Permission to reuse and alter/modify "CheckpointLoaderAdvancedDisTorch2MultiGPU" to merge with "DualCLIPLoaderGGUFDisTorch2MultiGPU"

[JuPillet]

Hey,
Thanks a lot for ComfyUI-MultiGPU, great work!
I'm the author of ComfyUI-Multi-Route-Tools where I made some Bypass Route nodes.
I'm trying to create a switch between the native checkpoint CLIP and a GGUF CLIP using my routers. The problem is that it gets complicated when the models have different dimensions.
I’d like to create a new node called CheckpointDualCLIPLoaderGGUFDisTorch2MultiGPU — an altered version of your CheckpointLoaderAdvancedDisTorch2MultiGPU merged with DualCLIPLoaderGGUFDisTorch2MultiGPU — loading checkpoint like CheckpointLoaderAdvancedDisTorch2MultiGPU and merge the CLIP from checkpoint and with an other CLIP like DualCLIPLoaderGGUFDisTorch2MultiGPU do it.
It would depend on your MultiGPU extension. Would you be okay if I reuse and adapt part of your code for this?
I'll properly credit you and link your repo of course.
Thanks in advance!
JuPillet

[Symbiomatrix]

Not sure I follow 100%, but if the issue is just combining unet and clip nodes, then no need to do that in code, this is done with a simple native subgraph, less bug prone and maintenance.
Also, you know about rgthree's fast group bypasser node, right? Just put a group around some nodes, call it whatever, then switch it on and off from a menu. You can also apply multiple groups to the same nodes, though it's a bit more finicky (eg I have a group that controls t2i / i2i over the latent application nodes, and above it a group for the whole gen as the workflow contains ollama prompt gen on the side).