huggingface / huggingface/datasets

GDPval dataset Word docs corrupted

Open
#7,915 1 comment 0 reactions 0 assignees View on GitHub
Dominant language
Python
Stars
22k
Forks
3.4k
Avg merge
5d 7h
Merged PRs (30d)
17

Description

The [openai/gdpval](https://huggingface.co/datasets/openai/gdpval) dataset on Hugging Face contains Word .docx files with two types of corruption that cause Microsoft Word to display an "unreadable content" error.

### Root Causes

1. **Corrupted settings.xml**: The `word/settings.xml` file uses incorrect namespace prefixes (`ns0:`, `ns1:`, etc.) instead of the proper prefixes (`w:`, `mc:`, `m:`, etc.)

2. **Malformed TargetMode attributes**: Some files have `TargetMode="External"` attributes missing their closing `/>` tag in hyperlink relationships

Both issues cause Word to reject the files even though the XML structure is technically valid.

I have a fix for the issue here https://github.com/alexheat/gdpval-docx-fix

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.