Published July 18, 2024 | Version v1
Dataset Open

Results outputs for "Beyond the Hype: Identifying and Analyzing Math Word Problem-Solving Challenges for Large Language Models"

Creators

  • 1. ROR icon Universitat de València

Description

The provided files contain outputs generated by various Large Language Models (LLMs) for solving problems in the SVAMP dataset. Additionally, they include tagged statements of problems that LLMs incorrectly resolved.

This repository includes the following two files:

  • all_data.json --> Contains the generated samples for the SVAMP dataset.
  • df_combined.pkl --> Contains the tagged SVAMP statements of problems that CodeLlama failed to resolve.

Files

all_data.json

Files (249.2 MB)

Name Size Download all
md5:f981dc4a9b382f0216361921b5a6363b
249.0 MB Preview Download
md5:6a7c637c34a8ca98937abf7c37619f0f
169.1 kB Download

Additional details

Related works

References
Dataset: 10.5281/zenodo.11126655 (DOI)

Funding

Generalitat Valenciana
Grupos Emergentes CIGE/2023/063
Ministerio de Ciencia, Innovación y Universidades
TED TED2021-129485B-C42
Ministerio de Ciencia, Innovación y Universidades
Ayudas para contratos predoctorales para la formación de doctores/as PRE2019-090854
Generalitat Valenciana
CIAPOS CIAPOS/2022/163