metadata
language:
- en
license: apache-2.0
base_model: openai/whisper-small
tags:
- speaker-diarization
- speaker-segmentation
- generated_from_trainer
model-index:
- name: speaker-segmentation-eng
results: []
speaker-segmentation-eng
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: 0.4666
- Der: 0.1827
- False Alarm: 0.0590
- Missed Detection: 0.0715
- Confusion: 0.0522
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: 64
- eval_batch_size: 64
- 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 | Der | False Alarm | Missed Detection | Confusion |
|---|---|---|---|---|---|---|---|
| 0.4224 | 1.0 | 181 | 0.4837 | 0.1939 | 0.0599 | 0.0764 | 0.0576 |
| 0.409 | 2.0 | 362 | 0.4692 | 0.1884 | 0.0618 | 0.0724 | 0.0543 |
| 0.3919 | 3.0 | 543 | 0.4700 | 0.1875 | 0.0638 | 0.0698 | 0.0540 |
| 0.3693 | 4.0 | 724 | 0.4718 | 0.1848 | 0.0602 | 0.0714 | 0.0533 |
| 0.358 | 5.0 | 905 | 0.4606 | 0.1810 | 0.0544 | 0.0754 | 0.0512 |
| 0.355 | 6.0 | 1086 | 0.4631 | 0.1826 | 0.0638 | 0.0677 | 0.0512 |
| 0.3563 | 7.0 | 1267 | 0.4646 | 0.1809 | 0.0587 | 0.0716 | 0.0505 |
| 0.347 | 8.0 | 1448 | 0.4682 | 0.1820 | 0.0581 | 0.0720 | 0.0519 |
| 0.3463 | 9.0 | 1629 | 0.4684 | 0.1827 | 0.0586 | 0.0718 | 0.0523 |
| 0.3299 | 10.0 | 1810 | 0.4666 | 0.1827 | 0.0590 | 0.0715 | 0.0522 |
Framework versions
- Transformers 4.40.1
- Pytorch 2.2.1+cu121
- Datasets 2.19.1
- Tokenizers 0.19.1