List your company
ResearchFormerly WizWand

Logical Reasoning on LogiQA (Accuracy)

80.4AccuracyQwen3-8B-thinking
19.45635.27851.166.922Dec 4, 2025Jan 5, 2026Feb 6, 2026Mar 11, 2026Apr 12, 2026May 14, 2026Jun 16, 2026
Updated 4mo ago

Evaluation results

MethodLinks
Qwen3-8B-thinkingModel Parameters=8B, Reasoning Mode=Thinking2026.0180.4
DenserModel=GPT-4o2025.1279.1
Process SupervisionModel=GPT-4o2025.1278.9
Reflection-CoTModel=GPT-4o2025.1278.4
Self-VerificationModel=GPT-4o2025.1277.8
Tree-of-ThoughtModel=GPT-4o2025.1277.7
Process SupervisionModel=DeepSeek-V32025.1277.4
Self-ConsistencyModel=GPT-4o2025.1277.1
Reflection-CoTModel=DeepSeek-V32025.1276.9
DenserModel=DeepSeek-V32025.1276.9
Think-to-ThinkModel=GPT-4o2025.1276.4
GPT-5.12026.0176.34
DenserModel=o1-mini2025.1275.8
Tree-of-ThoughtModel=DeepSeek-V32025.1275.6
Self-VerificationModel=DeepSeek-V32025.1275.6
Process SupervisionModel=o1-mini2025.1275.3
Self-ConsistencyModel=DeepSeek-V32025.1275
Reflection-CoTModel=o1-mini2025.1274.8
Chain-of-ThoughtModel=GPT-4o2025.1274.6
Self-VerificationModel=o1-mini2025.1274.3
Think-to-ThinkModel=DeepSeek-V32025.1274.3
Tree-of-ThoughtModel=o1-mini2025.1273.4
Qwen3-8BModel Parameters=8B2026.0173.2
Self-ConsistencyModel=o1-mini2025.1272.8
Chain-of-ThoughtModel=DeepSeek-V32025.1272.5
Think-to-ThinkModel=o1-mini2025.1271.9
Chain-of-ThoughtModel=o1-mini2025.1270.2
SGR-LLaMA-3.3-70BBackbone=LLaMA-3.3-70B2026.0169.91
AceMADBase Model=Qwen3-235B-A22B-Instruct, Iteration Rounds (T)=2, Number of Agents (N)=52026.0359.69
AceMADBase Model=Qwen3-235B-A22B-Instruct, Iteration Rounds (T)=3, Number of Agents (N)=52026.0359.69
Sparse MADBase Model=Qwen3-235B-A22B-Instruct, Number of Agents (N)=52026.0347.81
Decentralized MADBase Model=Qwen3-235B-A22B-Instruct, Number of Agents (N)=52026.0345.62
Centralized MADBase Model=Qwen3-235B-A22B-Instruct, Number of Agents (N)=52026.0345.62
Majority VotingBase Model=Qwen3-235B-A22B-Instruct, Number of Agents (N)=52026.0343.12
Single AgentBase Model=Qwen3-235B-A22B-Instruct, Number of Agents (N)=52026.0342.5
MMKEBackbone=Qwen2.5-7B Instruct, Consolidation=false2025.1241
MMKE+Con.Backbone=Qwen2.5-7B Instruct, Consolidation=true2025.1239.2
PreBackbone=Qwen2.5-7B Instruct2025.1238.7
EtConBackbone=Qwen2.5-7B Instruct2025.1238.4
FT-M+Con.Backbone=Qwen2.5-7B Instruct, Consolidation=true2025.1237.8
AceMADBase Model=GPT-4o-mini, Iteration Rounds (T)=3, Number of Agents (N)=52026.0337.74
FT-MBackbone=Qwen2.5-7B Instruct, Consolidation=false2025.1237
ConSA (head-wise, single-layer)Target Sparsity (ρ)=0.50, Granularity=head-wise, Constraint Scope=single-layer, Model=1.7B2026.0636.92
Qwen2.5-7BModel Type=Dense, Active Parameters=7B, Total Parameters=7B2026.0236.4
ConSA (head-wise, all-layers)Target Sparsity (ρ)=0.50, Granularity=head-wise, Constraint Scope=all-layers, Model=1.7B2026.0635.08
Rule (head-wise)Target Sparsity (ρ)=0.50, Granularity=head-wise, Model=1.7B2026.0634.62
Dense FATarget Sparsity (ρ)=0, Model=1.7B2026.0634.31
Qwen2.5-3BModel Type=Dense, Active Parameters=3B, Total Parameters=3B2026.0233.5
ConSA (layer-wise)Target Sparsity (ρ)=0.50, Granularity=layer-wise, Model=1.7B2026.0632.92
Qwen2.5-1.5BModel Type=Dense, Active Parameters=1.5B, Total Parameters=1.5B2026.0231.9
