{"arxiv_id":"2608.27384","incoming":{"papers":0,"sentences":0,"groups":[],"unlabelled_sentences":0},"outgoing":{"papers":16,"sentences":35,"groups":[{"kind":"extends","label":"Builds on","papers":3,"sentences":3,"quotes":[{"arxiv_id":"2505.13211","title":"MAGI-1: Autoregressive Video Generation at Scale","url":"/paper/2505.13211","section":"Introduction","sentence":"Inspired by streaming chunk-wise diffusion for long video [ref] , maintains a streaming buffer of action chunks at staggered noise levels and advances them jointly under chunk-wise causal attention.","intent_p":0.81,"contests_p":null},{"arxiv_id":"2407.01392","title":"Diffusion Forcing: Next-token Prediction Meets Full-Sequence Diffusion","url":"/paper/2407.01392","section":"Introduction","sentence":"Inspired by streaming chunk-wise diffusion for long video [ref] , maintains a streaming buffer of action chunks at staggered noise levels and advances them jointly under chunk-wise causal attention.","intent_p":0.81,"contests_p":null},{"arxiv_id":"2507.03745","title":null,"url":"/paper/2507.03745","section":"Introduction","sentence":"Inspired by streaming chunk-wise diffusion for long video [ref] , maintains a streaming buffer of action chunks at staggered noise levels and advances them jointly under chunk-wise causal attention.","intent_p":0.81,"contests_p":null}]},{"kind":"method_or_tool","label":"Uses the method or tool of","papers":2,"sentences":2,"quotes":[{"arxiv_id":"2601.18692","title":"A Pragmatic VLA Foundation Model","url":"/paper/2601.18692","section":"Setup","sentence":"Following LingBot-VLA [ref] , we group the RoboTwin 2.0 tasks into short-, medium-, and long-horizon subsets for horizon-wise analysis.","intent_p":0.95,"contests_p":null},{"arxiv_id":"1709.07871","title":"FiLM: Visual Reasoning with a General Conditioning Layer","url":"/paper/1709.07871","section":"Chunk-wise Autoregressive Formulation","sentence":"We apply FiLM [ref] to enhance the multi-level timestep conditioning in the action expert.","intent_p":0.99,"contests_p":null}]},{"kind":"comparison","label":"Compares with","papers":4,"sentences":4,"quotes":[{"arxiv_id":"2510.26742","title":null,"url":"/paper/2510.26742","section":"Setup","sentence":"For decoding latency we compare against Realtime-VLA [ref] , the strongest system-level acceleration baseline and an instance of the per-pass-cheaper family identified in [ref].","intent_p":0.99,"contests_p":null},{"arxiv_id":"2512.01031","title":"VLASH: Real-Time VLAs via Future-State-Aware Asynchronous Inference","url":"/paper/2512.01031","section":"Setup","sentence":"For asynchronous execution we compare against VLASH [ref] and StreamingVLA [ref] , which represent the two dominant strategies for mitigating prediction-execution mismatch (future-state conditioning and action flow matching).","intent_p":0.94,"contests_p":null},{"arxiv_id":"2603.28565","title":null,"url":"/paper/2603.28565","section":"Setup","sentence":"For asynchronous execution we compare against VLASH [ref] and StreamingVLA [ref] , which represent the two dominant strategies for mitigating prediction-execution mismatch (future-state conditioning and action flow matching).","intent_p":0.94,"contests_p":null},{"arxiv_id":"2603.19199","title":"FASTER: Rethinking Real-Time Flow VLAs","url":"/paper/2603.19199","section":"Setup","sentence":"For reaction speed we compare against FASTER [ref] .","intent_p":1.0,"contests_p":null}]},{"kind":"background","label":"Cites as background","papers":8,"sentences":9,"quotes":[{"arxiv_id":"2512.01031","title":"VLASH: Real-Time VLAs via Future-State-Aware Asynchronous Inference","url":"/paper/2512.01031","section":"Introduction","sentence":"Efficient-inference methods reduce per-step latency through lightweight architectures [ref] , compression [ref] , or token pruning [ref] , while asynchronous methods condition on future states or planned actions [ref] to suppress mismatch.","intent_p":0.93,"contests_p":null},{"arxiv_id":"2512.05964","title":null,"url":"/paper/2512.05964","section":"Introduction","sentence":"Efficient-inference methods reduce per-step latency through lightweight architectures [ref] , compression [ref] , or token pruning [ref] , while asynchronous methods condition on future states or planned actions [ref] to suppress mismatch.","intent_p":0.93,"contests_p":null},{"arxiv_id":"2602.03782","title":null,"url":"/paper/2602.03782","section":"Introduction","sentence":"Efficient-inference methods reduce per-step latency through lightweight architectures [ref] , compression [ref] , or token pruning [ref] , while asynchronous methods condition on future states or planned actions [ref] to suppress mismatch.","intent_p":0.93,"contests_p":null},{"arxiv_id":"2603.19199","title":"FASTER: Rethinking Real-Time Flow VLAs","url":"/paper/2603.19199","section":"Related Work","sentence":"Most efforts to accelerate VLA inference make each forward pass cheaper but leave the iterative decoding loop intact: lightweight backbones [ref] , weight and activation compression [ref] , visual-token pruning [ref] , and faster action heads [ref] .","intent_p":0.88,"contests_p":null},{"arxiv_id":"2603.28565","title":null,"url":"/paper/2603.28565","section":"Related Work","sentence":"Most efforts to accelerate VLA inference make each forward pass cheaper but leave the iterative decoding loop intact: lightweight backbones [ref] , weight and activation compression [ref] , visual-token pruning [ref] , and faster action heads [ref] .","intent_p":0.88,"contests_p":null}]}],"unlabelled_sentences":17},"claims":[],"thresholds":{"intent_p":0.8,"supports_p":0.95,"contests_p":0.7},"method":"Citing sentences come from each paper's LaTeX source; every \\cite marks the sentence and the works it cites. Pith's classifier typed each sentence. A label is shown at intent probability 0.8 or more (0.95 for support), a contest at 0.7 or more. Measured accuracy against two blind model families is at https://pith.science/accuracy. The quoted sentences are the citing authors' own.","held_up":null,"_links":{"html":"https://pith.science/paper/2608.27384"}}