Skip to content

Commit c0e5f14

Browse files
committed
update github docs
1 parent 0d88b28 commit c0e5f14

File tree

1 file changed

+3
-1
lines changed

1 file changed

+3
-1
lines changed

docs/usage.md

Lines changed: 3 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -111,4 +111,6 @@ The mode option has no effect on the RAG performance, but controls how much extr
111111

112112
### Github Repository
113113

114-
After choosing the configuration options, you have the choice to also include a Github repository URL to include in the indexing process. The URL provided will automatically be parsed for the repository owner and repository name information. This will supplement the PDF data ingestion to provide more specific output for workflow specific steps in the description and parametric domains.
114+
After choosing the configuration options, you have the choice to also include a Github repository URL to include in the indexing process. The URL provided will automatically be parsed for the repository owner and repository name information. This will supplement the PDF data ingestion to provide more specific output for workflow specific steps in the description and parametric domains. If a github URL is entered, you'll be asked to confirm the branch of the repo to index (if none is entered, will default to `main`). You will also have the choice to specify directory and filter extension filters.
115+
116+
For each filter, you will have the option to specify whether to conditionally exclude certain directories and file types or to inclusively include certain directories and file types. Specify the directory path for directories to include in the filter. For file types, include the file extension, for example, `".txt", ".md"` with the `"include"` filter type will only include files that are of type text and markdown. Note, the filters are potentially important for large repositories. Indexing repositories with large output, log, or data files can incur signficant performance overhead and additionally can lower output quality by polluting the retrieval step with noise.

0 commit comments

Comments
 (0)