Rule (layer-wise)Target Sparsity (ρ)=0.50, Granularity=layer-wise, Model=1.7B2026.0631.71
Qwen2.5-1.5B#Params=1.5B/1.5B, #Tokens=18T2026.0231.5
Qwen3-1.7B#Params=1.7B/1.7B, #Tokens=36T2026.0231.5
LLaMA-MoE-v2-3.5BModel Type=MoE, Active Parameters=3.5B2026.0230.7
LLaMA-MoE-3.0B#Params=3.0B/7B, #Tokens=2.2T2026.0230.6
Llama-3.2-3BModel Type=Dense, Active Parameters=3B, Total Parameters=3B2026.0230.6
SPES-9B#Params=3.1B/9B, #Tokens=400B2026.0230.4
SmolLM2-1.7BModel Type=Dense, Active Parameters=1.7B, Total Parameters=1.7B2026.0230.1
SmolLM2-1.7B#Params=1.7B/1.7B, #Tokens=11T2026.0229.8
LLaMA-MoE-v1-3.5BModel Type=MoE, Active Parameters=3.5B2026.0229.7
Qwen2.5-0.5B#Params=0.5B/0.5B, #Tokens=18T2026.0229.5
OLMoE-1B-7BModel Type=MoE, Active Parameters=1B, Total Parameters=7B2026.0229.3
Qwen3-0.6B#Params=0.6B/0.6B, #Tokens=36T2026.0229
ExpertWeaver-E64-A14-S2(Qwen2.5-7B)Model Type=MoE, Active Parameters=3.5B, Total Parameters=7B, Training Budget=200B tokens2026.0229
OLMoE-1B-7B*Model Type=MoE, Active Parameters=1B, Total Parameters=7B, Training Budget=500B tokens2026.0228.7
ExpertWeaver-E64-A14-S2(OLMo-7B)Model Type=MoE, Active Parameters=1B, Total Parameters=7B, Training Budget=200B tokens2026.0228.5
AceMADBase Model=Qwen3-235B-A22B-Instruct, Iteration Rounds (T)=5, Number of Agents (N)=52026.0328.44
OLMoE-1B-7B#Params=1.3B/7B, #Tokens=5T2026.0228.4
MoE++ 7B#Params=1.2B/7B, #Tokens=1T2026.0228.4
Sheared-LLaMA-2.7BModel Type=Dense, Active Parameters=2.7B, Total Parameters=2.7B2026.0228.3
Pythia-2.8BModel Type=Dense, Active Parameters=2.8B, Total Parameters=2.8B2026.0228.1
Open-LLaMA-3B-v2Model Type=Dense, Active Parameters=3B, Total Parameters=3B2026.0228.1
Pythia-2.8B#Params=2.8B/2.8B, #Tokens=300B2026.0228
SPES-7B#Params=1.6B/7B, #Tokens=500B2026.0227.5
OLMo-7BModel Type=Dense, Active Parameters=7B, Total Parameters=7B2026.0227.5
INCITE-Base-3BModel Type=Dense, Active Parameters=3B, Total Parameters=3B2026.0227.5
Pythia-1.4B#Params=1.4B/1.4B, #Tokens=300B2026.0227.3
SPES-2B#Params=0.8B/2.1B, #Tokens=500B2026.0227.2
OPT-1.3B#Params=1.3B/1.3B, #Tokens=180B2026.0226.9
TinyLlama-1.1B#Params=1.1B/1.1B, #Tokens=3T2026.0226.3
OPT-2.7B#Params=2.7B/2.7B, #Tokens=180B2026.0226
OPT-2.7BModel Type=Dense, Active Parameters=2.7B, Total Parameters=2.7B2026.0225.8
SOLARTraining Stage=SFT-1B2025.1223.81
OpT-DeUSTraining Stage=SFT-1B2025.1223.81
Avg-DeUSTraining Stage=SFT-1B2025.1223.66
OpT-DeUSTraining Stage=CPT-1B2025.1223.35
LESATraining Stage=SFT-1B2025.1223.35
MIDUS-HMLTraining Stage=CPT-1B2025.1223.2
Llama ProTraining Stage=SFT-1B2025.1222.89
Avg-DeUSTraining Stage=CPT-1B2025.1222.73
Gemma-2-2bModel Type=Dense, Active Parameters=2B, Total Parameters=2B2026.0222.7
SOLARTraining Stage=CPT-1B2025.1222.58
Llama ProTraining Stage=CPT-1B2025.1222.43
BaseTraining Stage=CPT-1B2025.1222.27
Decentralized MADBase Model=GPT-4o-mini, Number of Agents (N)=52026.0322.19
Centralized MADBase Model=GPT-4o-mini, Number of Agents (N)=52026.0322.19
Sparse MADBase Model=GPT-4o-mini, Number of Agents (N)=52026.0322.19
LESATraining Stage=CPT-1B2025.1221.97
Majority VotingBase Model=GPT-4o-mini, Number of Agents (N)=52026.0321.88
ALPHABackbone=Qwen2.5-7B Instruct, Consolidation=false2025.1221.8