speaker-segmentation-fine-tuned-callhome-jpn
This model is a fine-tuned version of openai/whisper-small on the diarizers-community/callhome dataset. It achieves the following results on the evaluation set:
- Loss: 0.4792
- Der: 0.1886
- False Alarm: 0.0606
- Missed Detection: 0.0736
- Confusion: 0.0544
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 | Der | False Alarm | Missed Detection | Confusion |
|---|---|---|---|---|---|---|---|
| 0.3877 | 1.0 | 362 | 0.4746 | 0.1886 | 0.0603 | 0.0737 | 0.0547 |
| 0.4206 | 2.0 | 724 | 0.4904 | 0.1956 | 0.0656 | 0.0733 | 0.0566 |
| 0.3933 | 3.0 | 1086 | 0.4704 | 0.1855 | 0.0589 | 0.0737 | 0.0529 |
| 0.3708 | 4.0 | 1448 | 0.4759 | 0.1879 | 0.0600 | 0.0739 | 0.0540 |
| 0.3516 | 5.0 | 1810 | 0.4792 | 0.1886 | 0.0606 | 0.0736 | 0.0544 |
Framework versions
- Transformers 4.40.2
- Pytorch 2.2.1+cu121
- Datasets 2.19.1
- Tokenizers 0.19.1
- Downloads last month
- 2
Model tree for mmufti/speaker-segmentation-fine-tuned-callhome-jpn
Base model
openai/whisper-small