MathCritique-76k

Citation Author(s):
Dingwen
Yang
Zhiheng
Xi
Jiafu
Tang
Jixuan
Huang
Submitted by:
Dingwen Yang
Last updated:
Thu, 01/09/2025 - 08:54
DOI:
10.21227/ncx4-9j08
License:
Creative Commons Attribution
17 Views
Categories:
Artificial Intelligence
Keywords:
Critique model, inference compute scaling, large language models, reasoning, scalable oversight, self-improvement
0
0 ratings - Please login to submit your rating.
ACCESS DATASET CITE SHARE/EMBED

Abstract 

To train critique models capable of delivering step-level supervision and constructive feedback for reasoning, we introduce AutoMathCritique—an automated and scalable framework for collecting critique data.
This framework consists of three main stages: flawed reasoning path construction, critique generation, and data filtering. Using AutoMathCritique, we create a dataset containing $76,321$ samples named MathCritique-76k.

Instructions: 

The data is in sharegpt format.

Dataset Files

LOGIN TO ACCESS DATASET FILES

 

QUESTIONS?