Hi, trying to use the vlm_chat app with Qwen3-VL-2B-Instruct (by adding the model as an ‘extra’ entry in resources_config.yaml) but have found it is not a drop-in replacement for the default Qwen2-VL model.
The download was successful (using the –hef-path flag) and it is now listed as an extra model using the –list-models flag.
Please see the hailort.log output below for runtime errors. Could you please help debug this?
I’m using the Hailo-10H.
Thanks
Chris
[2026-05-14 10:12:34.177] [4847] [HailoRT] [info] [vdevice.cpp:228] [create] Creating vdevice with params: device_count: 1, scheduling_algorithm: ROUND_ROBIN, multi_process_service: false
[2026-05-14 10:12:34.208] [4847] [HailoRT] [info] [device.cpp:54] [Device] OS Version: Linux 6.12.75+rpt-rpi-2712 #1 SMP PREEMPT Debian 1:6.12.75-1+rpt1 (2026-03-11) aarch64
[2026-05-14 10:12:34.268] [4847] [HailoRT] [info] [vlm.cpp:129] [create_unique] Sending 6 HEF chunks to server
[2026-05-14 10:13:12.156] [4847] [HailoRT] [error] [vlm.cpp:684] [generate_impl] CHECK_SUCCESS failed with status=HAILO_INVALID_OPERATION(6) - Failed to generate. Make sure the input data matches what the model expects and there is no other generation in progress
[2026-05-14 10:13:39.027] [4847] [HailoRT] [error] [minja_wrapper.cpp:83] [render] CHECK failed - System role messages can only be provided on the first prompt
[2026-05-14 10:13:39.028] [4847] [HailoRT] [error] [vlm.cpp:599] [generate] CHECK_SUCCESS failed with status=HAILO_INVALID_OPERATION(6)