• Bio
  • Papers
  • Talks
  • News
  • Experience
  • Projects
  • Teaching
  • News
    • 🎉 RogueAI accepted at CLiC-it 2026 (Poster)
    • 🎤 LM Taboo accepted at CLiC-it 2026 (Oral)
    • 🧩 Participating in the NDIF Aletheia's Quest Competition
    • 🎤 Paper accepted at NeSy 2026 (Oral)
    • 🗣️ Invited talk at the CLCG Linguistics Lunch
    • Visiting researcher period at CLCG, University of Groningen concluded
  • Publications
    • Beyond the Commitment Boundary: Probing Epiphenomenal Chain-of-Thought in Large Reasoning Models
    • RogueAI: A Reverse Turing Test for Detecting Licensed AI Deception in Dialogue
    • Distilling Formal Logic into Neural Spaces: A Kernel Alignment Approach for Signal Temporal Logic
    • A Dialectic Pipeline for Improving LLM Robustness
    • Bridging Logic and Learning: Decoding Temporal Logic Embeddings via Transformers
  • Recent & Upcoming Talks
    • Reading Between the Tokens: Uncovering the Semantic Minima of AI Monologues
    • Large Language Models: Potenzialità, Limiti e Sistemi Multi-Agent
    • Bridging Logic and Learning: Decoding Temporal Logic Embeddings via Transformers
  • Projects
    • Distilling Formal Logic into Neural Spaces
    • STLDec: Decoding Temporal Logic Embeddings via Transformers
    • OverRef: Studying Over-Refusal in Large Language Models
  • Teaching
    • Probabilistic Machine Learning
  • Projects
  • Experience
  • Blog

OverRef: Studying Over-Refusal in Large Language Models

Jan 1, 2025 · 1 min read
Go to Project Site

Ongoing project on over-refusal in LLMs: studying when and why models refuse legitimate user queries, with benchmarking and dataset resources.

Last updated on Jan 1, 2025
Large Language Models Evaluation Benchmarking
Sara Candussio
Authors
Sara Candussio
PhD Student (2nd year)
← STLDec: Decoding Temporal Logic Embeddings via Transformers Jul 10, 2025

© 2026 Me. This work is licensed under CC BY NC ND 4.0

Published with Hugo Blox Builder — the free, open source website builder that empowers creators.