Replies: 1 comment 1 reply
-
|
The default backend as of docling 2.14.0 is docling-parse v2, and we recommend using it. You can switch to pypdfium if you see issues with fonts or encodings. pypdfium is about the same speed as docling-parse v2. |
Beta Was this translation helpful? Give feedback.
1 reply
-
|
@cau-git I noticed for both v1 and v2 backend of docling parse , we are using some functionalities of the pypdfium2 library. Is this by design? |
Beta Was this translation helpful? Give feedback.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
There are 3 pdf backends available here and from a previous discussion I understand that the docling v2 parser is much faster than the v1 parser, but how about the pypdfium backend?
The docs page uses the pypdfium backend in the example DocumentConverter config. Is that the recommended approach? How does the docling parser compare with pypdfium? I tried looking through the docs and discussions but I can't seem to find any related information.
Beta Was this translation helpful? Give feedback.
All reactions