Tmux and Imported servers

[dysrea]

Got it set up on windows
Tried installing qwen
Realised it requires tmux
Downloaded the qwen model from growser but cant find any place to import it (.gguf)

If not that, is there a way to install.tmux on windows, if yes please help me out

[Michiel-VandeVelde]

Hey, thanks for reporting this.

tmux usually isn’t something you install directly in native Windows. The cleaner route is probably WSL, Docker, or a Linux setup for anything that depends on it.

For the .gguf import: could you share a bit more detail about how you installed Odysseus, where you downloaded the model from, and what flow you expected for importing it? If this isn’t supported clearly yet, it may be something we need to document or track as a Windows/model-import issue.

[dysrea]

Tmux thingy: okay....will look into it

I followed the native windows installation procedure (by hand)
Rest I have batch script to startup the server.

The gguf i downloaded from huggingface website
Qwen2.5-Coder-3B-Instruct.Q4_0.gguf

I moved the file to root dir of odysseus n set the dir to be scanned for it but cookbook cant seem to find the file in that directory. It seems to recognise the \data folder in it, tried moving it in there too but no luck.

[dysrea]

I do feel im doing something wrong here i just can't seem to figure out what exactly it is.

[tronbonjovi]

2 things -

-the tmux thing is likely due to installing odysseus via native windows installation. this means you're using the native windows terminal, where tmux is not typically available.
the other option is the docker install method. what this really means is instead of it using the native windows terminal, it uses the terminal inside a docker container (you can think of this like a mini virtual machine/vm). docker containers/images are typically based off linux, where tmux is supported.

-gguf is a format for models.
the question to ask is what are you using to run your models? (ollama, vllm, etc) that will help determine which format you need.

[dysrea]

Ill be using ollama...
(Thanks for the tmux insight)

[tronbonjovi]

ollama doesn't use gguf, from my understanding. I believe gguf is typically for vllm.
app should help getting ollama installed.

if downloading models for it doesn't seem to work, you can use the terminal: 'ollama pull qwen2.5-coder:3b'.

have fun! :)

[Michiel-VandeVelde]

Can i close this issue?

[dysrea]

Yeap....thanks for the help 👍🏼