Is there a user friendly self hosted AI guide? Low or No Code

[Despo]

I want to begin learning more about AI and some generic use cases so I don’t have super specific objectives here. But I want a user friendly guide on how to set up a self hosted AI solution where I can leverage models from hugging face or whatever and just start the creative juices flowing. This more of an academic exercise than anything.

Possible ideas, image recognition, DIY LLM, Organize File Folder Contents, etc etc.

[anon75264233]

Thank you for asking your question. Just letting you know that I changed the category.

The wiki section is where the actual wiki’s go, not requests.

---

Happy forum’ing!

[vivante]

That’s a bit of a broad request, but some basics are:

• HuggingFace is de-facto the place for publishing LLMs
• HuggingFace has malicious models, be careful
• You can’t do DIY LLM, that costs millions of dollars. You can do some fine-tuning
• Automatic 1111 is easy to use UI for image generation with stable diffusion
• LM studio is new, looks interesting. Thanks to @shadragon for binging it to our attention
• If you don’t have somewhat capable GPU it’s gonna suck
• You can do simple things by utilizing “simple” neural networks such as Tesseract’s OCR engine to solve practical problems, get paid more and piss off your IT department in the process. And all it takes is few lines of python and a can-do attitude

[bedHedd]

Might be out of your comfort zone, but if you want to dip your toe into linux and the command line interface, ollama is great.

Wendell wrote a guide for Ubuntu

Ubuntu 22.04 - From Zero to 70b Llama (with BOTH Nvidia and AMD 7xxx series GPUs)

Steps I took # first fully update 22.04 LTS apt update && apt upgrade -y # reboot ... you probably got a newer kernel... # ensure remote access Since we are updating the video driver, and it is likely you don't have more than one gpu in the system, ensure you can ```ssh``` into the system from another system. This is useful for both setup and t…

[Zszywany]

This might be a good video on the topic.
Chuck shows how to quickly setup ollama and a Web UI for it.

host ALL your AI locally

[Despo]

thanks! Thank you so much for the concise response. it was designed to be a broad request so I get that. I was more looking for good examples of “homelab” AI uses cases just to get my feet wet.

[Despo]

Amazing, I have no idea how I missed this guide. Thanks so much!

[shadragon]

This is how I started, by setting up discrete LLM’s one at a time. It works, no doubts there, but is cumbersome and inflexible, usually needing specific versions of files that conflict with other installs and the dancing around those issues that goes with that.

Check out LMStudio to try thousands of different LLM locally depending on what your system can support. Windows, Mac and Linux versions.

👾 LM Studio - Discover and run local LLMs

Matthew Berman on YouTube does an excellent job of showing you how to run many LLM locally as well as using ties to established online resources.

Matthew Berman

[drbsg]

Another option to consider is AnythingLLM. It can work with most LLM servers such as ollama or LM Studio, and can also help you explore the world of using RAG and vector databases.