MathConstraint generates scalable, automatically verifiable combinatorial problems where LLMs achieve 18.5-66.9% accuracy without tools but roughly double that with solver access.
hub Canonical reference
Arc prize 2024: Technical report
Canonical reference. 100% of citing Pith papers cite this work as background.
27
Pith papers citing it
3
external citations · Pith
Background
100% of classified citations
abstract
As of December 2024, the ARC-AGI benchmark is five years old and remains unbeaten. We believe it is currently the most important unsolved AI benchmark in the world because it seeks to measure generalization on novel tasks -- the essence of intelligence -- as opposed to skill at tasks that can be prepared for in advance. This year, we launched ARC Prize, a global competition to inspire new ideas and drive open progress towards AGI by reaching a target benchmark score of 85\%. As a result, the state-of-the-art score on the ARC-AGI private evaluation set increased from 33\% to 55.5\%, propelled by several frontier AGI reasoning techniques including deep learning-guided program synthesis and test-time training. In this paper, we survey top approaches, review new open-source implementations, discuss the limitations of the ARC-AGI-1 dataset, and share key insights gained from the competition.
hub tools
citation-role summary
background 5
citation-polarity summary
roles
background 5polarities
background 5representative citing papers
TAC is a bandit curriculum for multi-domain RLVR that prioritizes domains whose gradient updates align with and benefit other domains, yielding up to 2.8-point macro accuracy gains over learnability-only baselines on Qwen3-1.7B and Llama3.2-3B.
Introduces KINA benchmark with 899 items over 261 disciplines, formal (1-1/e) coverage guarantee and bonus-on-bar tournament theorem, plus evaluations of 42 models with top score 53.17%.
Formal Conjectures is a Lean 4 benchmark containing 2615 formalized problems with 1029 open conjectures, designed to evaluate automated mathematical reasoning and proof discovery.
Explorer-Definer and Reflective Orchestrator harnesses raise DeepSeek V3.2 from 15.5% to 67.25% pass@2 on ARC-AGI-1 public eval at $0.25–$0.62 per task without ARC-specific training.
A modality-driven search system with holistic trace judging for ARC-AGI-2 reaches 72.9% on the semi-private set and 76.1% on the public set, outperforming GPT-5.2 Pro and Gemini 3 Pro by 18.7 points while releasing full code.
ARTS improves automated scientific discovery by using reasoning LMs with test-time training to separate hypothesis merit from execution quality in tree search, achieving 15.3% relative gains on 22 MLGym and MLEBench tasks.
FPRM is a Transformer-based model using fixed-point convergence for adaptive halting in looped architectures, claimed effective on Sudoku, Maze, state-tracking, and ARC-AGI benchmarks.
Loop-OWM uses color-prototype slots, demonstration-conditioned task summaries, and looped transitions to model ARC rules as visual-symbolic state changes and outperforms baselines on ARC-1 and ARC-2.
LRMs show a large production-evaluation gap on the VAIR dataset with valid answers but invalid reasoning, driven by answer confirmation bias as evidenced by CoT analysis, linear probes, and causal patching.
Learns state-conditioned commitment depth in a 7B vision-language policy that jointly predicts actions and replan intervals, outperforming fixed-depth baselines and larger models on Sliding Puzzle and Sokoban while providing a theoretical dominance result.
Denoising Recursion Models train multi-step noise reversal in looped transformers and outperform the prior Tiny Recursion Model on ARC-AGI.
Introduces group matching score for better evaluation of compositional reasoning and Test-Time Matching (TTM) algorithm for unsupervised self-improvement in multimodal models, achieving SOTA gains including surpassing GPT-4.1 and estimated human performance.
LLMs achieve higher accuracy than humans on compositional imagery tasks previously argued to require pictorial representations, supporting emergent propositional mental imagery in AI.
ARC-AGI-2 adds a larger, more complex set of tasks to the original ARC-AGI benchmark to give finer-grained measurement of fluid intelligence in AI.
L-VARC is a LUPI framework that refines crowd-sourced language descriptions with an LLM and uses cross-attention to guide visual ARC models during training only, yielding SOTA results with a lightweight 18M-parameter network.
The authors propose creating data probes—synthetic sequences from defined random processes—to reveal how data properties drive LLM behavior across workflow stages.
