---
library_name: transformers
license: mit
base_model: pyannote/segmentation-3.0
tags:
- speaker-diarization
- speaker-segmentation
- generated_from_trainer
datasets:
- objects76/synthetic-ja4-speaker-overlap-6400
model-index:
- name: full-ja4-2.25sec-big-rf
  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. -->

# full-ja4-2.25sec-big-rf

This model is a fine-tuned version of [pyannote/segmentation-3.0](https://huggingface.co/pyannote/segmentation-3.0) on the objects76/synthetic-ja4-speaker-overlap-6400 dataset.
It achieves the following results on the evaluation set:
- eval_loss: 1.3497
- eval_der: 0.3641
- eval_false_alarm: 0.0785
- eval_missed_detection: 0.2430
- eval_confusion: 0.0426
- eval_runtime: 2.1638
- eval_samples_per_second: 295.775
- eval_steps_per_second: 0.462
- step: 0

## 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: 2048
- eval_batch_size: 2048
- seed: 42
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- num_epochs: 200

### Framework versions

- Transformers 4.51.3
- Pytorch 2.6.0+cu124
- Datasets 3.5.0
- Tokenizers 0.21.1