5 papers
SteerCheck: Attribution Specificity and Alignment Leakage in Activation-Steering Audits
Daming Luo, Christy Liang, Junyu Xuan
Activation steering can change behaviour without establishing that the effect is specific to the intended concept. We introduce SteerCheck, a preregistered attribution audit that m…
How Query Visibility Changes KV-Cache Compression Rankings: A Matched-Budget Audit
Daming Luo, Christy Liang, Junyu Xuan
KV-cache compression methods are predominantly evaluated with the query appended to the context before compression -- a query-aware protocol. Yet the economic case for a compressed…
InforME: Improving Informativeness of Abstractive Text Summarization With Informative Attention Guided by Named Entity Salience
Jianbin Shen, Christy Jie Liang, Junyu Xuan
Abstractive text summarization is integral to the Big Data era, which demands advanced methods to turn voluminous and often long text data into concise but coherent and informative…
A Non-Monolithic Policy Approach of Offline-to-Online Reinforcement Learning
JaeYoon Kim, Junyu Xuan, Christy Liang +1
Offline-to-online reinforcement learning (RL) leverages both pre-trained offline policies and online policies trained for downstream tasks, aiming to improve data efficiency and ac…
Decoupling Exploration and Exploitation for Unsupervised Pre-training with Successor Features
JaeYoon Kim, Junyu Xuan, Christy Liang +1
Unsupervised pre-training has been on the lookout for the virtue of a value function representation referred to as successor features (SFs), which decouples the dynamics of the env…