List your company
ResearchFormerly WizWand

Math Reasoning on Gaokao En 2023

79AccuracyLegislator-Executor (Ours)
11.9229.33546.7564.165Aug 28, 2025Oct 12, 2025Nov 27, 2025Jan 11, 2026Feb 26, 2026Apr 12, 2026May 28, 2026
Updated 5mo ago

Evaluation results

MethodLinks
Legislator-Executor (Ours)Model=Qwen3-14B-Base, Decoding Strategy=Zero-shot greedy2026.0479
LIMOModel=Qwen3-14B-Base, Decoding Strategy=Zero-shot greedy2026.0476.9
S1KModel=Qwen3-14B-Base, Decoding Strategy=Zero-shot greedy2026.0476.1
Legislator-Executor (Ours)Model=Qwen3-8B-Base, Decoding Strategy=Zero-shot greedy2026.0475.8
NPG-Muse-8BPass@k protocol=avg@82025.0874.7
Full AttentionBase Model=DeepSeek-R1-Distill-Qwen-7B, Cache Correction=False2026.0274.2
LycheeDecodeBase Model=DeepSeek-R1-Distill-Qwen-7B, Cache Correction=False2026.0274.2
Legislator-Executor (Ours)Model=Qwen2.5-7B-QwQ, Decoding Strategy=Zero-shot greedy2026.0473.5
LycheeDecodeBase Model=DeepSeek-R1-Distill-Qwen-7B, Cache Correction=True2026.0272.7
S1KModel=Qwen3-8B-Base, Decoding Strategy=Zero-shot greedy2026.0471.7
DAPO + HistaModel Size=14B, RL Algorithm=DAPO2026.0571.1
Legislator-Executor (Ours)Model=Qwen2.5-Math-7B, Decoding Strategy=Zero-shot greedy2026.0470.4
LIMOModel=Qwen2.5-7B-QwQ, Decoding Strategy=Zero-shot greedy2026.0470.4
Legislator-Executor (Ours)Model=Qwen3-4B-Base, Decoding Strategy=Zero-shot greedy2026.0469.6
Full AttentionBase Model=DeepSeek-R1-Distill-Llama-8B, Cache Correction=False2026.0268.8
LycheeDecodeBase Model=DeepSeek-R1-Distill-Llama-8B, Cache Correction=False2026.0268.8
LycheeDecodeBase Model=DeepSeek-R1-Distill-Llama-8B, Cache Correction=True2026.0268.8
LIMOModel=Qwen3-8B-Base, Decoding Strategy=Zero-shot greedy2026.0468.6
S1KModel=Qwen3-4B-Base, Decoding Strategy=Zero-shot greedy2026.0468.6
DAPOModel Size=14B, RL Algorithm=DAPO2026.0568.5
Legislator-Executor (Ours)Model=Qwen2.5-7B, Decoding Strategy=Zero-shot greedy2026.0468.3
LIMOModel=Qwen3-4B-Base, Decoding Strategy=Zero-shot greedy2026.0468.1
BaseModel=Qwen3-14B-Base, Decoding Strategy=Zero-shot greedy2026.0467.5
BaseModel=Qwen2.5-7B-QwQ, Decoding Strategy=Zero-shot greedy2026.0466.8
Qwen2.5-14B-Instruct-1MPass@k protocol=avg@82025.0866
DAPO + HistaModel Size=7B, RL Algorithm=DAPO2026.0565.4
BaseModel=Qwen3-8B-Base, Decoding Strategy=Zero-shot greedy2026.0464.9
S1KModel=Qwen2.5-7B-QwQ, Decoding Strategy=Zero-shot greedy2026.0464.9
Qwen3-14B-BasePass@k protocol=avg@82025.0864.8
CSIPOModel Size=7B, RL Algorithm=CSIPO2026.0564.4
S1KModel=Qwen2.5-Math-7B, Decoding Strategy=Zero-shot greedy2026.0464.2
CSIPO + HistaModel Size=7B, RL Algorithm=CSIPO2026.0563.9
DAPOModel Size=7B, RL Algorithm=DAPO2026.0563.8
TidalDecodeBase Model=DeepSeek-R1-Distill-Qwen-7B, Cache Correction=True2026.0263.3
LIMOModel=Qwen2.5-Math-7B, Decoding Strategy=Zero-shot greedy2026.0463.1
S1KModel=Qwen2.5-7B, Decoding Strategy=Zero-shot greedy2026.0462.9
baseline-GroupAModel=Qwen2.5 (7B-Instruct)2026.0362.6
TidalDecodeBase Model=DeepSeek-R1-Distill-Llama-8B, Cache Correction=False2026.0262.5
ARE-GroupAModel=Qwen2.5 (7B-Instruct)2026.0361.8
LIMOModel=Qwen2.5-7B, Decoding Strategy=Zero-shot greedy2026.0461.8
ARE-GroupBModel=Qwen2.5 (7B-Instruct)2026.0361.6
baseline-GroupBModel=Qwen2.5 (7B-Instruct)2026.0361.3
NPG-Muse-7BPass@k protocol=avg@82025.0861.1
BaseModel=Qwen3-4B-Base, Decoding Strategy=Zero-shot greedy2026.0460.8
GRPOModel=qwen2.5-7b-instruct, Agent Configuration=Single-agent2026.0360.31
CCPOModel=qwen2.5-7b-instruct, Agent Configuration=Dual-Agent2026.0359.74
