0	875	0.2508	0.8860	0.8807	1.0	[[53, 0], [13, 48]]
0.2441	6.0	1050	0.2247	0.8860	0.8807	1.0	[[53, 0], [13, 48]]
0.2341	7.0	1225	0.1040	0.9825	0.9833	1.0	[[53, 0], [2, 59]]
0.2745	8.0	1400	0.1556	0.9474	0.9483	1.0	[[53, 0], [6, 55]]
0.3679	9.0	1575	0.1100	0.9737	0.9748	1.0	[[53, 0], [3, 58]]
0.2632	10.0	1750	0.1346	0.9561	0.9573	1.0	[[53, 0], [5, 56]]
0.2121	11.0	1925	0.0841	0.9912	0.9917	1.0	[[53, 0], [1, 60]]
0.1473	12.0	2100	0.1087	0.9737	0.9748	1.0	[[53, 0], [3, 58]]
0.1911	13.0	2275	0.1195	0.9649	0.9661	1.0	[[53, 0], [4, 57]]
0.0672	14.0	2450	0.1396	0.9474	0.9483	1.0	[[53, 0], [6, 55]]
0.263	15.0	2625	0.0510	1.0	1.0	1.0	[[53, 0], [0, 61]]
0.1076	16.0	2800	0.0940	0.9649	0.9661	1.0	[[53, 0], [4, 57]]
0.069	17.0	2975	0.1289	0.9561	0.9573	1.0	[[53, 0], [5, 56]]
0.0598	18.0	3150	0.1003	0.9649	0.9661	1.0	[[53, 0], [4, 57]]
0.2148	19.0	3325	0.1039	0.9649	0.9661	1.0	[[53, 0], [4, 57]]
0.2338	20.0	3500	0.1783	0.9035	0.9009	1.0	[[53, 0], [11, 50]]

Framework versions

• Transformers 4.56.0
• Pytorch 2.8.0+cu129
• Datasets 4.8.5
• Tokenizers 0.22.0