mlsanityData to 14 Oct 2026

Information-Gain Rewards over Diversity-Pruned Tests: GT-Anchored Verifier Co-Training for Reliable Code Generation

arXiv 2609.21208 ↗First posted 18 Sep 2026cs.AIcited by 0

Not on the map. Why?

Abstract (arXiv)

Self-play methods that co-train a single language model as both coder and test author promise to move code-generation RL beyond fixed test suites, but they suffer from two coupled pathologies: permissiveness collapse, where pass-rate rewards are maximised by trivial, non-discriminative tests, and concentration bias, where i.i.d. sampled tests cluster on modal inputs and inflate estimator variance. We introduce CoVer (Co-trained Coder and Verifier), a single-policy GRPO framework that addresses both failure modes. First, an information-gain (IG) reward scores each self-generated test by the mutual information between its pass/fail vector and a graded, ground-truth-anchored correctness signal y [0, 1] m, gated by the sign of their covariance so that only positively discriminative tests receive reward. Second, a three-stage diversity-aware selection step prunes a candidate pool to a behaviourally non-redundant suite (invalidity, input-string, execution-profile filtering), raising the effective sample size of the IG estimator at fixed execution budget. On five benchmarks (LiveBench, MBPP, LiveCodeBench, CodeContests, Code-Forces), CoVer raises one-shot pass@1 by +5.8 points at 7B and +7.1 points at 14B over the Qwen2.5-Instruct backbone, and achieves the highest macro-average among all compared methods at both scales. As a drop-in backbone inside the CodeT ranking pipeline, CoVer-7B adds +3.5 points, demonstrating the dual benefit of co-training for both generation and selection.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 15 of its 23 references that are in the map. First posted on arXiv 18 Sep 2026.

Builds on

15 of 23 references.

References · 15

  1. 2017Proximal Policy Optimization Algorithmscited by 15,215
  2. 2023Efficient Memory Management for Large Language Model Serving with PagedAttentioncited by 5,944
  3. 2021Program Synthesis with Large Language Modelscited by 3,637
  4. 2025DAPO: An Open-Source LLM Reinforcement Learning System at Scalecited by 2,810
  5. 2024LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Codecited by 2,004
  6. 2022CodeT: Code Generation with Generated Testscited by 384
  7. 2024LiveBench: A Challenging, Contamination-Limited LLM Benchmarkcited by 262
  8. 2024Meta-Rewarding Language Models: Self-Improving Alignment with LLM-as-a-Meta-Judgecited by 204
  9. 2024RLEF: Grounding Code LLMs in Execution Feedback with Reinforcement Learningcited by 151
  10. 2024Code Generation with AlphaCodium: From Prompt Engineering to Flow Engineeringcited by 103
  11. 2025S*: Test Time Scaling for Code Generationcited by 98
  12. 2023Enhancing Large Language Models in Coding Through Multi-Perspective Self-Consistencycited by 46
3 more
  1. 2025SPC: Evolving Self-Play Critic via Adversarial Games for LLM Reasoningcited by 42
  2. 2025DiverseGRPO: Mitigating Mode Collapse in Image Generation via Diversity-Aware GRPOcited by 17
  3. 2025Propose, Solve, Verify: Self-Play Through Formal Verificationcited by 13

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026CodeScaler: Scaling Code LLM Training and Test-Time Inference via Reward Models
  2. 2026CVeDRL: An Efficient Code Verifier via Difficulty-aware Reinforcement Learning
  3. 2025Klear-CodeTest: Scalable Test Case Generation for Code Reinforcement Learning
  4. 2026Code-A1: Adversarial Evolving of Code LLM and Test LLM via Reinforcement Learning
  5. 2025Dynamic Scaling of Unit Tests for Code Reward Modeling
  6. 2026Improving Small Language Models for Code Generation with Reinforcement Learning from Verification Feedback
  7. 2025Rethinking Verification for LLM Code Generation: From Generation to Testing
  8. 2026Robust Code RL via Faulty-Code-Driven Test case Synthesis and Dense Reward Shaping