Skip to content
Discussion options

You must be logged in to vote

Yes, OFTv2 should be used with a lower LR than a normal lora.
Prodigy, in general, does not work well with OFTv2. Many people have reported that Prodigy struggles to find a LR for OFTv2 runs.
Normally you do not train the text encoder with Prodigy, unless you can use split groups, the text encoder in general likes to have a very low LR.
Whether OFTv2 follows historical info regarding TE training is an open discussion I suppose.

Replies: 1 comment 3 replies

Comment options

You must be logged in to vote
3 replies
@yamatazen
Comment options

@Calamdor
Comment options

@yamatazen
Comment options

Answer selected by yamatazen
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Category
Q&A
Labels
None yet
2 participants