CUDA=1 on startup

[LordEschatus]

So i have downloaded the prebuilt 1.4 for linux, and when running:

user@host:~/Downloads/colibri-v1.4.0-linux-x86_64$ ./coli chat --model ./GLM-5.2-colibri-int4-g64-with-int8-mtp/ --gpu auto

     ▄▀▀▀▄  ▄        colibri v1.4.0
  ▄▄▄▄▀▀▀▀▄▀▀        tiny engine, immense model
      ▀▀▀▀▀▀▀        GLM-5.2 · 744B MoE · int4 · streaming CPU
        ▀▀▀▀         chat ·  · ram -GB · topp off
          ▀          
  ──────────────────────────────────────────────────────────
  ↺ resuming a saved conversation: 253 tokens, last written just now
    it steers tone, language and topic. :reset clears it · KVSAVE=0 disables saving · delete ./GLM-5.2-colibri-int4-g64-with-int8-mtp/.coli_kv to start clean
**--gpu needs the CUDA build: make colibri CUDA=1 (this binary is CPU-only)**
user@host:~/Downloads/colibri-v1.4.0-linux-x86_64$

whether I run --gpu auto --gpu 0 --gpu 0,1 -r --auto-tier --gpu x,y etc... I continue to encounter the CPU-only warning.

I have tried cloning the repository to make it myself, but Im not sure what command I would pass to setup.sh

I am aware of https://github.com/JustVugg/colibri/actions/runs/30677132862 , but please forgive me , I am not a developer, can read code (badly) , but not sure what the resolution is here.

[brad-evony]

If you want the Cuda support just like if you want to try the latest updates and bug fixes, you need to compile the executable program yourself. I didn't know how to do that either until I asked AI how to do it.