Rahul Sajnani, Diego Romeres, Mingyuan Zhang, Kai Gao, Carl Qi, Baichuan Huang, Srinath Sridhar, Dylan Randle · 2026 · arXiv
DOI: 10.48550/arXiv.2610.28626 · open-access PDF
Motivados pela capacidade humana de integrar a experiência através da linguagem, percepção e ação, apresentamos o UniWALM, um modelo unificado de linguagem de ação mundial que aprende vídeo, ação e linguagem em uma única representação compartilhada, onde cada modalidade é tanto um sinal condicionador quanto um alvo de supervisão. Perguntamos: pode a geração conjunta de linguagem enriquecer essa representação compartilhada, melhorando a manipulação e apoiando o comportamento dependente da memória?
Machine translation (Meta M2M100); the English original follows.
Motivated by the human ability to integrate experience across language, perception, and action, we introduce UniWALM, a unified world-action-language model that learns video, action, and language in a single shared representation, where each modality is both a conditioning signal and a supervision target. We ask: can jointly generating language enrich this shared representation, improving manipulation and supporting memory-dependent behavior? Unlike prior approaches that attach a reasoning or vision-language model to a video-action backbone and run them as distinct stages, UniWALM unifies all
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=2610.28626, 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