speaker-segmentation-fine-tuned-callhome-eng
This model is a fine-tuned version of pyannote/segmentation-3.0 on the diarizers-community/callhome eng dataset. It achieves the following results on the evaluation set:
- Loss: 0.4570
- Der: 0.1803
- False Alarm: 0.0556
- Missed Detection: 0.0731
- Confusion: 0.0516
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: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- num_epochs: 5.0
Training results
| Training Loss | Epoch | Step | Validation Loss | Der | False Alarm | Missed Detection | Confusion |
|---|---|---|---|---|---|---|---|
| 0.4257 | 1.0 | 362 | 0.4789 | 0.1918 | 0.0573 | 0.0786 | 0.0559 |
| 0.3889 | 2.0 | 724 | 0.4660 | 0.1866 | 0.0556 | 0.0760 | 0.0549 |
| 0.3758 | 3.0 | 1086 | 0.4587 | 0.1807 | 0.0548 | 0.0755 | 0.0503 |
| 0.3643 | 4.0 | 1448 | 0.4564 | 0.1805 | 0.0555 | 0.0734 | 0.0515 |
| 0.3511 | 5.0 | 1810 | 0.4570 | 0.1803 | 0.0556 | 0.0731 | 0.0516 |
Framework versions
- Transformers 4.40.1
- Pytorch 2.2.0+cu121
- Datasets 2.17.0
- Tokenizers 0.19.1