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-deu
results: []
speaker-segmentation-fine-tuned-callhome-deu
This model is a fine-tuned version of pyannote/segmentation-3.0 on the diarizers-community/callhome deu dataset. It achieves the following results on the evaluation set:
- Loss: 0.3780
- Der: 0.1415
- False Alarm: 0.0724
- Missed Detection: 0.0490
- Confusion: 0.0201
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 |
|---|---|---|---|---|---|---|---|
| 0.4622 | 1.0 | 330 | 0.3844 | 0.1439 | 0.0653 | 0.0562 | 0.0223 |
| 0.4306 | 2.0 | 660 | 0.4004 | 0.1519 | 0.0763 | 0.0515 | 0.0241 |
| 0.4069 | 3.0 | 990 | 0.3775 | 0.1407 | 0.0707 | 0.0496 | 0.0204 |
| 0.3949 | 4.0 | 1320 | 0.3771 | 0.1408 | 0.0710 | 0.0498 | 0.0200 |
| 0.3879 | 5.0 | 1650 | 0.3780 | 0.1415 | 0.0724 | 0.0490 | 0.0201 |
Framework versions
- Transformers 4.40.0
- Pytorch 2.2.2+cu121
- Datasets 2.18.0
- Tokenizers 0.19.1