uncleMehrzad/speaker-segmentation-fine-tuned-cv11-3000

speaker-segmentation-fine-tuned-cv11-fa

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

• Loss: 0.5676
• Model Preparation Time: 0.0027
• Der: 0.1956
• False Alarm: 0.0566
• Missed Detection: 0.0374
• Confusion: 0.1016

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: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Model Preparation Time	Der	False Alarm	Missed Detection	Confusion
0.6772	1.0	602	0.6496	0.0027	0.2250	0.0574	0.0494	0.1182
0.6198	2.0	1204	0.6310	0.0027	0.2182	0.0580	0.0439	0.1164
0.5988	3.0	1806	0.6111	0.0027	0.2109	0.0587	0.0411	0.1111
0.5849	4.0	2408	0.6054	0.0027	0.2093	0.0589	0.0393	0.1111
0.595	5.0	3010	0.5949	0.0027	0.2074	0.0572	0.0404	0.1097
0.5711	6.0	3612	0.5875	0.0027	0.2047	0.0599	0.0371	0.1076
0.5655	7.0	4214	0.5831	0.0027	0.2044	0.0571	0.0393	0.1080
0.5577	8.0	4816	0.5824	0.0027	0.2011	0.0571	0.0385	0.1055
0.5632	9.0	5418	0.5749	0.0027	0.1994	0.0566	0.0389	0.1040
0.5566	10.0	6020	0.5741	0.0027	0.1975	0.0571	0.0381	0.1023
0.5499	11.0	6622	0.5732	0.0027	0.1970	0.0574	0.0375	0.1021
0.5487	12.0	7224	0.5692	0.0027	0.1959	0.0570	0.0373	0.1016
0.5461	13.0	7826	0.5684	0.0027	0.1961	0.0571	0.0372	0.1019
0.5475	14.0	8428	0.5674	0.0027	0.1947	0.0566	0.0375	0.1006
0.536	15.0	9030	0.5682	0.0027	0.1956	0.0567	0.0374	0.1015
0.5397	16.0	9632	0.5680	0.0027	0.1959	0.0565	0.0377	0.1017
0.5443	17.0	10234	0.5673	0.0027	0.1957	0.0565	0.0375	0.1017
0.5423	18.0	10836	0.5676	0.0027	0.1957	0.0566	0.0374	0.1017
0.5499	19.0	11438	0.5676	0.0027	0.1956	0.0566	0.0374	0.1016
0.5451	20.0	12040	0.5676	0.0027	0.1956	0.0566	0.0374	0.1016

Framework versions

• Transformers 4.52.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1