List your company

Question Answering, Reasoning, and Coding on ExpBench Generic

84ARC-C AccuracyExpWeaver
61.577667.398873.2279.0412May 31, 2026

Evaluation results

MethodLinks
ExpWeaverBackbone=Qwen2.5-3B-Instruct, Training=LoRA (rank 32), Algorithm=GRPO2026.05848235688089.788565.33727878.25
Search-R1Method Category=LLM-Centric Experience Learning Baseline2026.0580.4476.6723.3367.7873.3385.1177.5659.7871.7967.573.27
Search-o1Method Category=LLM-Centric Experience Learning Baseline2026.05807228.3363.7871.5666.8963.1130.895457.0263.1
HRPOMethod Category=General Reasoning-Based Baseline2026.0579.3375.332067.117288.2280.8963.1174.3670.8974.04
R1Method Category=General Reasoning-Based Baseline2026.0578.4474.6721.6766.2271.3387.1179.7861.7873.0868.7573.1
ExpeLMethod Category=Retrieval-Centric Experience Learning Baseline2026.0576.4466.4426.6767.5469.5661.3358.89366067.5460.94
ReasoningBankMethod Category=Retrieval-Centric Experience Learning Baseline2026.0574.8963.333060.4467.566863.1132.895252.6360.57
LLM-RMethod Category=Retrieval-Centric Experience Learning Baseline2026.0574.8967.5616.6761.3371.7873.7865.1135.336467.5463.52
IRCoTMethod Category=LLM-Centric Experience Learning Baseline2026.0573.5664.2233.3360.2268.8965.1163.1128.675255.2659.82
CoTMethod Category=General Reasoning-Based Baseline2026.0562.4456.4426.6758.2264.2271.1166.4453.116672.8161.4