speaker-segmentation-fine-tuned-callfriend-jpn
This model is a fine-tuned version of pyannote/segmentation-3.0 on the diarizers-community/callfriend jpn dataset. It achieves the following results on the evaluation set:
- Loss: 0.6570
- Der: 0.2815
- False Alarm: 0.1035
- Missed Detection: 0.1082
- Confusion: 0.0698
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 |
|---|---|---|---|---|---|---|---|
| 0.6836 | 1.0 | 237 | 0.6465 | 0.2865 | 0.1045 | 0.1095 | 0.0724 |
| 0.6265 | 2.0 | 474 | 0.6455 | 0.2772 | 0.1023 | 0.1062 | 0.0687 |
| 0.6199 | 3.0 | 711 | 0.6615 | 0.2879 | 0.0950 | 0.1161 | 0.0768 |
| 0.6007 | 4.0 | 948 | 0.6574 | 0.2823 | 0.1051 | 0.1066 | 0.0705 |
| 0.5979 | 5.0 | 1185 | 0.6570 | 0.2815 | 0.1035 | 0.1082 | 0.0698 |
Framework versions
- Transformers 4.40.0
- Pytorch 2.2.2+cu121
- Datasets 2.18.0
- Tokenizers 0.19.1
- Downloads last month
- 13
Model tree for kamilakesbi/speaker-segmentation-fine-tuned-callfriend-jpn
Base model
pyannote/segmentation-3.0