You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on Jan 9, 2024. It is now read-only.
USE CASE
User would like to test different LORAs clip strengh vs model strenght. ... BUT User uses their own workflow and loader becaues it's linked to a workflow that took 3 weeks to build.
PROBLEM:
XY graph won't start unless fed dependencies.
It is not possible to consistantly obtain said dependencies due to them being created at the loader level.
User is no able to provide those because of said existing workflow (custom prompting loops, etc) and manual generation is impossible given the large amount of data it requires.
OUTCOME:
User can either:
Use efficiency nodes ONLY when there's a need for the graphs that don't require dependencies, making the workflow so complex in branching (even using R3 nodes) that it's error prone
Use separate workflow pages for certain kinds of testing, but loses deterministic output (because the CLIP_L/CLIP_G magic happens inside the loader they don't use elsewhere).
In simpler words, efficiency nodes provides bar-none the BEST XY solution at the moment because it's able to overide things like Seeds defined earlier in the workflow, but the user is completely bound to the loader for most (but not all) XY graphs. for example i am able to do seed graphs, cfg graphs because those do not require dependancies.
Thank you for considering this. I appreciate it.
The text was updated successfully, but these errors were encountered:
Sign up for freeto subscribe to this conversation on GitHub.
Already have an account?
Sign in.
USE CASE
User would like to test different LORAs clip strengh vs model strenght.
... BUT User uses their own workflow and loader becaues it's linked to a workflow that took 3 weeks to build.
PROBLEM:
OUTCOME:
User can either:
In simpler words, efficiency nodes provides bar-none the BEST XY solution at the moment because it's able to overide things like Seeds defined earlier in the workflow, but the user is completely bound to the loader for most (but not all) XY graphs. for example i am able to do seed graphs, cfg graphs because those do not require dependancies.
Thank you for considering this. I appreciate it.
The text was updated successfully, but these errors were encountered: