{"paper":{"title":"GIFT: Guided Intermediate Feature Training via Action-Oriented Structural Supervision for Robotic Manipulation","arxiv_id":"2609.04193","primary_cat":"cs.RO","submitted_at":"2026-09-03T17:59:03+00:00","version_compiled":"v1"},"notice":"Every quoted sentence is the authors' text from the arXiv listing abstract. Labels and probabilities are the outputs of classifier-1.13.0 on the question bank abstract-p1-20260923, exactly as returned. No sentence in this record was written by a language model.","schema":"pith.compiled-record/v0","status":"compiled","abstract":{"source":"arXiv listing abstract v1","sentences":[{"i":0,"p":0.84,"id":"s0","own":false,"role":"background","text":"Vision-language pre-training and predictive world modeling provide robot policies with rich semantic and dynamic visual features, but their native action and visual-prediction objectives may omit critical physical and task structure while retaining control-irrelevant visual redundancy.","hedge":{"p":0.79,"label":"hedged_possibility"},"own_p":0.09,"main_claim_p":0.03,"quantitative":false,"quantitative_p":0.05,"role_probabilities":{"gap":0.16,"method":0.0,"result":0.0,"background":0.84,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":1,"p":0.85,"id":"s1","own":false,"role":"gap","text":"We call this mismatch between visual richness and control utility the action-sufficiency gap.","hedge":{"p":0.97,"label":"asserted"},"own_p":0.43,"main_claim_p":0.03,"quantitative":false,"quantitative_p":0.04,"role_probabilities":{"gap":0.85,"method":0.0,"result":0.0,"background":0.12,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.03}},{"i":2,"p":0.85,"id":"s2","own":false,"role":"contribution","text":"We investigate whether this gap can be bridged by guiding intermediate features to preserve three control-relevant structure in robotic manipulation: geometry governing motion feasibility, affordance encoding instruction-relevant entities, and goals grounding instructions in task-relevant regions.","hedge":{"p":0.76,"label":"hedged_possibility"},"own_p":0.22,"main_claim_p":0.06,"quantitative":false,"quantitative_p":0.39,"role_probabilities":{"gap":0.06,"method":0.05,"result":0.0,"background":0.04,"limitation":0.0,"future_work":0.0,"contribution":0.85,"interpretation":0.0}},{"i":3,"p":0.63,"id":"s3","own":true,"role":"method","text":"To this end, we present GIFT (Guided Intermediate Feature Training), an architecture-flexible framework for learning intermediate features that translates these structures into training-time constraints through geometry alignment, affordance prediction, and goal-region reconstruction.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.89,"main_claim_p":0.06,"quantitative":false,"quantitative_p":0.03,"role_probabilities":{"gap":0.0,"method":0.63,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.37,"interpretation":0.0}},{"i":4,"p":0.83,"id":"s4","own":true,"role":"method","text":"We instantiate GIFT in a Vision-Language-Action (VLA) policy, a direct-action World-Action Model (WAM), and an inverse-dynamics WAM while retaining each model's action formulation.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.92,"main_claim_p":0.03,"quantitative":false,"quantitative_p":0.04,"role_probabilities":{"gap":0.0,"method":0.83,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.17,"interpretation":0.0}},{"i":5,"p":1.0,"id":"s5","own":true,"role":"result","text":"Under zero-shot transfer to LIBERO-Plus, GIFT-VLA, GIFT-WAM-Fast, and GIFT-WAM-IDM outperform StarVLA-OFT, Fast-WAM, and Fast-WAM-IDM by 4.6, 12.6, and 5.2 points, reaching 79.6%, 72.6%, and 87.8%, respectively.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.96,"main_claim_p":0.11,"quantitative":true,"quantitative_p":0.99,"role_probabilities":{"gap":0.0,"method":0.0,"result":1.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":6,"p":1.0,"id":"s6","own":true,"role":"result","text":"On RoboCasa, the three GIFT variants reach 61.4%, 83.6%, and 82.3%, outperforming their counterparts by 12.6, 9.0, and 8.4 points, respectively.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.96,"main_claim_p":0.06,"quantitative":true,"quantitative_p":0.99,"role_probabilities":{"gap":0.0,"method":0.0,"result":1.