This model is a fine-tuned version of pyannote/speaker-diarization-3.1 on the speaker-segmentation dataset. It achieves the following results on the evaluation set:

Loss: 0.3830
Model Preparation Time: 0.0032
Der: 0.1285
False Alarm: 0.0204
Missed Detection: 0.0113
Confusion: 0.0968

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: 0.0003
train_batch_size: 64
eval_batch_size: 64
seed: 42
optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_ratio: 0.15
num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Model Preparation Time	Der	False Alarm	Missed Detection	Confusion
0.5423	1.0	400	0.5831	0.0032	0.1971	0.0254	0.0205	0.1511
0.4998	2.0	800	0.5031	0.0032	0.1721	0.0219	0.0149	0.1353
0.4588	3.0	1200	0.4654	0.0032	0.1577	0.0206	0.0134	0.1237
0.4467	4.0	1600	0.4401	0.0032	0.1511	0.0202	0.0130	0.1179
0.4192	5.0	2000	0.4276	0.0032	0.1432	0.0205	0.0122	0.1105
0.3991	6.0	2400	0.4157	0.0032	0.1413	0.0205	0.0118	0.1090
0.4042	7.0	2800	0.4016	0.0032	0.1355	0.0207	0.0114	0.1034
0.3972	8.0	3200	0.3959	0.0032	0.1330	0.0205	0.0113	0.1012
0.4001	9.0	3600	0.3914	0.0032	0.1324	0.0204	0.0114	0.1006
0.3733	10.0	4000	0.3892	0.0032	0.1310	0.0204	0.0114	0.0992
0.3835	11.0	4400	0.3855	0.0032	0.1300	0.0204	0.0113	0.0983
0.3716	12.0	4800	0.3859	0.0032	0.1293	0.0206	0.0111	0.0976
0.3655	13.0	5200	0.3836	0.0032	0.1286	0.0204	0.0113	0.0969
0.3743	14.0	5600	0.3833	0.0032	0.1284	0.0204	0.0113	0.0967
0.3592	15.0	6000	0.3830	0.0032	0.1285	0.0204	0.0113	0.0968

Framework versions

Transformers 4.51.1
Pytorch 2.6.0+cu124
Datasets 3.5.0
Tokenizers 0.21.1

whitneyten/pydiarize-Dataset-2_4-updated_03

speaker-segmentation-fine-tuned-datasetID-hugging_2_4_updated_03

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

whitneyten/
pydiarize-Dataset-2_4-updated_03