-
Notifications
You must be signed in to change notification settings - Fork 400
feat: add prediction argument #334
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
base: master
Are you sure you want to change the base?
Conversation
e995b3b
to
1b00b0c
Compare
24c784b
to
b9064ac
Compare
b9064ac
to
3beb06f
Compare
3beb06f
to
300aa58
Compare
Maybe it would make sense to also add support for the recently added |
I haven't been up to date with the recent development so I'm trying to catch up for now. Has it already been added to the repo right? I may be able to add support for it later. |
Yes it has been added with #683 Also there's a distinction between Flux Flow and Sd3 (Discrete) Flow: |
I've added the new prediction methods. I'm unable to check them as I'm on a laptop right now but the logic is the same as with the others so it should work. |
66e121d
to
08aec64
Compare
@stduhpf Feel free to push this PR on your bleedingedge fork or let me know if you think I could improve the code in any way (I'm still not sure if the description is clear enough). |
I think I will re-create the bleedingedge branch from current master with the new c API (#728), I will try to include this PR when I do so. |
Will you include the image preview PR? I'm building a webui (more like a command generator + gallery) and that would be very helpful. I've been thinking of adding a toggle to switch between a mode that works on the mainline branch and another that works on your fork (I still need to add all the new features from mainline but at least that's the plan). |
This PR adds an argument to manually select the prediction type. This allows the user to override the prediction type of the model on some experimental models that use non standard prediction types.
This argument aims to have a similar way to achieve what ComfyUI has with their ModelSamplingDiscrete node.