List your company

Program Synthesis on CodeContests (test)

0.2045Pass@1GRPO-RLVR
0.068260.103630.1390.17437Mar 17, 2026

Evaluation results

MethodLinks
GRPO-RLVRBackbone=QWEN2.5-72B2026.030.20450.3697
LEAFEBackbone=QWEN2.5-72B2026.030.17120.4788
LEAFEBackbone=LLAMA3-70B2026.030.14090.3394
GRPO-RLVRBackbone=LLAMA3-70B2026.030.13640.2788
BASE (NO FT)Backbone=QWEN2.5-72B2026.030.10.3394
BASE (NO FT)Backbone=LLAMA3-70B2026.030.07350.2485