Does PaddleOCR-VL SFT support LoRA (adapter or merged) in ERNIEKit? #1407
Unanswered
Salehjarad
asked this question in
Q&A
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
Hi, I’m fine-tuning PaddleOCR-VL-0.9B using the official SFT config in ERNIEKit.
Is LoRA (parameter-efficient fine-tuning) currently supported for PaddleOCR-VL?
If yes, is it expected to be used as an adapter at inference time or merged into a standard model directory for PaddleOCR inference?
If not supported yet, is full SFT the recommended approach for now?
Beta Was this translation helpful? Give feedback.
All reactions