FLUX.1-dev Black Forest Labs LocalLLM Image Inferencer

This LLM is to be used under  a Non-Commerial Black Forest License Usage

Easily pull and run it with two docker commands.

docker pull cnmcdee/black-forest
docker run -d \
--name black-forest \
--restart unless-stopped \
--gpus all \
-p 8080:8080 \
-e DEFAULT_MODE=cpu \
-e CPU_THREADS=12 \
-v black-forest:/data \
cnmcdee/black-forest:latest

--gpu all should always be specified.  Otherwise it will run but CPU only generation can take a very long time!

When it comes up it will be visible at port 8080

http://192.168.1.<your ip>:8080

This is a complete standalone powerful ImageLLM and big kudos to Black Forest for allowing it for non-commercial use.  Hosted here:

black-forest-labs/FLUX.1-dev · Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

Black Forest Lab image LLM's have showing incredible popularity with a stunning 775,000 downloads just last month.

Using one of the worlds most advanced LLM's (ChatGPT Astra) - and a separate dedicated building server, we had it build this entire system.

It is a Big Image (But it is a StandAlone)

• Like 46.2 GB, so make some room for it.
• Once it's pulled it's completely standalone, it does not require a Huggingface token, and does not need to pull models. It just works.

This kind of thing is HUGE because it lets the little guy quickly and easily stand up a high quality image generator. Use it for generating memes.

Don't know how to setup docker, it's pretty easy:

Docker Installation Script

Just another Docker Installation Script - Because you can never have enough of these!

Hot Configthinkmeltprotonmail.com

A Beautiful Interface

You can select different ways for it to load on the left, it offerse several options to ensure even minimal CPU's will load.

Over Loading Fault Tolerant

• You can try higher settings and if it fails simply lower the load and click Reload Model

On the right side you can see how your system is being loaded

After 108s, our 4080 system produces the following:

These are really nice images...