Issue with getting VLM Model on board

Hi Community,

I have tried to follow GitHub - msorenss/hailo-frigate-standalone: Standalone Hailo-10H Frigate and VLM scaffold for Raspberry Pi 5 with upstream attribution to mikehailodev/frigate-hass-addons-h10 · GitHub and every went well.

Frigate is running fine and Motion detection is doing well.

But I am not succeeding in having the Hailo VLM Chat working; I only get the Demo_Msg answer. The logs tell me there is an issue to get the model loaded. I have with the Qwen2-VL-2b-Instruct.hef et Qwen3-VL-2b-Instruct.hef.

Is it because I got the Hailo10 4Go and not the 8Go version.

My setup: Raspberry PI 5/8Go, SEEED dual NVME hat, Hailo10/4Go+250Go SSD, Debian 13.7, Hailoxx 5.4.

Thanks for your support