Help with Pulsar2: I don’t know if it’s my hardware, my lack of experience, or just bad luck

[s00pa]

Hi everyone,

I’ve been trying for days to convert my LLM model into a format compatible with the LLM module using Pulsar2 inside Docker. I followed a tutorial (link below), and while I’ve managed to make great progress in my project—learning things I never thought I could—I’ve hit a complete wall here.

To give you some context: I’ve set up the environment, prepared my model in safetensor, loaded Docker, and executed the commands step by step. But I still can’t get a working axmodel. At this point, I don’t know if my hardware is the issue, if I’m making some critical mistake, or if this process is just way harder than I expected.

I’ve followed the tutorial as closely as possible, even seeking help from AI (which sometimes helps, sometimes makes things worse). But I’m now completely exhausted and honestly on the verge of giving up. Is this supposed to be this difficult? Has anyone else struggled with this?

If anyone has experience with Pulsar2 and converting LLM models, I’d really appreciate any advice. Here’s the tutorial I’ve been following: https://pulsar2-docs.readthedocs.io/en/latest/user_guides_quick/quick_start_prepare.html

Thanks in advance, any help would be greatly appreciated!