speaker-segmentation-fine-tuned-callhome-jpn
This model is a fine-tuned version of pyannote/segmentation-3.0 on the diarizers-community/callhome jpn dataset. It achieves the following results on the evaluation set:
- Loss: 0.7433
- Der: 0.2234
- False Alarm: 0.0478
- Missed Detection: 0.1328
- Confusion: 0.0428
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 |
---|---|---|---|---|---|---|---|
0.5771 | 1.0 | 328 | 0.7534 | 0.2321 | 0.0564 | 0.1261 | 0.0496 |
0.5388 | 2.0 | 656 | 0.7503 | 0.2261 | 0.0485 | 0.1347 | 0.0429 |
0.5061 | 3.0 | 984 | 0.7486 | 0.2248 | 0.0475 | 0.1350 | 0.0423 |
0.4883 | 4.0 | 1312 | 0.7374 | 0.2227 | 0.0492 | 0.1315 | 0.0421 |
0.493 | 5.0 | 1640 | 0.7433 | 0.2234 | 0.0478 | 0.1328 | 0.0428 |
Framework versions
- Transformers 4.40.0
- Pytorch 2.2.2+cu121
- Datasets 2.18.0
- Tokenizers 0.19.1
- Downloads last month
- 11
Unable to determine this model’s pipeline type. Check the
docs
.