metadata
library_name: transformers
language:
- fr
license: mit
base_model: pyannote/segmentation-3.0
tags:
- speaker-diarization
- speaker-segmentation
- generated_from_trainer
datasets:
- MPF
model-index:
- name: speaker-segmentation-ft-mpf
results: []
speaker-segmentation-ft-mpf
This model is a fine-tuned version of pyannote/segmentation-3.0 on the MPF dataset. It achieves the following results on the evaluation set:
- Loss: 0.8806
- Model Preparation Time: 0.004
- Der: 0.3412
- False Alarm: 0.0953
- Missed Detection: 0.1247
- Confusion: 0.1211
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: 10
Training results
| Training Loss | Epoch | Step | Validation Loss | Model Preparation Time | Der | False Alarm | Missed Detection | Confusion |
|---|---|---|---|---|---|---|---|---|
| 0.9823 | 1.0 | 1302 | 0.9267 | 0.004 | 0.3729 | 0.0896 | 0.1466 | 0.1367 |
| 0.9422 | 2.0 | 2604 | 0.9247 | 0.004 | 0.3695 | 0.0991 | 0.1282 | 0.1422 |
| 0.915 | 3.0 | 3906 | 0.8984 | 0.004 | 0.3549 | 0.0878 | 0.1397 | 0.1273 |
| 0.8706 | 4.0 | 5208 | 0.9055 | 0.004 | 0.3543 | 0.1036 | 0.1171 | 0.1336 |
| 0.8511 | 5.0 | 6510 | 0.8776 | 0.004 | 0.3424 | 0.0932 | 0.1279 | 0.1213 |
| 0.834 | 6.0 | 7812 | 0.8896 | 0.004 | 0.3485 | 0.1004 | 0.1220 | 0.1261 |
| 0.8505 | 7.0 | 9114 | 0.8777 | 0.004 | 0.3418 | 0.0959 | 0.1260 | 0.1199 |
| 0.8452 | 8.0 | 10416 | 0.8802 | 0.004 | 0.3430 | 0.0975 | 0.1246 | 0.1209 |
| 0.837 | 9.0 | 11718 | 0.8792 | 0.004 | 0.3407 | 0.0950 | 0.1247 | 0.1211 |
| 0.8459 | 10.0 | 13020 | 0.8806 | 0.004 | 0.3412 | 0.0953 | 0.1247 | 0.1211 |
Framework versions
- Transformers 4.45.2
- Pytorch 2.4.1+cu121
- Datasets 3.0.1
- Tokenizers 0.20.0