Compile llama on local PC

[bowangnio]

hi,
currently qai_hub_model only can compile model on cloud, if we can do it on local pc???

because the qnn sdk version is changing on cloud, and bugs are also.
sometime I built the mode successfully, but sometime failed, due to bugs of cloud.
if we can do it locally, i can keep the sdk version unchanged, and get compiling successfully every time.

[shreyajn]

Multiple QNN versions supported on AI Hub: We support a few versions of QNN. You can compile with any of them.
https://app.aihub.qualcomm.com/docs/hub/api.html#common-options LOOK AT --qairt_version here in common options. These do keep moving ahead though.

Use old job to download the previously compiled model: If you have compiled it before, you can always store the job id and download from them again and again. In compile logs, we have all the compile logs as well, you can copy them and run locally to get the same asset as produced by Hub.