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":7,"p":0.99,"id":"s7","own":true,"role":"interpretation","text":"Together, these results establish learning functionally structured intermediate features as a reusable principle across model-specific action formulations, with especially large gains on articulated-object tasks and high-precision real-world manipulation under unseen visual and spatial perturbations.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.94,"main_claim_p":0.91,"quantitative":false,"quantitative_p":0.19,"role_probabilities":{"gap":0.0,"method":0.0,"result":0.01,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.99}},{"i":8,"p":0.49,"id":"s8","own":false,"role":"future_work","text":"Project page: https://openphoenix-team.github.io/GIFT-pages.","hedge":{"p":0.98,"label":"unclear"},"own_p":0.13,"main_claim_p":0.02,"quantitative":false,"quantitative_p":0.02,"role_probabilities":{"gap":0.0,"method":0.02,"result":0.01,"background":0.12,"limitation":0.0,"future_work":0.49,"contribution":0.34,"interpretation":0.02}}]},"compiled":{"main_claim":{"p":1.0,"text":"Together, these results establish learning functionally structured intermediate features as a reusable principle across model-specific action formulations, with especially large gains on articulated-object tasks and high-precision real-world manipulation under unseen visual and spatial perturbations.","location":"abstract s7","hedge_strength":{"p":1.0,"label":"asserted"}},"provenance":{"route":{"via":"TypeSafe API","model":"classifier-1.13.0","endpoint":"https://api.typesafe.ai/v1/systemone","provider":"TypeSafe"},"packing":"one request per paper","requests":1,"input_tokens":7359,"cost_usd_reported":0.0},"compiled_at":"2026-09-23","bank_version":"abstract-p1-20260923","model_version":"classifier-1.13.0","abstract_states":{"claims_first":0.06,"states_limitation":0.13,"human_participants":0.04,"controlled_comparison":0.96,"randomized_assignment":0.01,"states_quantitative_result":0.99,"compares_with_prior_results":0.98,"mentions_code_or_data_availability":0.31},"evidential_basis":{"p":0.93,"value":"observation or experiment","probabilities":{"argument":0.0,"literature":0.0,"proof_or_derivation":0.0,"analysis_of_existing_data":0.0,"observation_or_experiment":0.93,"simulation_or_computation":0.07}},"contribution_type":{"p":1.0,"value":"method or tool","probabilities":{"review":0.0,"method_or_tool":1.0,"empirical_study":0.0,"simulation_study":0.0,"theoretical_result":0.0,"dataset_or_benchmark":0.0,"position_or_commentary":0.0}},"contribution_statement":{"p":1.0,"text":"To this end, we present GIFT (Guided Intermediate Feature Training), an architecture-flexible framework for learning intermediate features that translates these structures into training-time constraints through geometry alignment, affordance prediction, and goal-region reconstruction.","location":"abstract s3"}},"coverage":{"tier":"abstract","levels_run":["P abstract"],"not_examined":["full text","figures and tables","references and citing sentences","relations to other papers"],"sources_read":["arXiv listing abstract v1"],"levels_not_run":["S sections","T body sentences","C citations","R cross-paper relations"],"sentences_examined":9,"sentences_in_abstract":9},"measured_accuracy":{"url":"https://pith.science/accuracy","measured_on":"2026-10-02","method":"two independent model families labelled a fresh random sample blind; the figure is how often the published label matches their consensus","sentence_role_agreement":0.92,"main_claim_agreement":0.908,"typed_citation_agreement":0.933,"typed_citation_agreement_95_interval":[0.917,0.945],"review_comments_accurate":0.758,"review_comments_accurate_95_interval":[0.684,0.82],"review_caught_admitted_error_found_or_related":0.448,"review_comments_judged":{"judged_by":["gpt-5.6-luna","grok-4.7"],"judged_against":"Pith's own text copy of each paper: text extracted from the PDF, the first 90,000 characters","measured_on":"2026-10-04","accurate":113,"decided":149,"sampled_reviews":200,"split_by_lane":false,"by_lane":"https://pith.science/accuracy#by-lane"}}}