metadata
license: mit
base_model: pyannote/segmentation-3.0
tags:
- speaker-diarization
- speaker-segmentation
- generated_from_trainer
datasets:
- diarizers-community/callhome
model-index:
- name: speaker-segmentation-fine-tuned-callhome-eng
results: []
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: 0.4654
- Der: 0.1832
- False Alarm: 0.0599
- Missed Detection: 0.0724
- Confusion: 0.0508
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: 5.0
Training results
| Training Loss | Epoch | Step | Validation Loss | Der | False Alarm | Missed Detection | Confusion |
|---|---|---|---|---|---|---|---|
| 0.4455 | 1.0 | 181 | 0.4776 | 0.1948 | 0.0699 | 0.0691 | 0.0558 |
| 0.4049 | 2.0 | 362 | 0.4746 | 0.1916 | 0.0590 | 0.0763 | 0.0562 |
| 0.3856 | 3.0 | 543 | 0.4631 | 0.1843 | 0.0565 | 0.0754 | 0.0524 |
| 0.3796 | 4.0 | 724 | 0.4634 | 0.1834 | 0.0593 | 0.0726 | 0.0515 |
| 0.3727 | 5.0 | 905 | 0.4654 | 0.1832 | 0.0599 | 0.0724 | 0.0508 |
Framework versions
- Transformers 4.41.2
- Pytorch 2.3.1+cu121
- Datasets 2.19.2
- Tokenizers 0.19.1