README.md

---
license: mit
base_model: pyannote/segmentation-3.0
tags:
- speaker-diarization
- speaker-segmentation
- generated_from_trainer
datasets:
- KMayanja/backup_uganda
model-index:
- name: speaker-segmentation-fine-tuned-backup-uganda
  results: []
---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

# speaker-segmentation-fine-tuned-backup-uganda

This model is a fine-tuned version of [pyannote/segmentation-3.0](https://huggingface.co/pyannote/segmentation-3.0) on the KMayanja/backup_uganda default dataset.
It achieves the following results on the evaluation set:
- Loss: 0.2271
- Der: 0.0667
- False Alarm: 0.0188
- Missed Detection: 0.0260
- Confusion: 0.0219

## 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 |
|:-------------:|:-----:|:----:|:---------------:|:------:|:-----------:|:----------------:|:---------:|
| 0.1819        | 1.0   | 266  | 0.2174          | 0.0663 | 0.0186      | 0.0249           | 0.0228    |
| 0.1659        | 2.0   | 532  | 0.2177          | 0.0669 | 0.0169      | 0.0278           | 0.0221    |
| 0.1549        | 3.0   | 798  | 0.2170          | 0.0659 | 0.0181      | 0.0261           | 0.0217    |
| 0.1535        | 4.0   | 1064 | 0.2222          | 0.0666 | 0.0195      | 0.0251           | 0.0220    |
| 0.1541        | 5.0   | 1330 | 0.2271          | 0.0667 | 0.0188      | 0.0260           | 0.0219    |


### Framework versions

- Transformers 4.42.4
- Pytorch 2.3.1+cu121
- Datasets 2.20.0
- Tokenizers 0.19.1