4 papers · 1 filter
Detecting High-Stakes Interactions with Activation Probes
Alex McKenzie, Urja Pawar, Phil Blandfort +4
Monitoring is an important aspect of safely deploying Large Language Models (LLMs). This paper examines activation probes for detecting ``high-stakes'' interactions -- where the te…
Robust Multi-Objective Controlled Decoding of Large Language Models
Seongho Son, William Bankes, Sangwoong Yoon +3
We introduce Robust Multi-Objective Decoding (RMOD), a novel inference-time algorithm that robustly aligns Large Language Models (LLMs) to multiple human objectives (e.g., instruct…
Right Now, Wrong Then: Non-Stationary Direct Preference Optimization under Preference Drift
Seongho Son, William Bankes, Sayak Ray Chowdhury +2
Current Large Language Model (LLM) preference optimization algorithms do not account for temporal preference drift, which can lead to severe misalignment. To address this limitatio…
REDUCR: Robust Data Downsampling Using Class Priority Reweighting
William Bankes, George Hughes, Ilija Bogunovic +1
Modern machine learning models are becoming increasingly expensive to train for real-world image and text classification tasks, where massive web-scale data is collected in a strea…