Execution-Grounded Credit Assignment for GRPO in Code Generation
This addresses a bottleneck in code generation for developers by improving efficiency and accuracy in reinforcement learning methods, though it is incremental as it builds on GRPO-style updates.
The paper tackles the problem of coarse credit assignment in critic-free reinforcement learning for code generation by proposing Execution-Grounded Credit Assignment (EGCA), which localizes updates using execution traces to assign advantage to token spans where semantic errors occur, resulting in 82.1% pass@1 on HumanEval (+3.1 over GRPO) and 68.9% on MBPP (+1.5).
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.