callhome_finetuned_v1

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

• Loss: 0.4727
• Model Preparation Time: 0.004
• Der: 0.1798
• False Alarm: 0.0603
• Missed Detection: 0.0699
• Confusion: 0.0496

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: 10.0

Training results

Training Loss	Epoch	Step	Validation Loss	Model Preparation Time	Der	False Alarm	Missed Detection	Confusion
0.4176	1.0	362	0.4821	0.004	0.1938	0.0608	0.0745	0.0584
0.3931	2.0	724	0.4564	0.004	0.1853	0.0552	0.0760	0.0541
0.3662	3.0	1086	0.4535	0.004	0.1812	0.0521	0.0765	0.0526
0.3575	4.0	1448	0.4660	0.004	0.1815	0.0590	0.0699	0.0526
0.3449	5.0	1810	0.4605	0.004	0.1802	0.0629	0.0670	0.0503
0.3288	6.0	2172	0.4659	0.004	0.1794	0.0573	0.0717	0.0504
0.3215	7.0	2534	0.4676	0.004	0.1803	0.0593	0.0709	0.0501
0.3261	8.0	2896	0.4667	0.004	0.1786	0.0597	0.0702	0.0488
0.3193	9.0	3258	0.4696	0.004	0.1796	0.0597	0.0705	0.0494
0.3203	10.0	3620	0.4727	0.004	0.1798	0.0603	0.0699	0.0496

Framework versions

• Transformers 4.49.0
• Pytorch 2.7.0+cu126
• Datasets 3.5.0
• Tokenizers 0.21.0