List your company
ResearchFormerly WizWand

Mathematical Reasoning on AMC 23 (Avg@8)

80.6Avg@8Legislator-Executor (Ours)
-0.93620.23241.462.568Oct 10, 2025Nov 17, 2025Dec 25, 2025Feb 2, 2026Mar 12, 2026Apr 19, 2026May 28, 2026
Updated 5mo ago

Evaluation results

MethodLinks
Legislator-Executor (Ours)Model=Qwen3-14B-Base, Decoding Strategy=8 sampling iterations2026.0480.6
R1-Distill-Qwen-32B# Trainable Params=32B2025.1076.9
S1KModel=Qwen3-14B-Base, Decoding Strategy=8 sampling iterations2026.0475.9
DeepToolBackbone=Qwen3-4B, Training Stage=+RL2026.0575.3
Legislator-Executor (Ours)Model=Qwen3-8B-Base, Decoding Strategy=8 sampling iterations2026.0474.7
DeepToolBackbone=Qwen3-4B, Training Stage=SFT2026.0574.4
Legislator-Executor (Ours)Model=Qwen2.5-7B-QwQ, Decoding Strategy=8 sampling iterations2026.0470.6
LIMOModel=Qwen3-14B-Base, Decoding Strategy=8 sampling iterations2026.0470
s1.1-32B# Trainable Params=32B2025.1070
DeepToolBackbone=Qwen2.5-7B, Training Stage=SFT2026.0568.1
ToRLBackbone=Qwen2.5-7B, Training Stage=RL-enhanced tool reasoning2026.0567.5
BaseModel=Qwen3-14B-Base, Decoding Strategy=8 sampling iterations2026.0466.6
Legislator-Executor (Ours)Model=Qwen2.5-Math-7B, Decoding Strategy=8 sampling iterations2026.0465.3
Legislator-Executor (Ours)Model=Qwen3-4B-Base, Decoding Strategy=8 sampling iterations2026.0465
DeepToolBackbone=Qwen2.5-7B, Training Stage=+RL2026.0564
Target + THINKLOGIT-DPO# Trainable Params=78M2025.1063.7
ReToolBackbone=Qwen2.5-7B, Training Stage=RL-enhanced tool reasoning2026.0562.5
SimpleRL-Zoo-7BBackbone=Qwen2.5-7B, Training Stage=Zero-RL2026.0562.5
Target + THINKLOGIT# Trainable Params=02025.1062.2
LIMOModel=Qwen3-8B-Base, Decoding Strategy=8 sampling iterations2026.0460.3
S1KModel=Qwen3-8B-Base, Decoding Strategy=8 sampling iterations2026.0460.3
BaseModel=Qwen3-8B-Base, Decoding Strategy=8 sampling iterations2026.0458.4
LIMOModel=Qwen2.5-7B-QwQ, Decoding Strategy=8 sampling iterations2026.0457.5
Qwen2.5-32B# Trainable Params=-2025.1057.2
LIMOModel=Qwen3-4B-Base, Decoding Strategy=8 sampling iterations2026.0456.9
Legislator-Executor (Ours)Model=Qwen2.5-7B, Decoding Strategy=8 sampling iterations2026.0455.9
BaseModel=Qwen2.5-7B-QwQ, Decoding Strategy=8 sampling iterations2026.0455
S1KModel=Qwen2.5-7B-QwQ, Decoding Strategy=8 sampling iterations2026.0455
S1KModel=Qwen3-4B-Base, Decoding Strategy=8 sampling iterations2026.0453.1
R1-Distill-Qwen-1.5B# Trainable Params=-2025.1051.2
DeepToolBackbone=Qwen3-4B, Training Stage=w/o thinking2026.0550.9
BaseModel=Qwen3-4B-Base, Decoding Strategy=8 sampling iterations2026.0450.3
S1KModel=Qwen2.5-Math-7B, Decoding Strategy=8 sampling iterations2026.0449.7
Qwen3-4B-BaseBackbone=Qwen3-4B, Training Stage=Base2026.0548.1
LIMOModel=Qwen2.5-Math-7B, Decoding Strategy=8 sampling iterations2026.0447.2
LIMOModel=Qwen2.5-7B, Decoding Strategy=8 sampling iterations2026.0445.9
DeepToolBackbone=Qwen2.5-7B, Training Stage=w/o thinking2026.0545.6
Search-R1Backbone=Qwen2.5-7B, Training Stage=Search-based2026.0545
S1KModel=Qwen2.5-7B, Decoding Strategy=8 sampling iterations2026.0443.8
BaseModel=Qwen2.5-Math-7B, Decoding Strategy=8 sampling iterations2026.0438.8
Search-o1Backbone=Qwen2.5-7B, Training Stage=Search-based2026.0537.5
BaseModel=Qwen2.5-7B, Decoding Strategy=8 sampling iterations2026.0436.9
Legislator-Executor (Ours)Model=Gemma-2-9B, Decoding Strategy=8 sampling iterations2026.0423.8
Qwen2.5-7B-BaseBackbone=Qwen2.5-7B, Training Stage=Base2026.0521.7
LIMOModel=Gemma-2-9B, Decoding Strategy=8 sampling iterations2026.0420.6
S1KModel=Gemma-2-9B, Decoding Strategy=8 sampling iterations2026.0420.3
Legislator-Executor (Ours)Model=Llama-3.1-8B, Decoding Strategy=8 sampling iterations2026.0410.9
Qwen2.5-7B-TIRBackbone=Qwen2.5-7B, Training Stage=TIR2026.0510.8
LIMOModel=Llama-3.1-8B, Decoding Strategy=8 sampling iterations2026.048.8
S1KModel=Llama-3.1-8B, Decoding Strategy=8 sampling iterations2026.047.8
BaseModel=Gemma-2-9B, Decoding Strategy=8 sampling iterations2026.047.5
BaseModel=Llama-3.1-8B, Decoding Strategy=8 sampling iterations2026.044.7
BaseModel=Llama-3.2-3B, Decoding Strategy=8 sampling iterations2026.043.8
Legislator-Executor (Ours)Model=Llama-3.2-3B, Decoding Strategy=8 sampling iterations2026.043.8
S1KModel=Llama-3.2-3B, Decoding Strategy=8 sampling iterations2026.043.4
Legislator-Executor (Ours)Model=Mistral-7B-v0.3, Decoding Strategy=8 sampling iterations2026.043.1
S1KModel=Mistral-7B-v0.3, Decoding Strategy=8 sampling iterations2026.042.8
LIMOModel=Llama-3.2-3B, Decoding Strategy=8 sampling iterations2026.042.5
BaseModel=Mistral-7B-v0.3, Decoding Strategy=8 sampling iterations2026.042.5
LIMOModel=Mistral-7B-v0.3, Decoding Strategy=8 sampling iterations2026.042.2