ComfyUI Workflow
required Custom Nodes:
if you need help check the readme files in the repo readme folder at github.
llama-cpp-python is needed for the loading_method: GGUF (llama-cpp-python) in the Smart LML v2 Node
switch the template in the Smart Language Model Loader v2 to download the model. the node detects if the model already exists, it is also not downloading florence2 anymore when models\florence2 already exists, except for models that are not found in this folder.
load images from folder
Smart LML v2 loads the selected model for the image description
ReplaceString v3 removes unwanted parts
save prompt (.txt, .csv, .json)
if you create one file e.g. random_prompts you can use this file in the wildcard processor. for that:
copy the file to the wildcards folder e.g. comfyui\models\wildcards. reload the page or restart comfyui if that didnt work and load the file from the list.
Description
requires SmartLML 2.4.4 (with docker amd support)
FAQ
Comments (1)
From the looks of it; a lots of LLM are supported through docker loaded tools it's nice to have wide choice but apparently it's an NVIDIA only setup ?
Is it gonna break if i install those with my amd setup (linux) ?