LAWA keeps the benefits of test-time future imagination for world action models while replacing expensive future-observation generation with compact latent intentions, yielding an effective trade-off among performance, generalization, and latency.</p>\n","updatedAt":"2026-08-26T09:45:51.393Z","author":{"_id":"669bd845d77a8b28703ad05f","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/vtvxX5pw4OH3-7jEXvpdj.png","fullname":"Xiang Li","name":"FortyTwoo","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.8832771182060242},"editors":["FortyTwoo"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/vtvxX5pw4OH3-7jEXvpdj.png"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2608.24882","authors":[{"_id":"6a8eb59a5673886739f198b8","name":"Xiang Li","hidden":false},{"_id":"6a8eb59a5673886739f198b9","name":"Yupeng Zheng","hidden":false},{"_id":"6a8eb59a5673886739f198ba","name":"Songen Gu","hidden":false},{"_id":"6a8eb59a5673886739f198bb","name":"Huailiang Ma","hidden":false},{"_id":"6a8eb59a5673886739f198bc","name":"Feng Yu","hidden":false},{"_id":"6a8eb59a5673886739f198bd","name":"Xian Nie","hidden":false},{"_id":"6a8eb59a5673886739f198be","name":"Shanshuai Yuan","hidden":false},{"_id":"6a8eb59a5673886739f198bf","name":"Yujie Zang","hidden":false},{"_id":"6a8eb59a5673886739f198c0","name":"Weize Li","hidden":false},{"_id":"6a8eb59a5673886739f198c1","name":"Shuai Tian","hidden":false},{"_id":"6a8eb59a5673886739f198c2","name":"Moyang Liu","hidden":false},{"_id":"6a8eb59a5673886739f198c3","name":"Ya-Qin Zhang","hidden":false},{"_id":"6a8eb59a5673886739f198c4","name":"Wenchao Ding","hidden":false}],"publishedAt":"2026-08-25T00:00:00.000Z","submittedOnDailyAt":"2026-08-26T00:00:00.000Z","title":"Latent Action as Intention Enables Efficient Future Imagination for World Action Models","submittedOnDailyBy":{"_id":"669bd845d77a8b28703ad05f","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/vtvxX5pw4OH3-7jEXvpdj.png","isPro":false,"fullname":"Xiang Li","user":"FortyTwoo","type":"user","name":"FortyTwoo"},"summary":"World action models (WAMs) improve robot control by modeling how observations evolve, but generating future observations at test time incurs substantial latency. Fast-WAM removes this process for efficiency; however, our matched implementations show lower generalization for Fast-WAM than for future-aware alternatives, especially with scarce robot demonstrations and in out-of-distribution scenarios. To bridge this gap, we introduce **LAWA**, a WAM architecture that uses compact latent actions as an operational representation of future intentions, enabling efficient test-time future imagination without generating future observations. Specifically, a discrete tokenizer enhanced by action-free pre-training produces manipulation-centric codebook targets. LAWA jointly denoises a continuous latent state anchored to these targets with executable action chunks while omitting the future-video branch at inference. On RoboCasa, LAWA achieves state-of-the-art average success rates of 65.6% and 80.8% in the few-shot and full data settings, improving over the matched Fast-WAM baseline by 9.6 and 4.5 points, respectively. It also preserves the performance level of the matched Joint-WAM variant while requiring 42.9% lower inference latency. LAWA also demonstrates competitive zero-shot robustness on LIBERO-Plus and superior performance on real-world tasks. These results show that future imagination need not be discarded: retaining it with compact latent actions yields an effective trade-off among performance, generalization, and latency. Code and models will be released.","upvotes":0,"discussionId":"6a8eb59a5673886739f198c5","projectPage":"https://getterupper.github.io/LAWA","ai_summary":"LAWA improves robot control by using compact latent actions to retain efficient future imagination without generating observations, achieving strong performance with lower latency.","ai_keywords":["world action models","latent actions","discrete tokenizer","action-free pre-training","codebook","latent state","denoising","action chunks","future-video branch","inference latency"],"ai_summary_model":"thinkingmachines/Inkling-Small"},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[],"acceptLanguages":["en"],"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2608/2608.24882.md","query":{}}">
Latent Action as Intention Enables Efficient Future Imagination for World Action Models
Abstract
LAWA improves robot control by using compact latent actions to retain efficient future imagination without generating observations, achieving strong performance with lower latency.
World action models (WAMs) improve robot control by modeling how observations evolve, but generating future observations at test time incurs substantial latency. Fast-WAM removes this process for efficiency; however, our matched implementations show lower generalization for Fast-WAM than for future-aware alternatives, especially with scarce robot demonstrations and in out-of-distribution scenarios. To bridge this gap, we introduce **LAWA**, a WAM architecture that uses compact latent actions as an operational representation of future intentions, enabling efficient test-time future imagination without generating future observations. Specifically, a discrete tokenizer enhanced by action-free pre-training produces manipulation-centric codebook targets. LAWA jointly denoises a continuous latent state anchored to these targets with executable action chunks while omitting the future-video branch at inference. On RoboCasa, LAWA achieves state-of-the-art average success rates of 65.6% and 80.8% in the few-shot and full data settings, improving over the matched Fast-WAM baseline by 9.6 and 4.5 points, respectively. It also preserves the performance level of the matched Joint-WAM variant while requiring 42.9% lower inference latency. LAWA also demonstrates competitive zero-shot robustness on LIBERO-Plus and superior performance on real-world tasks. These results show that future imagination need not be discarded: retaining it with compact latent actions yields an effective trade-off among performance, generalization, and latency. Code and models will be released.
Community
LAWA keeps the benefits of test-time future imagination for world action models while replacing expensive future-observation generation with compact latent intentions, yielding an effective trade-off among performance, generalization, and latency.
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Cite arxiv.org/abs/2608.24882 in a model README.md to link it from this page.
Cite arxiv.org/abs/2608.24882 in a dataset README.md to link it from this page.
Cite arxiv.org/abs/2608.24882 in a Space README.md to link it from this page.
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.