---
language: en
tags:
- disfluency-detection
- token-classification
- modernbert
- speech-pathology
datasets:
- arielcerdap/disfluency-fluencybank
---

# ModernBERT Disfluency Detection — Exp A (Real Only)

Fine-tuned from [answerdotai/ModernBERT-base](https://huggingface.co/answerdotai/ModernBERT-base) on **real-only** FluencyBank Timestamped data.

## Dataset
- Config: `real_only` de `arielcerdap/disfluency-fluencybank`
- Train: 2737 segmentos (100% reales)
- Val/Test: idénticos a Exp B para comparación directa

## Labels
O · FP (filled pause) · RP (repetition) · RV (revision) · PW (partial word)

## Test Results
| Label | P | R | F1 | Support |
|---|---|---|---|---|
| Label | P | R | F1 | Support |
|---|---|---|---|---|
| O | 0.9783 | 0.9860 | 0.9821 | 3704 |
| FP | 0.9888 | 1.0000 | 0.9944 | 176 |
| RP | 0.7784 | 0.8276 | 0.8022 | 174 |
| RV | 0.3425 | 0.2907 | 0.3145 | 86 |
| PW | 0.9510 | 0.8326 | 0.8879 | 233 |

**Macro F1 (4 disfluencias):** 0.7497  
**Binary F1:** 0.8902

## Hyperparameters
- learning_rate: 5e-05
- epochs: 15
- warmup_steps: 191
- weight_decay: 0.1
- focal_loss_gamma: 3.0 (adaptive)
- class_weights: O=1.0, FP=3.0, RP=6.0, RV=20.0, PW=5.0