Research

Paper

TESTING March 17, 2026

Execution-Grounded Credit Assignment for GRPO in Code Generation

Authors

Abhijit Kumar, Natalya Kumar, Shikhar Gupta

Abstract

Critic-free reinforcement learning with verifiable rewards (RLVR) improves code generation by optimizing unit-test pass rates, but GRPO-style updates suffer from coarse credit assignment: a single outcome signal is spread uniformly across long programs even when failure stems from a localized semantic error. We propose Execution-Grounded Credit Assignment (EGCA), which localizes GRPO updates using execution traces. For programs that satisfy algorithmic constraints but fail tests, EGCA executes the candidate and a canonical reference solution (curated once offline; used for analysis, not supervision) under identical instrumentation, identifies the earliest semantic divergence, and assigns advantage only to the corresponding token span while masking downstream tokens. EGCA is a drop-in modification requiring no critic, auxiliary loss, or learned verifier, yielding 82.1% pass@1 on HumanEval (+3.1 over GRPO) and 68.9% on MBPP (+1.5) with 18% wall-clock overhead.

Metadata

arXiv ID: 2603.16158
Provider: ARXIV
Primary Category: cs.LG
Published: 2026-03-17
Fetched: 2026-03-18 06:02

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2603.16158v1</id>\n    <title>Execution-Grounded Credit Assignment for GRPO in Code Generation</title>\n    <updated>2026-03-17T06:22:44Z</updated>\n    <link href='https://arxiv.org/abs/2603.16158v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2603.16158v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>Critic-free reinforcement learning with verifiable rewards (RLVR) improves code generation by optimizing unit-test pass rates, but GRPO-style updates suffer from coarse credit assignment: a single outcome signal is spread uniformly across long programs even when failure stems from a localized semantic error. We propose Execution-Grounded Credit Assignment (EGCA), which localizes GRPO updates using execution traces. For programs that satisfy algorithmic constraints but fail tests, EGCA executes the candidate and a canonical reference solution (curated once offline; used for analysis, not supervision) under identical instrumentation, identifies the earliest semantic divergence, and assigns advantage only to the corresponding token span while masking downstream tokens. EGCA is a drop-in modification requiring no critic, auxiliary loss, or learned verifier, yielding 82.1% pass@1 on HumanEval (+3.1 over GRPO) and 68.9% on MBPP (+1.5) with 18% wall-clock overhead.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.LG'/>\n    <published>2026-03-17T06:22:44Z</published>\n    <arxiv:comment>Accepted at SPOT ICLR 2026 (https://openreview.net/forum?id=nqkVB5EVXJ)</arxiv:comment>\n    <arxiv:primary_category term='cs.LG'/>\n    <author>\n      <name>Abhijit Kumar</name>\n    </author>\n    <author>\n      <name>Natalya Kumar</name>\n    </author>\n    <author>\n      <name>Shikhar Gupta</name>\n    </author>\n  </entry>"
}