[help] Am I just blind? Where's the LLM model connection or setup information?

Description

I truly have been poking around in the manual, the interface, and the other docs here and I must be having a stroke every time I come across information on just how I'm supposed to connect an LLM, where to set the URL to it, hell I don't even know what providers your system can interact with, or what site to go to in order to obtain an API key.

Is ANYONE else just as lost when trying to find this information within the README, or the docs/manual? Maybe I really am having a full on stroke or other episode of some sort whenever I do come across it because I'm tellin' ya'll, I literally cannot remember ever seeing a single answer to the most basic setup items for the LLM model and I'm just freaking out because I'd think that the information I'm looking for would be pretty much one of the very first things the devs would write up when creating an AI personal Assistant, your documentation would kind of need information on how to connect an LLM and the assistant so the assistant can have some intelligence (as well as a vector search database, the RAG and MCP implementations, etc)

Please point me in the right direction, and be easy on this old guy. I've had some major things going on lately that have me discombobulated and I'm a little out of sorts. Now, I truly believe that while I'm a bit off right now, there is no way that I've been so out of it I'd miss this...

Thanks, S. McGowan - @rushowr

Hi @rushowr

Thank you for reaching out and sharing your confusion. I understand it can be frustrating when you can't find the information you need easily.

Currently, this plugin only supports Alibaba Cloud's Qwen model(for more details of API token please access https://bailian.console.aliyun.com/#/home). As for the API Token configuration, you can find the relevant settings in the settings page in Obsidian.

Sorry for the lack of clarity in our documentation and will work on improving it to make such basic setup information more accessible. Let me know if you run into any further issues while setting things up or if you want the personal-assistant plugin to support the models you can access.

Ah, thank you for the clarification! I did see the api token settings as shown in your screenshots. The issue was the fact that I didn't know it was locked to Alibaba Cloud.

I have many models available on my servers to save money, and use multiple methods to access and interact with them including Ollama, 5ire, AnythingLLM, LMSTUDIO, and OpenWebUI among others. Any chance of getting local API connectivity? Most of the local AI APIs I use are OpenAI API compatible, if that helps.

If you need further or more specific information about what models and/or servers I currently intend to use, etc, please connect with me and I'll gladly supply information async, or I can also connect in a messaging app to work and test things synchronously with you.

Hi @rushowr. It's great! I understand your problem.

Currently, the personal-assistant plugin requires the support of the LLM model and the embedding model. Since I'm using the langchain framework, it's okay to support the local Ollama model etc.

However, I don't know much about the local text embedding model. The hard requirement for the embedding is that the dimensionality of the embedding vectors should support 512. So I need to spend some time researching this.

As a full-time employee, I can't afford a personal server to run a local model for testing. I will release a beta version to help you access the local LLM. If you have any problem that needs to be fixed, please let me know.

thank you, I appreciate the assistance. If you'd like, by way of thanks, I can set you up with an account on my Dell R720 that I've set up with a GeForce GTX1060 and possibly even assist by setting up a model or models that you wish to test against on my main workstation which is running an Nvidia RTX 3060. I should be able to set you up with the model(s) running on the more powerful GPU on my workstation and you having access to a docker container for development that you could configure for the API calls to my workstation's Ollama server, or even the AnythingLLM or LM Studio instances (LM Studio, IIRC, allows for embedding models in addition the LLMs. In fact I think AnythingLLM does as well).

hi @rushowr , please try 1.5.6-beta to check if it works for you to run with local llm server.

here is the AI setting page:

Thank you @edonyzpc , I'll pull an update and check it out 👍

Hello, i have all the models and connections (i checked with page assist, the Firefox extension), but it doesn't react to anything sent in chat, and i get an error instead:

I pulled the update but due to some large work interrupting things going on out here on my end, I've been unable to actually sit down and test the update with an actual chat. Sorry, I do hope to get back to test soon.

Hello, i have all the models and connections (i checked with page assist, the Firefox extension), but it doesn't react to anything sent in chat, and i get an error instead:

Could you please provide a bit more detail so we can better reproduce and diagnose the problem? For example:

What model are you using?

• Are you self-hosting or using a public endpoint?
• Can you share the full error message (text is preferred over screenshots if possible)?
• Does this happen with all queries or only specific ones?
• Any relevant logs from the console or network tab?

This information will help us identify the root cause faster. Thanks again for your patience!