speaker-segmentation-fine-tuned-callhome-spa
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: 0.5724
- Der: 0.3391
- False Alarm: 0.2612
- Missed Detection: 0.0776
- Confusion: 0.0003
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: 64
- eval_batch_size: 64
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- num_epochs: 5
Training results
| Training Loss | Epoch | Step | Validation Loss | Der | False Alarm | Missed Detection | Confusion |
|---|---|---|---|---|---|---|---|
| 0.4184 | 1.0 | 230 | 0.4700 | 0.2893 | 0.2341 | 0.0546 | 0.0006 |
| 0.4075 | 2.0 | 460 | 0.5348 | 0.3197 | 0.2567 | 0.0625 | 0.0005 |
| 0.3941 | 3.0 | 690 | 0.5296 | 0.3134 | 0.2608 | 0.0525 | 0.0001 |
| 0.3902 | 4.0 | 920 | 0.5936 | 0.3624 | 0.2612 | 0.1009 | 0.0003 |
| 0.389 | 5.0 | 1150 | 0.5724 | 0.3391 | 0.2612 | 0.0776 | 0.0003 |
Framework versions
- Transformers 4.45.1
- Pytorch 2.4.1
- Datasets 3.0.1
- Tokenizers 0.20.0