AI Paper Atlas

Research map / Benchmarks / LogiQA

LogiQA leaderboard

13 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on LogiQA. Best accuracy: 81 (TypedThinker + SC @5, ICLR 2025).

Numbers are as each paper reports them, read by AI from its results tables and not checked by hand. Papers can differ in setting (training data, model size, extra data); the topic column shows what each paper is about. Open the paper before you compare or quote a number.

LogiQA · accuracy

13 papers, each with its best reported number for its own method; higher is better.

#accuracy ↑MethodPaperTopic
181TypedThinker + SC @5
GPT-4o
TypedThinker: Diversify Large Language Model Reasoning with Typed Thinking
ICLR 2025 · Table: GPT-4o
Agents, code and math
263.2+LogicReward
Qwen3-8B
LogicReward: Incentivizing LLM Reasoning via Step-Wise Logical Supervision
ICLR 2026 · Table 1
Reasoning and chain of thought
361.14DKT-large
LLaMA 3-70B
Collaborative Enhancement of Large and Small Models for Question Answering via Dual Knowledge Transfer
AAAI 2026 · Table 2
Visual question answering
459.5CP-Router
Qwen-2.5-14B + R1-Distill-Qwen-14B
CP-Router: An Uncertainty-Aware Router Between LLM and LRM
AAAI 2026 · Table 1
Reasoning and chain of thought
545.8FOCAL REASONER
DeBERTa-xlarge
Fact-Driven Logical Reasoning for Machine Reading Comprehension
AAAI 2024 · Table 2
Retrieval and knowledge
642.8ZeroTuning
Llama-3.1-8B-Instruct
ZeroTuning: Unlocking the Initial Token's Power to Enhance Large Language Models Without Training
ICLR 2026 · Table 2
Language, safety and interpretability
742.24DAGNs + LReasoner
RoBERTa-Large
Discourse-Aware Graph Networks for Textual Logical Reasoning
TPAMI 2023 · Table 4
Knowledge and temporal graphs
840.55LogiGAN largeLogiGAN: Learning Logical Reasoning via Adversarial Pre-training
NeurIPS 2022 · Table 1
Language, safety and interpretability
932.35DiffuMamba-H
1.3B
DiffuMamba: High-Throughput Diffusion LMs with Mamba Backbone
ICML 2026 · Table 3
Attention and transformers
1028.3MoE++ 7B/(16+4)EMoE++: Accelerating Mixture-of-Experts Methods with Zero-Computation Experts
ICLR 2025 · Table 1
Mixture of experts
3 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

Not affiliated with any conference, journal or the benchmark's authors. Corrections: contact@latenticlabs.com.