facebookresearch / facebookresearch/CRAG
Full CRAG Document Corpus
- Dominant language
- Jupyter Notebook
- Stars
- 303
- Forks
- 39
- PR merge metrics
- No merged PRs in 30d
Description
Hi, could you please help me with a link to the entire CRAG document corpus?
For each of the three CRAG tasks, the dataset links provide JSONL files where each query includes a search_results field containing the top search hits (usually 5 or 50 pages of HTML text).
What I am looking for is:
A single downloadable corpus that contains all search result documents for all queries combined — essentially, the full set of retrieved pages used across the entire CRAG benchmark.I want to download this complete corpus so I can build a global index and evaluate my own RAG pipeline over it.Do you have such a consolidated corpus available, or a recommended source for obtaining it?
Contributor guide
Assessment
This issue has not been assessed yet.