works on
fine-tuning 1hypernetwork 1language model safety 1post hoc safety 1prompt classification 1

From the 1 of 16 linked papers with an AI index.

activity
20242026
most citedDistractor-Aware Video Object Segmentation

3 citations · 3 across the 9 of their papers we have counts for

collaborators
Showing cs.LGShow all

1 paper · 1 filter

cs.LG2026

HyperSafe: Inference-Time Safety Recovery for Fine-Tuned Language Models

Aznaur Aliev, Carlos Hinojosa, Abdelrahman Eldesokey +3

HyperSafe introduces a post‑hoc, model‑specific safe side network generated by a hypernetwork that classifies prompts using activation fingerprints, allowing fine‑tuned language mo…