ReMAModel=qwen3-4b-base, Agent Configuration=Dual-Agent2026.0358.7
Qwen2.5-7B-Ins-1MPass@k protocol=avg@82025.0858.3
DAPOModel Size=3B, RL Algorithm=DAPO2026.0558.2
TidalDecodeBase Model=DeepSeek-R1-Distill-Qwen-7B, Cache Correction=False2026.0257.8
ReMAModel=qwen2.5-7b-instruct, Agent Configuration=Dual-Agent2026.0357.7
UntrainedModel=qwen2.5-7b-instruct, Agent Configuration=Dual-Agent2026.0357.4
CCPOModel=qwen3-4b-base, Agent Configuration=Dual-Agent2026.0357.4
TidalDecodeBase Model=DeepSeek-R1-Distill-Llama-8B, Cache Correction=True2026.0257
Qwen3-8B-BasePass@k protocol=avg@82025.0856.9
BaseModel=Qwen2.5-7B, Decoding Strategy=Zero-shot greedy2026.0456.1
BaseModel=Qwen2.5-Math-7B, Decoding Strategy=Zero-shot greedy2026.0456.1
CSIPO + HistaModel Size=3B, RL Algorithm=CSIPO2026.0555.8
DAPO + HistaModel Size=3B, RL Algorithm=DAPO2026.0555.6
CSIPOModel Size=3B, RL Algorithm=CSIPO2026.0552.7
DAPO + HistaModel Size=1.5B, RL Algorithm=DAPO2026.0551.2
GRPOModel=qwen3-4b-base, Agent Configuration=Single-agent2026.0351.06
GSPO + HistaModel Size=1.5B, RL Algorithm=GSPO2026.0549.8
DAPOModel Size=1.5B, RL Algorithm=DAPO2026.0549.4
GSPOModel Size=1.5B, RL Algorithm=GSPO2026.0549.3
CSIPOModel Size=1.5B, RL Algorithm=CSIPO2026.0548.3
CSIPO + HistaModel Size=1.5B, RL Algorithm=CSIPO2026.0548.1
ARE-GroupAModel=Qwen2.5 (1.5B-Instruct)2026.0346
baseline-GroupAModel=Qwen2.5 (1.5B-Instruct)2026.0345.5
baseline-GroupCModel=Qwen2.5 (1.5B-Instruct)2026.0345.5
ARE-GroupCModel=Qwen2.5 (1.5B-Instruct)2026.0345.5
ReMAModel=qwen2.5-1.5b-instruct, Agent Configuration=Dual-Agent2026.0343.9
Legislator-Executor (Ours)Model=Gemma-2-9B, Decoding Strategy=Zero-shot greedy2026.0443.4
S1KModel=Gemma-2-9B, Decoding Strategy=Zero-shot greedy2026.0443.1
GRPOModel=qwen2.5-1.5b-instruct, Agent Configuration=Single-agent2026.0342.82
ARE-GroupBModel=Llama3.1 (8B-Instruct)2026.0342.6
UntrainedModel=qwen2.5-1.5b-instruct, Agent Configuration=Dual-Agent2026.0342.08
UntrainedModel=qwen2.5-7b-instruct, Agent Configuration=Single-agent2026.0342.04
Base ModelModel=Qwen2.5 (7B-Instruct)2026.0342.04
CCPOModel=qwen2.5-1.5b-instruct, Agent Configuration=Dual-Agent2026.0341.6
ARE-GroupCModel=Llama3.1 (8B-Instruct)2026.0340.3
baseline-GroupBModel=Llama3.1 (8B-Instruct)2026.0339.7
CCPOModel=llama3.1-8b-instruct, Agent Configuration=Dual-Agent2026.0339.48
baseline-GroupCModel=Llama3.1 (8B-Instruct)2026.0339.2
ReMAModel=llama3.1-8b-instruct, Agent Configuration=Dual-Agent2026.0338.7
LIMOModel=Gemma-2-9B, Decoding Strategy=Zero-shot greedy2026.0438.4
UntrainedModel=llama3.1-8b-instruct, Agent Configuration=Dual-Agent2026.0336.1
Legislator-Executor (Ours)Model=Llama-3.1-8B, Decoding Strategy=Zero-shot greedy2026.0430.9
DAPO + HistaModel Size=0.5B, RL Algorithm=DAPO2026.0530.6
S1KModel=Llama-3.1-8B, Decoding Strategy=Zero-shot greedy2026.0428.8
DAPOModel Size=0.5B, RL Algorithm=DAPO2026.0528
LIMOModel=Llama-3.1-8B, Decoding Strategy=Zero-shot greedy2026.0424.9
BaseModel=Gemma-2-9B, Decoding Strategy=Zero-shot greedy2026.0424.7
UntrainedModel=qwen2.5-1.5b-instruct, Agent Configuration=Single-agent2026.0323.5
Base ModelModel=Qwen2.5 (1.5B-Instruct)2026.0323.5
UntrainedModel=qwen3-4b-base, Agent Configuration=Dual-Agent2026.0322.08
Legislator-Executor (Ours)Model=Llama-3.2-3B, Decoding Strategy=Zero-shot greedy2026.0417.1
Legislator-Executor (Ours)Model=Mistral-7B-v0.3, Decoding Strategy=Zero-shot greedy2026.0415.6
BaseModel=Llama-3.1-8B, Decoding Strategy=Zero-shot greedy2026.0415.1
LIMOModel=Mistral-7B-v0.3, Decoding Strategy=Zero-shot greedy2026.0414.5