dts_ESLO.06.05.25_exp.ft.dia.1.A_mdl.no1

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

• Loss: 0.7479
• Model Preparation Time: 0.0044
• Der: 0.4638
• False Alarm: 0.1470
• Missed Detection: 0.2281
• Confusion: 0.0888

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Model Preparation Time	Der	False Alarm	Missed Detection	Confusion
0.8676	1.0	270	0.7954	0.0044	0.5104	0.1441	0.2546	0.1117
0.8147	2.0	540	0.7855	0.0044	0.5000	0.1319	0.2714	0.0967
0.8085	3.0	810	0.7573	0.0044	0.4754	0.1503	0.2216	0.1036
0.7639	4.0	1080	0.7556	0.0044	0.4719	0.1445	0.2282	0.0993
0.7599	5.0	1350	0.7472	0.0044	0.4685	0.1595	0.2048	0.1041
0.7611	6.0	1620	0.7520	0.0044	0.4670	0.1463	0.2260	0.0947
0.7299	7.0	1890	0.7917	0.0044	0.4768	0.1359	0.2535	0.0874
0.7409	8.0	2160	0.7331	0.0044	0.4661	0.1432	0.2254	0.0974
0.7153	9.0	2430	0.7479	0.0044	0.4662	0.1583	0.2068	0.1010
0.7247	10.0	2700	0.7446	0.0044	0.4619	0.1527	0.2192	0.0899
0.7187	11.0	2970	0.7425	0.0044	0.4637	0.1427	0.2328	0.0882
0.7259	12.0	3240	0.7391	0.0044	0.4601	0.1453	0.2278	0.0870
0.6628	13.0	3510	0.7496	0.0044	0.4610	0.1486	0.2250	0.0874
0.6894	14.0	3780	0.7429	0.0044	0.4598	0.1428	0.2312	0.0858
0.681	15.0	4050	0.7454	0.0044	0.4603	0.1502	0.2193	0.0907
0.6821	16.0	4320	0.7409	0.0044	0.4621	0.1496	0.2213	0.0911
0.6868	17.0	4590	0.7472	0.0044	0.4647	0.1478	0.2263	0.0907
0.6821	18.0	4860	0.7472	0.0044	0.4628	0.1473	0.2268	0.0887
0.6987	19.0	5130	0.7477	0.0044	0.4637	0.1466	0.2287	0.0884
0.7158	20.0	5400	0.7479	0.0044	0.4638	0.1470	0.2281	0.0888

Framework versions

• Transformers 4.45.2
• Pytorch 2.4.1+cu121
• Datasets 3.0.1
• Tokenizers 0.20.0