…Existing visual RL post-training trains on static curated datasets, with fixed image-question-answer samples bounded by their collection budget. In this work, we introduce TRON (Targeted, Rule-verifiable Online eNvironments…
…Furthermore, we show that a top-k approximation of the baseline further lowers cost without compromising performance. Across mathematical and scientific reasoning benchmarks, vOPD consistently outperforms vanilla OPD and matches the most…
…TrOPD performs OPD only in regions where the teacher provides reliable supervision, mitigating the optimization difficulty of the K1 reverse-KL estimator under distribution mismatch . 2) Outlier Estimation: For outlier regions, we…
…Xin Li , , , , Abstract On-policy distillation with reward extrapolation exhibits a safety threshold beyond which structured output tasks lose format preservation, with empirical validation showing performance parity at reduced parameter count when…
…Yunho Choi , Jongwon Lim , , , , Abstract POISE enables stable and efficient policy optimization for large reasoning models by estimating baselines using internal model signals, reducing computational overhead while maintaining performance comparable to existing…
…Generated by Qwen/Qwen2.5-Coder-32B-Instruct Role-playing language agents (RPLAs) should play characters whose values and behavior evolve as the story progresses, not maintain a fixed persona. Existing benchmarks…
…Experiments across multiple model scales and architectures show that SEIF consistently improves instruction-following performance, suggesting strong generality. Further analyses reveal the sources of improvement and identify an effective training strategy for…
…However, these outputs are inherently flattened, entangling foreground elements, background, and text within a fixed canvas. As a result, flexible post-generation editing remains challenging, revealing a clear last-mile gap toward…
…Empirical Failure Modes and Simple Fixes (2026) Self-Induced Outcome Potential: Turn-Level Credit Assignment for Agents without Verifiers (2026) Skip-Connected Policy Optimization for Implicit Advantage (2026) Demystifying OPD: Length Inflation…
…Existing methods mainly curate memory with predefined KV-cache schedules, fixed-ratio heuristic compression, or inference-time RoPE adaptation. These designs inevitably lose historical information and amplify compounding errors due to their…