A CPST-based taxonomy sorts autonomous systems into Confined Actors, Socially-Aware Interactors, and CPST-Integrated Agents to enable proportional governance from enhanced liability to qualified personhood.
Factorization Regret — the divergence between joint and naive Bayesian goal beliefs — tracks why reservoir networks plateau at naive accuracy, and a classifier-generator chain learns interactions from sparse reward to enable offline generalization.
HRM is a recurrent architecture with high-level planning and low-level execution modules that reaches near-perfect accuracy on complex Sudoku, maze navigation, and ARC benchmarks using 27M parameters and 1000 samples without pre-training or CoT supervision.
The paper unifies perspectives on Long CoT in reasoning LLMs by introducing a taxonomy, detailing characteristics of deep reasoning and reflection, and discussing emergence phenomena and future directions.
Humanity's Last Exam is a new 2,500-question benchmark at the frontier of human knowledge where state-of-the-art LLMs show low accuracy.
Open-ended intelligence is formalized as the compositional closure L(P,C) of primitives P under operators C, with next primitive prediction proposed as an objective to acquire reusable primitives and grammar for lifelong adaptation.
Technical report announcing Ling-2.6 and Ring-2.6 models with hybrid linear attention, evolutionary CoT, and KPop RL for efficient agentic intelligence at scale.
citing papers explorer
-
MathConstraint: Automated Generation of Verified Combinatorial Reasoning Instances for LLMs
MathConstraint generates scalable, automatically verifiable combinatorial problems where LLMs achieve 18.5-66.9% accuracy without tools but roughly double that with solver access.
-
Transferability for General Reasoning: An Automated Curriculum for Multi-Domain RLVR
TAC is a bandit curriculum for multi-domain RLVR that prioritizes domains whose gradient updates align with and benefit other domains, yielding up to 2.8-point macro accuracy gains over learnability-only baselines on Qwen3-1.7B and Llama3.2-3B.
-
Knowledge Index of Noah's Ark
Introduces KINA benchmark with 899 items over 261 disciplines, formal (1-1/e) coverage guarantee and bonus-on-bar tournament theorem, plus evaluations of 42 models with top score 53.17%.
-
Formal Conjectures: An Open and Evolving Benchmark for Verified Discovery in Mathematics
Formal Conjectures is a Lean 4 benchmark containing 2615 formalized problems with 1029 open conjectures, designed to evaluate automated mathematical reasoning and proof discovery.
-
Cost-Effective Agent Harnesses for Abstract Reasoning and Generalization on ARC-AGI-1
Explorer-Definer and Reflective Orchestrator harnesses raise DeepSeek V3.2 from 15.5% to 67.25% pass@2 on ARC-AGI-1 public eval at $0.25–$0.62 per task without ARC-specific training.
-
Modality-Driven Search with Holistic Trace Judging for ARC-AGI-2
A modality-driven search system with holistic trace judging for ARC-AGI-2 reaches 72.9% on the semi-private set and 76.1% on the public set, outperforming GPT-5.2 Pro and Gemini 3 Pro by 18.7 points while releasing full code.
-
Learning the ARTS of Search for Automated Discovery
ARTS improves automated scientific discovery by using reasoning LMs with test-time training to separate hypothesis merit from execution quality in tree search, achieving 15.3% relative gains on 22 MLGym and MLEBench tasks.
-
Fixed-Point Reasoners: Stable and Adaptive Deep Looped Transformers
FPRM is a Transformer-based model using fixed-point convergence for adaptive halting in looped architectures, claimed effective on Sudoku, Maze, state-tracking, and ARC-AGI benchmarks.
-
Slots, Transitions, Loops: Learning Composable World Models for ARC
Loop-OWM uses color-prototype slots, demonstration-conditioned task summaries, and looped transitions to model ARC rules as visual-symbolic state changes and outperforms baselines on ARC-1 and ARC-2.
-
An Enigma of Artificial Reason: Investigating the Production-Evaluation Gap in Large Reasoning Models
LRMs show a large production-evaluation gap on the VAIR dataset with valid answers but invalid reasoning, driven by answer confirmation bias as evidenced by CoT analysis, linear probes, and causal patching.
