[Question] Llama-v2-7B-Chat Qnn-Gpu model generation

[sparkleholic]

According to the updates of Qualcomm Direct SDK 2.29.0.241129 release, Llama-v2-7B-Chat model can be supported by Qnn-Gpu Backend as well as Qnn-HTP backends.

As far as I know, when I build a Llama-v2-7B-Chat via ai-hub, it supports Qnn-HTP.
Could you explain how to make a Llama-v2-7B-Chat model for supporting Qnn-Gpu backend?

[mestrona-3]

Hi @sparkleholic , thanks for the question! We'd recommend asking questions via our Slack Community for faster response times :)

That being said, we (AI Hub team) haven't validated our Llama-v2 recipe on the QNN-Gpu backend.