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

Loss: 0.3828
Model Preparation Time: 0.0078
Der: 0.1460
False Alarm: 0.0572
Missed Detection: 0.0634
Confusion: 0.0254

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
0.4499	1.0	359	0.3920	0.0078	0.1518	0.0434	0.0774	0.0310
0.4236	2.0	718	0.3838	0.0078	0.1500	0.0492	0.0711	0.0296
0.4148	3.0	1077	0.3855	0.0078	0.1494	0.0468	0.0733	0.0293
0.3918	4.0	1436	0.3786	0.0078	0.1464	0.0538	0.0654	0.0272
0.3737	5.0	1795	0.3766	0.0078	0.1468	0.0576	0.0624	0.0267
0.3677	6.0	2154	0.3750	0.0078	0.1449	0.0553	0.0642	0.0253
0.3539	7.0	2513	0.3649	0.0078	0.1413	0.0530	0.0656	0.0227
0.3447	8.0	2872	0.3689	0.0078	0.1425	0.0517	0.0661	0.0246
0.3387	9.0	3231	0.3716	0.0078	0.1412	0.0483	0.0676	0.0252
0.3314	10.0	3590	0.3758	0.0078	0.1429	0.0503	0.0667	0.0259
0.3257	11.0	3949	0.3795	0.0078	0.1435	0.0533	0.0656	0.0246
0.3245	12.0	4308	0.3799	0.0078	0.1450	0.0550	0.0638	0.0262
0.3155	13.0	4667	0.3792	0.0078	0.1452	0.0558	0.0631	0.0263
0.3099	14.0	5026	0.3835	0.0078	0.1478	0.0585	0.0627	0.0266
0.3152	15.0	5385	0.3799	0.0078	0.1445	0.0546	0.0642	0.0257
0.3081	16.0	5744	0.3846	0.0078	0.1464	0.0567	0.0638	0.0259
0.299	17.0	6103	0.3838	0.0078	0.1451	0.0557	0.0644	0.0250
0.3023	18.0	6462	0.3828	0.0078	0.1464	0.0581	0.0630	0.0253
0.2971	19.0	6821	0.3824	0.0078	0.1458	0.0572	0.0633	0.0253
0.2984	20.0	7180	0.3828	0.0078	0.1460	0.0572	0.0634	0.0254

Framework versions

Transformers 4.56.2
Pytorch 2.7.1+cu118
Datasets 3.6.0
Tokenizers 0.22.1

echojosh/speaker-segmentation-fine-tuned-callhome-zho-0927

speaker-segmentation-fine-tuned-callhome-zho-0927

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

echojosh/
speaker-segmentation-fine-tuned-callhome-zho-0927