Outcome-Guided On-Policy Distillation is proposed, which applies trajectory-relative weighting to teacher supervision and calibrates these weights using final task outcomes from paired student continuations and selectively strengthens supervision on the student's original trajectories at turns where teacher guidance be...
Tong Zhang, Zhou Liu, Yi-Hao Liu et al.· 0 citations
This work introduces Visual Skill Cards, a state-conditioned memory representation that binds reusable procedures with applicability cues, visual evidence, and verification signals and supports CardDistill, which uses VSC evidence as privileged teacher context to train a student that acts without runtime card retrieval...
Zhou Liu, Ligang Huang, Zeli Su et al.· 2 citations
SemanticAlign-Bench (SA-Bench), a diagnostic benchmark covering 30 papers from ICLR, ICML and NeurIPS 2025, is introduced and indicates that scaffolds optimized for executability provide limited leverage for scientific reproduction; narrowing the gap requires scaffolds that prioritize semantic specification verificatio...
Xueyan Hu, Ze-Wei Pan, Zeli Su et al.· 0 citations
ReproAgent is introduced, a four-stage Prepare--Plan--Generate--Repair pipeline built around a persistent implementation contract with two channels: an implementation-requirement channel that turns paper snippets into code obligations, and a reference-evidence channel that retrieves content and structure evidence from...
Xueyan Hu, Ze-Wei Pan, Zhongyuan Wang et al.· 0 citations
ExRole is introduced, a trajectory-to-role framework that learns future-aware role prototypes from prefix-local team traces, resolves them into readable instructions and token-aligned role markers, and optionally routes shared LoRA rank slots with turn-aligned credit.
Zhou Liu, Chaoyang Han, Ze-Wei Pan et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
