speaker-segmentation-fine-tuned
This model is a fine-tuned version of pyannote/segmentation-3.0 on the syvai/synthetic-diarization-mixed-speakers dataset. It achieves the following results on the evaluation set:
- Loss: 0.2088
- Model Preparation Time: 0.0013
- Der: 0.0754
- False Alarm: 0.0218
- Missed Detection: 0.0180
- Confusion: 0.0356
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 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: 5.0
Training results
| Training Loss | Epoch | Step | Validation Loss | Model Preparation Time | Der | False Alarm | Missed Detection | Confusion |
|---|---|---|---|---|---|---|---|---|
| 0.2579 | 1.0 | 1232 | 0.2571 | 0.0013 | 0.0941 | 0.0253 | 0.0190 | 0.0498 |
| 0.2437 | 2.0 | 2464 | 0.2483 | 0.0013 | 0.0898 | 0.0230 | 0.0196 | 0.0471 |
| 0.2169 | 3.0 | 3696 | 0.2200 | 0.0013 | 0.0787 | 0.0232 | 0.0179 | 0.0376 |
| 0.206 | 4.0 | 4928 | 0.2104 | 0.0013 | 0.0760 | 0.0218 | 0.0181 | 0.0361 |
| 0.1928 | 5.0 | 6160 | 0.2088 | 0.0013 | 0.0754 | 0.0218 | 0.0180 | 0.0356 |
Framework versions
- Transformers 4.53.0
- Pytorch 2.7.1+cu126
- Datasets 3.6.0
- Tokenizers 0.21.2