Hongcan Guo, Yian Zhao, Feng Wang, Shen Nie, Guoqiang Wei, Yan Zeng, Hengshuang Zhao · 2026 · arXiv
DOI: 10.48550/arXiv.2610.31305 · open-access PDF
A geração hierárquica de linguagem separa a modelagem semântica contínua da realização de texto discreto, mas sua eficácia depende do que o espaço latente comprime e de como o gerador discreto o usa. propomos Cola2, um quadro que estuda conjuntamente a compressão semântica contínua, o equilíbrio entre reconstrução e geração e a co-adaptação contínua-discreta. Cola2 segmenta dinamicamente o texto em espaços semânticos, controla a taxa de compressão através de feedback negativo auxiliar-livre de perdas e combina a reconstrução e supervisão de geração para aprender a codificação compacta, preditiva.
Machine translation (Meta M2M100); the English original follows.
Hierarchical language generation separates continuous semantic modeling from discrete text realization, but its effectiveness depends on what the latent space compresses and how the discrete generator uses it. We propose Cola2, a framework that jointly studies continuous semantic compression, the balance between reconstruction and generation, and continuous-discrete co-adaptation. Cola2 dynamically segments text into semantic spans, controls the compression rate through auxiliary-loss-free negative feedback, and combines reconstruction and generation supervision to learn compact, predictive la
Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese
GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.31305, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txtSources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange