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

Loss: 0.7571
Model Preparation Time: 0.007
Der: 0.2254
False Alarm: 0.0476
Missed Detection: 0.1333
Confusion: 0.0444

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: 43
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
num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Model Preparation Time	Der	False Alarm	Missed Detection	Confusion
0.5711	1.0	328	0.7534	0.007	0.2332	0.0465	0.1395	0.0471
0.5219	2.0	656	0.7573	0.007	0.2306	0.0459	0.1384	0.0463
0.5102	3.0	984	0.7540	0.007	0.2254	0.0491	0.1329	0.0433
0.5439	4.0	1312	0.7564	0.007	0.2273	0.0494	0.1336	0.0443
0.4975	5.0	1640	0.7571	0.007	0.2254	0.0476	0.1333	0.0444

Framework versions

Transformers 4.50.0
Pytorch 2.6.0+cu124
Datasets 3.5.0
Tokenizers 0.21.1

whitneyten/pydiarize-synthetic-japan

speaker-segmentation-fine-tuned-callhome-jpn

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

whitneyten/
pydiarize-synthetic-japan