speaker-segmentation-fine-tuned-ami
This model is a fine-tuned version of pyannote/segmentation-3.0 on the diarizers-community/ami ihm dataset. It achieves the following results on the evaluation set:
- Loss: 0.3660
- Der: 0.1396
- False Alarm: 0.0503
- Missed Detection: 0.0578
- Confusion: 0.0314
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 |
|---|---|---|---|---|---|---|---|
| 0.4133 | 1.0 | 1427 | 0.3629 | 0.1388 | 0.0424 | 0.0646 | 0.0318 |
| 0.3907 | 2.0 | 2854 | 0.3638 | 0.1400 | 0.0492 | 0.0583 | 0.0324 |
| 0.3651 | 3.0 | 4281 | 0.3631 | 0.1403 | 0.0506 | 0.0581 | 0.0316 |
| 0.3692 | 4.0 | 5708 | 0.3643 | 0.1394 | 0.0489 | 0.0591 | 0.0314 |
| 0.3484 | 5.0 | 7135 | 0.3660 | 0.1396 | 0.0503 | 0.0578 | 0.0314 |
Framework versions
- Transformers 4.40.1
- Pytorch 2.2.0+cu121
- Datasets 2.17.0
- Tokenizers 0.19.1