{"arxiv_id":"2608.29937","incoming":{"papers":0,"sentences":0,"groups":[],"unlabelled_sentences":0},"outgoing":{"papers":18,"sentences":25,"groups":[{"kind":"extends","label":"Builds on","papers":1,"sentences":1,"quotes":[{"arxiv_id":"2606.15768","title":"LaWAM: Latent World Action Models for Efficient Dynamics-Aware Robot Policies","url":"/paper/2606.15768","section":"Related Work","sentence":"LaWAM showed that a latent visual subgoal can preserve the control benefit of future prediction without reconstructing pixels [ref] . builds on this result by exposing the policy/world boundary in the implementation and evaluating a compact policy configuration.","intent_p":0.91,"contests_p":null}]},{"kind":"method_or_tool","label":"Uses the method or tool of","papers":2,"sentences":2,"quotes":[{"arxiv_id":"2508.10104","title":"DINOv3","url":"/paper/2508.10104","section":"Stage 1: Learning an Action-Bearing Latent World Model","sentence":"For each transition pair $(o_t,o_t+Δ)$, a frozen DINOv3 encoder [ref] produces patch tokens $h_t=f_ψ(o_t)$ and $h_t+Δ=f_ψ(o_t+Δ)$, with $h∈ℝ^K×d_v$.","intent_p":1.0,"contests_p":null},{"arxiv_id":"2511.21631","title":"Qwen3-VL Technical Report","url":"/paper/2511.21631","section":"Stage 2: Predicting Latent Actions from Language and Vision","sentence":"The reported closed-loop models use Qwen-family backbones [ref] ; the more general VLM/VAM label denotes an interface target, not a completed cross-family benchmark.","intent_p":0.88,"contests_p":null}]},{"kind":"background","label":"Cites as background","papers":3,"sentences":4,"quotes":[{"arxiv_id":"2307.15818","title":"RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control","url":"/paper/2307.15818","section":"Introduction","sentence":"Vision–language–action (VLA) models inherit strong semantic representations from multimodal pretraining [ref] , while world–action models add an explicit prediction of how the scene should change [ref] .","intent_p":0.93,"contests_p":null},{"arxiv_id":"2506.21539","title":"WorldVLA: Towards Autoregressive Action World Model","url":"/paper/2506.21539","section":"Introduction","sentence":"Vision–language–action (VLA) models inherit strong semantic representations from multimodal pretraining [ref] , while world–action models add an explicit prediction of how the scene should change [ref] .","intent_p":0.93,"contests_p":null},{"arxiv_id":"2310.08864","title":"Open X-Embodiment: Robotic Learning Datasets and RT-X Models","url":"/paper/2310.08864","section":"Related Work","sentence":"OpenVLA and Octo subsequently provided open, broadly adaptable policy backbones trained on heterogeneous robot data [ref] .","intent_p":0.98,"contests_p":null}]}],"unlabelled_sentences":18},"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/2608.29937"}}