arxiv Moonshot AI (arXiv) · 838d ago

Deeply Understanding the Problems method achieves 97.1% on GSM8K

from English

Researchers propose Deeply Understanding the Problems (DUP), a method designed to improve Large Language Models' ability to solve math word problems by addressing semantic misunderstanding errors that often limit Chain-of-Thought reasoning.

DUP encourages models to deeply understand problems and extract key information for better reasoning.
The method consistently outperforms counterparts across 10 diverse reasoning benchmarks.
DUP achieves a new state-of-the-art accuracy of 97.1% on the GSM8K benchmark in a zero-shot setting.

This approach addresses a major factor limiting LLM reasoning performance by focusing on semantic understanding rather than just calculation or step-missing errors.

Importance 1/3 Moonshot AI (arXiv) Reasoning models Research paper

Benchmarks

Benchmark	Model	Score
GSM8K	DUP method	97.1%

Read original

media Don't Worry About the Vase · 5d ago · 5 views

OpenAI's Astra solves 10 major open mathematics problems

OpenAI has released results from its unreleased internal model, Astra, which successfully solved ten significant open problems in mathematics. The solutions were generated by the model and subsequently formalized into Lean certificates by human researchers.

blog Simon Willison · 7d ago · 12 views

OpenAI's Astra model solves ten long-standing math problems using GPT-5.6 Sol

OpenAI has released solutions to ten mathematical and theoretical computer science problems that had seen no progress on their main results for at least a decade, using an internal version of its next major model, Astra.

arxiv arXiv cs.AI · 17d ago · 1 view

PoTRE introduces heterogeneous multi-agent framework for test-time reasoning

The authors introduce PoTRE (Poly-Topological Reasoning Ensembles), a framework designed to address the limitations of standard single-stream prompting in complex reasoning tasks. PoTRE decouples inference into four distinct agents: an Adversarial Refinement Agent, a Hierarchical strategic Planning Agent, a Spectrum Search Agent, and a Direct Chain Agent.

arxiv arXiv cs.AI · 24d ago

Deep Interaction enables precise human correction of LLM reasoning errors

The authors propose Deep Interaction, an efficient human-AI interaction method designed to correct reasoning errors in large language models without requiring full response regeneration. This mechanism allows users to directly edit the original Chain-of-Thought response, preserving accurate steps while fixing mistakes.

arxiv arXiv cs.CL · 25d ago

Ring-Zero scales zero RL to 1T parameters for emergent reasoning

Researchers present Ring-Zero, a stable training pipeline that scales reinforcement learning with verifiable rewards to models with one trillion parameters, addressing issues like token redundancy and poor readability found in naive scaling.