Share your thoughts, 1 month free Claude Pro on usSee more

Logical Reasoning on LogiQA (Accuracy)

80.4Accuracy

Qwen3-8B-thinking

Updated 1mo ago

Evaluation Results

Method	Links
Qwen3-8B-thinking 2026.01		80.4
Denser 2025.12		79.1
Process Supervision 2025.12		78.9
Reflection-CoT 2025.12		78.4
Self-Verification 2025.12		77.8
Tree-of-Thought 2025.12		77.7
Process Supervision 2025.12		77.4
Self-Consistency 2025.12		77.1
Reflection-CoT 2025.12		76.9
Denser 2025.12		76.9
Think-to-Think 2025.12		76.4
GPT-5.1 2026.01		76.34
Denser 2025.12		75.8
Tree-of-Thought 2025.12		75.6
Self-Verification 2025.12		75.6
Process Supervision 2025.12		75.3
Self-Consistency 2025.12		75
Reflection-CoT 2025.12		74.8
Chain-of-Thought 2025.12		74.6
Self-Verification 2025.12		74.3
Think-to-Think 2025.12		74.3
Tree-of-Thought 2025.12		73.4
Qwen3-8B 2026.01		73.2
Self-Consistency 2025.12		72.8
Chain-of-Thought 2025.12		72.5
Think-to-Think 2025.12		71.9
Chain-of-Thought 2025.12		70.2
SGR-LLaMA-3.3-70B 2026.01		69.91
AceMAD 2026.03		59.69
AceMAD 2026.03		59.69
Sparse MAD 2026.03		47.81
Decentralized MAD 2026.03		45.62
Centralized MAD 2026.03		45.62
Majority Voting 2026.03		43.12
Single Agent 2026.03		42.5
MMKE 2025.12		41
MMKE+Con. 2025.12		39.2
Pre 2025.12		38.7
EtCon 2025.12		38.4
FT-M+Con. 2025.12		37.8
AceMAD 2026.03		37.74
FT-M 2025.12		37
ConSA (head-wise, single-layer) 2026.06		36.92
Qwen2.5-7B 2026.02		36.4
ConSA (head-wise, all-layers) 2026.06		35.08
Rule (head-wise) 2026.06		34.62
Dense FA 2026.06		34.31
Qwen2.5-3B 2026.02		33.5
ConSA (layer-wise) 2026.06		32.92
Qwen2.5-1.5B 2026.02		31.9
Rule (layer-wise) 2026.06		31.71
Qwen2.5-1.5B 2026.02		31.5
Qwen3-1.7B 2026.02		31.5
LLaMA-MoE-v2-3.5B 2026.02		30.7
LLaMA-MoE-3.0B 2026.02		30.6
Llama-3.2-3B 2026.02		30.6
SPES-9B 2026.02		30.4
SmolLM2-1.7B 2026.02		30.1
SmolLM2-1.7B 2026.02		29.8
LLaMA-MoE-v1-3.5B 2026.02		29.7
Qwen2.5-0.5B 2026.02		29.5
OLMoE-1B-7B 2026.02		29.3
Qwen3-0.6B 2026.02		29
ExpertWeaver-E64-A14-S2(Qwen2.5-7B) 2026.02		29
OLMoE-1B-7B* 2026.02		28.7
ExpertWeaver-E64-A14-S2(OLMo-7B) 2026.02		28.5
AceMAD 2026.03		28.44
OLMoE-1B-7B 2026.02		28.4
MoE++ 7B 2026.02		28.4
Sheared-LLaMA-2.7B 2026.02		28.3
Pythia-2.8B 2026.02		28.1
Open-LLaMA-3B-v2 2026.02		28.1
Pythia-2.8B 2026.02		28
SPES-7B 2026.02		27.5
OLMo-7B 2026.02		27.5
INCITE-Base-3B 2026.02		27.5
Pythia-1.4B 2026.02		27.3
SPES-2B 2026.02		27.2
OPT-1.3B 2026.02		26.9
TinyLlama-1.1B 2026.02		26.3
OPT-2.7B 2026.02		26
OPT-2.7B 2026.02		25.8
SOLAR 2025.12		23.81
OpT-DeUS 2025.12		23.81
Avg-DeUS 2025.12		23.66
OpT-DeUS 2025.12		23.35
LESA 2025.12		23.35
MIDUS-HML 2025.12		23.2
Llama Pro 2025.12		22.89
Avg-DeUS 2025.12		22.73
Gemma-2-2b 2026.02		22.7
SOLAR 2025.12		22.58
Llama Pro 2025.12		22.43
Base 2025.12		22.27
Decentralized MAD 2026.03		22.19
Centralized MAD 2026.03		22.19
Sparse MAD 2026.03		22.19
LESA 2025.12		21.97
Majority Voting 2026.03		21.88
ALPHA 2025.12		21.8

Showing 100 of 106 rows