amity-diarization-v02

This model is a fine-tuned version of pyannote/segmentation-3.0 on the amitysolution/sample-voice-dataset dataset. It achieves the following results on the evaluation set:

• Loss: 0.3846
• Model Preparation Time: 0.0077
• Der: 0.1686
• False Alarm: 0.0769
• Missed Detection: 0.0777
• Confusion: 0.0140

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: 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: 15.0

Training results

Training Loss	Epoch	Step	Validation Loss	Model Preparation Time	Der	False Alarm	Missed Detection	Confusion
0.7157	0.6173	300	0.5912	0.0077	0.2733	0.0783	0.1634	0.0315
0.5807	1.2346	600	0.5261	0.0077	0.2517	0.0783	0.1451	0.0283
0.5114	1.8519	900	0.4810	0.0077	0.2299	0.0841	0.1207	0.0250
0.4601	2.4691	1200	0.4629	0.0077	0.2098	0.0910	0.0960	0.0227
0.426	3.0864	1500	0.4443	0.0077	0.2016	0.0909	0.0894	0.0214
0.4077	3.7037	1800	0.4391	0.0077	0.1946	0.0866	0.0888	0.0192
0.3818	4.3210	2100	0.4287	0.0077	0.1891	0.0863	0.0839	0.0189
0.3687	4.9383	2400	0.4214	0.0077	0.1848	0.0838	0.0821	0.0188
0.357	5.5556	2700	0.4135	0.0077	0.1802	0.0849	0.0777	0.0175
0.3533	6.1728	3000	0.4106	0.0077	0.1768	0.0796	0.0809	0.0163
0.3357	6.7901	3300	0.3981	0.0077	0.1732	0.0821	0.0754	0.0157
0.3317	7.4074	3600	0.3957	0.0077	0.1724	0.0800	0.0777	0.0146
0.3278	8.0247	3900	0.3884	0.0077	0.1710	0.0800	0.0761	0.0148
0.3193	8.6420	4200	0.3859	0.0077	0.1696	0.0787	0.0765	0.0144
0.3218	9.2593	4500	0.3842	0.0077	0.1687	0.0790	0.0755	0.0142
0.3244	9.8765	4800	0.3795	0.0077	0.1674	0.0781	0.0751	0.0142
0.3121	10.4938	5100	0.3827	0.0077	0.1685	0.0762	0.0780	0.0144
0.31	11.1111	5400	0.3825	0.0077	0.1688	0.0768	0.0779	0.0140
0.3131	11.7284	5700	0.3855	0.0077	0.1688	0.0772	0.0775	0.0141
0.3108	12.3457	6000	0.3836	0.0077	0.1685	0.0772	0.0773	0.0141
0.3093	12.9630	6300	0.3853	0.0077	0.1687	0.0769	0.0779	0.0139
0.3131	13.5802	6600	0.3855	0.0077	0.1688	0.0767	0.0782	0.0139
0.3012	14.1975	6900	0.3847	0.0077	0.1687	0.0769	0.0777	0.0140
0.3108	14.8148	7200	0.3846	0.0077	0.1686	0.0769	0.0777	0.0140

Framework versions

• Transformers 4.51.2
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1