metadata
library_name: transformers
language:
- fr
license: mit
base_model: pyannote/speaker-diarization-3.1
tags:
- speaker-diarization
- speaker-segmentation
- generated_from_trainer
datasets:
- ESLO2
model-index:
- name: speaker-segmentation-fine-tuned_ESLO2
results: []
speaker-segmentation-fine-tuned_ESLO2
This model is a fine-tuned version of pyannote/speaker-diarization-3.1 on the ESLO2 dataset. It achieves the following results on the evaluation set:
- Loss: 1.0390
- Model Preparation Time: 0.004
- Der: 0.5468
- False Alarm: 0.1928
- Missed Detection: 0.2690
- Confusion: 0.0849
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
Training results
| Training Loss | Epoch | Step | Validation Loss | Model Preparation Time | Der | False Alarm | Missed Detection | Confusion |
|---|---|---|---|---|---|---|---|---|
| 1.0681 | 1.0 | 686 | 1.0642 | 0.004 | 0.5632 | 0.1931 | 0.2781 | 0.0921 |
| 1.0147 | 2.0 | 1372 | 1.0523 | 0.004 | 0.5551 | 0.1887 | 0.2772 | 0.0893 |
| 1.0044 | 3.0 | 2058 | 1.0445 | 0.004 | 0.5537 | 0.1887 | 0.2791 | 0.0859 |
| 0.9801 | 4.0 | 2744 | 1.0352 | 0.004 | 0.5464 | 0.1934 | 0.2677 | 0.0853 |
| 0.9866 | 5.0 | 3430 | 1.0390 | 0.004 | 0.5468 | 0.1928 | 0.2690 | 0.0849 |
Framework versions
- Transformers 4.45.2
- Pytorch 2.4.1+cu121
- Datasets 3.0.1
- Tokenizers 0.20.0