-
When to Re-Commit: Temporal Abstraction Discovery for Long-Horizon Vision-Language Reasoning
Learns state-conditioned commitment depth in a 7B vision-language policy that jointly predicts actions and replan intervals, outperforming fixed-depth baselines and larger models on Sliding Puzzle and Sokoban while providing a theoretical dominance result.
-
One Step Forward and K Steps Back: Better Reasoning with Denoising Recursion Models
Denoising Recursion Models train multi-step noise reversal in looped transformers and outperform the prior Tiny Recursion Model on ARC-AGI.
-
Test-Time Matching: Unlocking Compositional Reasoning in Multimodal Models
Introduces group matching score for better evaluation of compositional reasoning and Test-Time Matching (TTM) algorithm for unsupervised self-improvement in multimodal models, achieving SOTA gains including surpassing GPT-4.1 and estimated human performance.
-
Artificial Phantasia: Emergent Mental Imagery in Large Language Models
LLMs achieve higher accuracy than humans on compositional imagery tasks previously argued to require pictorial representations, supporting emergent propositional mental imagery in AI.
-
ARC-AGI-2: A New Challenge for Frontier AI Reasoning Systems
ARC-AGI-2 adds a larger, more complex set of tasks to the original ARC-AGI benchmark to give finer-grained measurement of fluid intelligence in AI.
-
Language-Guided Abstraction for Visual Reasoning
L-VARC is a LUPI framework that refines crowd-sourced language descriptions with an LLM and uses cross-attention to guide visual ARC models during training only, yielding SOTA results with a lightweight 18M-parameter network.
-
Position: Let's Develop Data Probes to Fundamentally Understand How Data Affects LLM Performance
The authors propose creating data probes—synthetic sequences from defined random processes—to reveal how data properties drive LLM behavior across workflow stages.
-
Beyond Tools and Persons: Who Are They? Classifying Robots and AI Agents for Proportional Governance
A CPST-based taxonomy sorts autonomous systems into Confined Actors, Socially-Aware Interactors, and CPST-Integrated Agents to enable proportional governance from enhanced liability to qualified personhood.
-
Factorization Regret mediates compositional generalization in latent space
Factorization Regret — the divergence between joint and naive Bayesian goal beliefs — tracks why reservoir networks plateau at naive accuracy, and a classifier-generator chain learns interactions from sparse reward to enable offline generalization.
-
Hierarchical Reasoning Model
HRM is a recurrent architecture with high-level planning and low-level execution modules that reaches near-perfect accuracy on complex Sudoku, maze navigation, and ARC benchmarks using 27M parameters and 1000 samples without pre-training or CoT supervision.
-
Towards Reasoning Era: A Survey of Long Chain-of-Thought for Reasoning Large Language Models
The paper unifies perspectives on Long CoT in reasoning LLMs by introducing a taxonomy, detailing characteristics of deep reasoning and reflection, and discussing emergence phenomena and future directions.
-
Humanity's Last Exam
Humanity's Last Exam is a new 2,500-question benchmark at the frontier of human knowledge where state-of-the-art LLMs show low accuracy.
-
A Compositional Framework for Open-ended Intelligence
Open-ended intelligence is formalized as the compositional closure L(P,C) of primitives P under operators C, with next primitive prediction proposed as an objective to acquire reusable primitives and grammar for lifelong adaptation.
-
Ling and Ring 2.6 Technical Report: Efficient and Instant Agentic Intelligence at Trillion-Parameter Scale
Technical report announcing Ling-2.6 and Ring-2.6 models with hybrid linear attention, evolutionary CoT, and KPop RL for efficient agentic intelligence at scale.
-
Customizing an LLM for Enterprise Software Engineering
Gemini for Google, customized via continued pre-training on proprietary Google engineering data, delivers measurable productivity gains in a large internal developer study.
-
Measuring AI Reasoning: A Guide for Researchers
Reasoning in language models should be measured by the faithfulness and validity of their multi-step search processes and intermediate traces, not final-answer accuracy.
-
OpenCompass: A Universal Evaluation Platform for Large Language Models
OpenCompass is presented as a one-stop, scalable, high-concurrency LLM evaluation platform with modular architecture supporting multiple domains and evaluator types.