This is a basic workflow for Auraflow 0.1. This is a very big model. it does run on 10-12gb, but it uses a T5 text encoder so it will be quite slow.
The model is available here: https://huggingface.co/fal/AuraFlow, please read the model card, that's the information people were given about the model, the model is available to download there as well.
VLM Nodes are used to get better prompts, you must look in the repo of nodes for instructions. https://github.com/gokayfem/ComfyUI_VLM_nodes. If you don't like using VLM nodes, chose another workflow or delete them like a normal person.
If you must use the exact same LLM I used, it can be downloaded here: https://huggingface.co/failspy/Meta-Llama-3-8B-Instruct-abliterated-v3-GGUF/tree/main
If you hate the model, don't use it. Comments about the model go to the author, I have no access to his training equipment or datasets. Use common sense.
Leave a tip to support more releases: https://ko-fi.com/marduk191
Discord Support: https://discord.gg/s3kj9VqpKc
Description
v1.0 initial upload
FAQ
Comments (4)
Hello ! I can't launch your workflow because i don't have this model in LLmLoader:
"meta-llama-3-8B-Instruct-Abliterated-v3_q5.gguf"
and i don't know WHERE i can download it !
Even Google and Huggingface don't know.....
But Thanks !
Error:
Error occurred when executing LLMLoader: LLMLoader.load_llm_checkpoint() missing 1 required positional argument: 'ckpt_name' File "D:\Data\Packages\ComfyUI\execution.py", line 152, in recursive_execute output_data, output_ui = get_output_data(obj, input_data_all) File "D:\Data\Packages\ComfyUI\execution.py", line 82, in get_output_data return_values = map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True) File "D:\Data\Packages\ComfyUI\execution.py", line 75, in map_node_over_list results.append(getattr(obj, func)(**slice_dict(input_data_all, i)))use whatever llm you want, check the VLM nodes repository for a selection of starter models. I did indeed get that one from huggingface lol.
her is a link to the one i use https://huggingface.co/failspy/Meta-Llama-3-8B-Instruct-abliterated-v3-GGUF/tree/main
@marduk191 thanks you ! <3 (Sorry for the late response, I'm on vacation!)