This work takes the other half of the classical transform-coding toolkit instead: keep every token and vary its rate, and preserves the model's output distribution and its answers better than uniform-rate coding, the closed-form water-fill and distortion-ranked pruning.
Hong-Bo Zhang, Zi-Hao Yang, Liu-Yang Song et al.· 0 citations
Global vision--language similarities compress an image and a caption into one vector, preserving semantics but not which word corresponds to which region or how those regions are arranged; a model can recognize every word and object yet prefer a compositionally incorrect caption. We argue that a frozen encoder retains...
Liu-Yang Song, Yi Zhang, Zhong-Yi Deng et al.· arXiv.org· 0 citations
An agent that remembers what it did on a web page must decide when two pages count as the same. Memories built on observation similarity merge pages that look alike but behave differently, and GUIs are full of such pages: two tabs of one widget or two rows of one menu answer the same click differently. We define the me...
Hong-Bo Zhang, Liu-Yang Song, Quan-Quan Li et al.· 0 citations
Memory Adaptation for Task-Conditioned Execution (MATE) is introduced, a deterministic post-retrieval procedure that converts trajectories into execution-oriented memory and supports memory adaptation as a distinct stage between retrieval and embodied execution.
Quan-Quan Li, Hong-Bo Zhang, Yi-He Chi et al.· arXiv.org· 1 citation
FutureBridge is presented, which ranks joint LLM-SLM token candidates according to how well they support the SLM's subsequent reasoning, and indicates that token selection benefits from modeling whether the receiving SLM can use each candidate to continue reasoning, rather than relying on the LLM's local preference alo...
Quanquan Li, Hongbo Zhang, Yihe Chi et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.