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

Loss: 0.2745
Model Preparation Time: 0.0069
Der: 0.0853
False Alarm: 0.0344
Missed Detection: 0.0236
Confusion: 0.0273

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: 15.0

Training results

Training Loss	Epoch	Step	Validation Loss	Model Preparation Time	Der	False Alarm	Missed Detection	Confusion
0.2419	1.0	791	0.2628	0.0069	0.0909	0.0363	0.0244	0.0302
0.2356	2.0	1582	0.2557	0.0069	0.0904	0.0361	0.0237	0.0306
0.2129	3.0	2373	0.2555	0.0069	0.0864	0.0334	0.0249	0.0280
0.2085	4.0	3164	0.2436	0.0069	0.0849	0.0306	0.0275	0.0269
0.1946	5.0	3955	0.2571	0.0069	0.0857	0.0330	0.0248	0.0278
0.1863	6.0	4746	0.2503	0.0069	0.0860	0.0356	0.0231	0.0273
0.1763	7.0	5537	0.2526	0.0069	0.0858	0.0351	0.0234	0.0273
0.1783	8.0	6328	0.2571	0.0069	0.0857	0.0319	0.0253	0.0285
0.1704	9.0	7119	0.2582	0.0069	0.0861	0.0326	0.0253	0.0281
0.1719	10.0	7910	0.2626	0.0069	0.0862	0.0323	0.0259	0.0281
0.1625	11.0	8701	0.2713	0.0069	0.0860	0.0330	0.0245	0.0284
0.1668	12.0	9492	0.2753	0.0069	0.0857	0.0345	0.0235	0.0277
0.1579	13.0	10283	0.2741	0.0069	0.0850	0.0345	0.0233	0.0271
0.1609	14.0	11074	0.2748	0.0069	0.0851	0.0344	0.0235	0.0272
0.1637	15.0	11865	0.2745	0.0069	0.0853	0.0344	0.0236	0.0273

Framework versions

Transformers 4.50.2
Pytorch 2.6.0+cu124
Datasets 3.5.0
Tokenizers 0.21.1

HV-Khurdula/speaker-segmentation-0.2-vox

speaker-segmentation-0.2-vox

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

HV-Khurdula/
speaker-segmentation-0.2-vox