EMNLP 2025: FLAMES: Improving LLM Math Reasoning via a Fine-Grained Analysis of the Data Synthesis Pipeline

Parker Seegmiller, Kartik Mehta, Soumya Saha, Chenyang Tao, Shereen Oraby, Arpit Gupta, Tagyoung Chung, Mohit Bansal, Nanyun Peng
August 2025
PDF Cite
FLAMES

Abstract

Discovering what works for math reasoning data synthesis

Type
Conference paper
Publication
The 2025 Conference on Empirical Methods in Natural Language Processing
Parker Seegmiller
Parker Seegmiller
Innovation Fellow, PhD Candidate @ Dartmouth CS | Applied Science Intern @ Amazon

nlp @ dartmouth college, llms @ amazon