Groupwise Reward Geometry Conditioning (GRGC) is proposed, a two-stage framework that improves advantage construction by shaping student-side reward groups during both critic training and policy optimization, and introduces negligible overhead over GAD.
Abstract
Black-box distillation is a practical route for transferring capabilities from API-accessible large language models that expose only text outputs into smaller student models. Recent on-policy adversarial methods such as GAD improve over SeqKD by forming an adversarial loop between a critic and a student, where the critic provides rewards for GRPO-based student policy optimization over the student's sampled responses. However, GRPO computes advantages from the within-group relative rewards of student samples for the same prompt, whereas the critic is trained primarily to distinguish teacher responses from student responses. This objective mismatch can produce reward groups with collapsed scale or fragile margins, leading to brittle grouped optimization signals. We propose Groupwise Reward Geometry Conditioning (GRGC), a two-stage framework that improves advantage construction by shaping student-side reward groups during both critic training and policy optimization. To improve critic-side conditioning, Gaussian groupwise Optimal Transport calibration regularizes the critic during training to produce reward groups with non-collapsed spread and smooth rank-wise gaps by matching sorted prompt-wise rewards to group-centered Gaussian quantiles. Building on this conditioned reward geometry, policy-side group power modulation reshapes the prompt-wise reward groups before they are converted into advantages, preserving the critic-induced ordering while increasing optimization-relevant margin separability. Extensive experiments across diverse teachers, student model families and scales, and training datasets demonstrate the effectiveness of GRGC on both in-distribution and out-of-distribution evaluations, while introducing negligible overhead over GAD. The code is available at https://github.com/2018cx/GRGC.
Persistent-negative adversarial distillation consistently improves performance over current methods at matched discriminator compute and yields smoother fresh-policy discriminator trajectories, with fewer below-chance dips.
Hai-Xu Ma, S. Lahrichi, Wei-Wei Li et al.· 0 citations
DiffGate is introduced, an outcome-gated objective that combines GRPO with selective, bounded teacher guidance, and improves pass@8 across all four model--domain settings, demonstrating improved solution coverage under the evaluation protocol.
Karn Tiwari, V. Chordia, P. PrathoshA· 0 citations
Reward-Aligned Reweighting for On-Policy Distillation (R$^{2}$-OPD), which uses outcome agreement and the magnitude of teacher--student disagreement to continuously reallocate teacher supervision, achieves the highest average accuracy among the compared training methods in both cross-size and same-size distillation.
Hao Xu, Junwei Su, Lan-Song Diao et al.· arXiv.org· 0 citations
The cross-signal NTK is introduced, a token-level statistic that measures the alignment between reward and distillation gradients at position n and an empirical threshold beyond which naive mixing can lead to persistent training collapse is revealed.
Xin-Ke Jiang, Tao Feng, Zhi-Bang Yang et al.· arXiv.org· 1 citation
Experiments spanning mathematical reasoning, multi-domain STEM, code generation, and multi-turn agentic tasks show that RISE outperforms RLVR-only training and on-policy self-distillation across all settings.
Yang Li, Semih Yavuz, Shafiq Joty· 5 citations· ⚡1
Reinforcement learning can substantially improve a reasoning teacher, but it is unclear which of those improvements survive when the teacher supervises a smaller on-policy student. We study this question in mathematical reasoning by comparing teacher lineages before and after GRPO, multiple student scales, direct GRPO,...
Xiao-Yu Chen, Bo Shao, Tian-Gang Zhu et al.· 0 citations
Related blog posts
Microsoft Research Blog· microsoft.comNov 16, 2026
AI coding agents are increasingly help developers fix bugs and build features largely on their own, but every interaction with their models consumes tokens. In each session, an agent sends instructions, tool descriptions, and session history to its model. Most efforts to reduce this cost focus on individual tasks. Yet developers return to the same […] The post TokenEconomizer: Learning from Experience to Make AI Coding Agents More Efficient appeared first on Microsoft Research.
Professor Paris Smaragdis develops computers with superhuman hearing for applications like speech recognition, music understanding, and environmental monitoring.
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.