Problem loading LLM and getting no responses to prompts

I have a Raspberry Pi 5 with 8 gig of memory. I have the AI Hat+ 2 attached to it. Power is supplied to them by a 45w supply from Canakit. I have sucessfully installed Hailo-Apps v25.10.0.

Hailo-ollama server is running on port 8000.

Listing models gives 5 in store.

I tried loading both deepseek and also qwen2:1.5b. After loading all I get is JSON response {“status”:”success”} and nothing else.

cURL gives error 18 or “transfer closed with outstanding read data remaining”.

Sending a prompt results in same thing {“status”:”success”} and nothing else.

Any clue what could be causing this possibly. Appreciate the help.

Hi @BryanSwie ,

Can you please confirm that those steps are not working for you? GitHub - hailo-ai/hailo_model_zoo_genai: Model zoo for Gen AI models for Hailo products

Error 18 means that the server started sending data but the connection was closed before all expected data was received.

Maybe altering "stream": false between true and false might help.

Thanks,

It’s all working fine. I was confused about whether the LLM needed to be downloaded everytime I rebooted the PI. It had downloaded the qwen2:1.5b LLM and keeps in on the MicroSD card. So, next time I tried to download it gave Status Success because I already had it on the SD card. I had previously downloaded it before I installed Hailo-Apps v25,10.0 so I didn’t need to download it again. Still learning. So, topic is closed. It is working fine. Going to install the Web UI next.

1 Like