speaker-segmentation-fine-tuned-callhome-eng-0927

This model is a fine-tuned version of pyannote/segmentation-3.0 on the diarizers-community/callhome eng dataset. It achieves the following results on the evaluation set:

• Loss: 0.4972
• Model Preparation Time: 0.007
• Der: 0.1808
• False Alarm: 0.0619
• Missed Detection: 0.0699
• Confusion: 0.0489

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: 0.001
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• num_epochs: 20.0

Training results

Training Loss	Epoch	Step	Validation Loss	Model Preparation Time	Der	False Alarm	Missed Detection	Confusion
0.4219	1.0	362	0.4806	0.007	0.1893	0.0543	0.0778	0.0572
0.3851	2.0	724	0.4724	0.007	0.1879	0.0480	0.0817	0.0581
0.3676	3.0	1086	0.4666	0.007	0.1849	0.0588	0.0712	0.0549
0.3554	4.0	1448	0.4667	0.007	0.1831	0.0598	0.0704	0.0530
0.3457	5.0	1810	0.4613	0.007	0.1802	0.0603	0.0692	0.0507
0.3319	6.0	2172	0.4586	0.007	0.1771	0.0541	0.0740	0.0490
0.3126	7.0	2534	0.4743	0.007	0.1787	0.0488	0.0787	0.0512
0.3144	8.0	2896	0.4885	0.007	0.1852	0.0595	0.0736	0.0520
0.3084	9.0	3258	0.4781	0.007	0.1819	0.0624	0.0698	0.0498
0.303	10.0	3620	0.4801	0.007	0.1819	0.0599	0.0727	0.0493
0.2907	11.0	3982	0.4893	0.007	0.1820	0.0581	0.0750	0.0490
0.293	12.0	4344	0.4798	0.007	0.1792	0.0559	0.0739	0.0493
0.2787	13.0	4706	0.4926	0.007	0.1823	0.0642	0.0675	0.0506
0.2741	14.0	5068	0.4928	0.007	0.1813	0.0647	0.0676	0.0490
0.2705	15.0	5430	0.4940	0.007	0.1807	0.0595	0.0717	0.0496
0.2638	16.0	5792	0.4955	0.007	0.1809	0.0632	0.0691	0.0486
0.2623	17.0	6154	0.4999	0.007	0.1808	0.0606	0.0707	0.0494
0.2704	18.0	6516	0.4977	0.007	0.1803	0.0605	0.0707	0.0491
0.2705	19.0	6878	0.4973	0.007	0.1808	0.0616	0.0701	0.0491
0.2683	20.0	7240	0.4972	0.007	0.1808	0.0619	0.0699	0.0489

Framework versions

• Transformers 4.56.2
• Pytorch 2.7.1+cu118
• Datasets 3.6.0
• Tokenizers 0.22.1