8 papers
Grounding latent algorithm routing in transformer reasoning
Xiangbo Zhang, Xiaoxu Ma
A central question in the in-context learning literature is whether transformers can organize episode-level adaptation around different inductive-bias families. We study this quest…
Attribute-Prompted Kernel Hashing for Unsupervised Data-Efficient Cross-Modal Retrieval
Runhao Li, Xiaoxu Ma, Zhenyu Weng +5
Unsupervised cross-modal hashing enables efficient retrieval of semantically related instances across different modalities without requiring manual semantic annotation. However, ex…
Unsupervised Data-Efficient Cross-Modal Retrieval with Global-Neighborhood Alignment Hashing
Runhao Li, Xiaoxu Ma, Zhenyu Weng +5
Compared to supervised cross-modal hashing (CMH), unsupervised CMH reduces the reliance on manual labeling by learning binary codes from unlabeled image-text pairs. However, existi…
Input-Envelope-Output: Auditable Generative Music Rewards in Sensory-Sensitive Contexts
Cong Ye, Songlin Shang, Xiaoxu Ma +1
Generative feedback in sensory-sensitive contexts poses a core design challenge: large individual differences in sensory tolerance make it difficult to sustain engagement without c…
UniHash: Unifying Pointwise and Pairwise Hashing Paradigms
Xiaoxu Ma, Runhao Li, Xiangbo Zhang +1
Effective retrieval across both seen and unseen categories is crucial for modern image retrieval systems. Retrieval on seen categories ensures precise recognition of known classes,…
Stable and Explainable Personality Trait Evaluation in Large Language Models with Internal Activations
Xiaoxu Ma, Xiangbo Zhang, Zhenyu Weng
Evaluating personality traits in Large Language Models (LLMs) is key to model interpretation, comparison, and responsible deployment. However, existing questionnaire-based evaluati…
Mutual Learning for Hashing: Unlocking Strong Hash Functions from Weak Supervision
Xiaoxu Ma, Runhao Li, Zhenyu Weng
Deep hashing has been widely adopted for large-scale image retrieval, with numerous strategies proposed to optimize hash function learning. Pairwise-based methods are effective in…
Mesa-Extrapolation: A Weave Position Encoding Method for Enhanced Extrapolation in LLMs
Xin Ma, Yang Liu, Jingjing Liu +1
Large language models (LLMs), although having revolutionized many fields, still suffer from the challenging extrapolation problem, where the inference ability of LLMs sharply decli…