Pith. sign in

REVIEW 11 cited by

MathScale: Scaling Instruction Tuning for Mathematical Reasoning

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2403.02884 v1 pith:HH6CAGFU submitted 2024-03-05 cs.CL cs.AIcs.LG

MathScale: Scaling Instruction Tuning for Mathematical Reasoning

classification cs.CL cs.AIcs.LG
keywords mathmathematicalreasoningllmsmathscaleproblemsaccuracyaverage
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Large language models (LLMs) have demonstrated remarkable capabilities in problem-solving. However, their proficiency in solving mathematical problems remains inadequate. We propose MathScale, a simple and scalable method to create high-quality mathematical reasoning data using frontier LLMs (e.g., {\tt GPT-3.5}). Inspired by the cognitive mechanism in human mathematical learning, it first extracts topics and knowledge points from seed math questions and then build a concept graph, which is subsequently used to generate new math questions. MathScale exhibits effective scalability along the size axis of the math dataset that we generate. As a result, we create a mathematical reasoning dataset (MathScaleQA) containing two million math question-answer pairs. To evaluate mathematical reasoning abilities of LLMs comprehensively, we construct {\sc MwpBench}, a benchmark of Math Word Problems, which is a collection of ten datasets (including GSM8K and MATH) covering K-12, college, and competition level math problems. We apply MathScaleQA to fine-tune open-source LLMs (e.g., LLaMA-2 and Mistral), resulting in significantly improved capabilities in mathematical reasoning. Evaluated on {\sc MwpBench}, MathScale-7B achieves state-of-the-art performance across all datasets, surpassing its best peers of equivalent size by 42.9\% in micro average accuracy and 43.7\% in macro average accuracy, respectively.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 11 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Beyond Accuracy: Diagnosing Algebraic Reasoning Failures in LLMs Across Nine Complexity Dimensions

    cs.CL 2026-04 unverdicted novelty 8.0

    A nine-dimension algebraic complexity framework shows that LLMs suffer a scale-invariant working memory bottleneck, collapsing at 20-30 parallel branches regardless of parameter count from 8B to 235B.

  2. InvEvolve: Evolving White-Box Inventory Policies via Large Language Models with Performance Guarantees

    cs.LG 2026-05 unverdicted novelty 7.0

    InvEvolve evolves white-box inventory policies from LLMs with statistical safety guarantees and outperforms classical and deep learning methods on synthetic and real retail data.

  3. $\Sigma$-Mem: An Online Reliability Memory for LLM-based Multi-Agent Systems

    cs.MA 2026-07 conditional novelty 6.0

    Online symmetric reliability memory for LLM multi-agent systems accumulates bounded competence and peer-relationship evidence and supports steering, routing, and weighted voting without retraining.

  4. GRAIL: Gradient-Reweighted Advantages for Reinforcement Learning with Verifiable Rewards

    cs.CL 2026-06 unverdicted novelty 6.0

    GRAIL reweights token advantages via gradient saliency in RLVR, outperforming GRPO by 3.60% accuracy and 3.05% Pass@3 on five LLM families.

  5. InvEvolve: Evolving White-Box Inventory Policies via Large Language Models with Performance Guarantees

    cs.LG 2026-05 unverdicted novelty 6.0

    InvEvolve uses LLMs and RL to generate certified inventory policies that outperform classical and deep learning methods on synthetic and real data while providing multi-period performance guarantees.

  6. InvEvolve: Evolving White-Box Inventory Policies via Large Language Models with Performance Guarantees

    cs.LG 2026-05 unverdicted novelty 6.0

    InvEvolve evolves inventory policies using LLMs with RL and provides statistical safety guarantees, outperforming classical and DL methods on synthetic and real data.

  7. PiERN: Token-Level Routing for Integrating High-Precision Computation and Reasoning

    cs.LG 2025-09 unverdicted novelty 6.0

    PiERN proposes token-level routing of physically-isolated experts to embed high-precision computation directly into LLMs, reporting higher accuracy and lower latency, token count, and energy use than fine-tuning or mu...

  8. Step-DPO: Step-wise Preference Optimization for Long-chain Reasoning of LLMs

    cs.LG 2024-06 conditional novelty 6.0

    Step-DPO performs preference optimization on individual reasoning steps rather than complete answers, producing nearly 3% accuracy gains on MATH for 70B+ parameter models with 10K preference pairs.

  9. The Periodic Table of LLM Reasoning: A Structured Survey of Reasoning Paradigms, Methods, and Failure Modes

    cs.CL 2026-06 unverdicted novelty 4.0

    A literature survey that introduces a taxonomy for LLM reasoning paradigms, analyzes methodological trends, and synthesizes failure modes from over 300 papers.

  10. Learning to Reason at the Frontier of Learnability

    cs.LG 2025-02 unverdicted novelty 4.0

    A curriculum sampling questions with high variance in success rate improves reinforcement learning performance for LLM reasoning tasks.

  11. Step-Video-T2V Technical Report: The Practice, Challenges, and Future of Video Foundation Model

    cs.CV 2025-02 unverdicted novelty 4.0

    Step-Video-T2V describes a 30B-parameter text-to-video model with custom Video-VAE, 3D DiT, flow matching, and Video-DPO that claims state-of-the-art results on a new internal benchmark.