Gustav Larsson

Hi @My-captain, great question! Apologies for the slow response. We are working on making quantization recipes publicly available. It will take some time for us to get there, but this...

Thanks for letting us know! This helps us prioritize. We do not have a concrete timeline for this at this time.

It looks like you do not have Git installed. Can you try installing Git first, ensuring that you can run the `git` command successfully, and then try again? We will...

Hi @zhenchong, right now you have two options: * Split the model into parts (you may need this anyway to execute it on the NPU) * Use QAIRT SDK offline...

@zhenchong Unfortunately we don't have any ways to automatically split at this time. What I mean is you will have to split the model first manually, and then submit each...

@zhenchong We currently don't have a recipe for Qwen3. We hope to add that in the future. Unfortunately I have nothing to share until that is ready.

We do not have an example app for Stable Diffusion. That is something that we hope to add in the future (see https://github.com/quic/ai-hub-apps/issues/18 for instance). Until then, here is a...

By 8gen4, I assume you mean Snapdragon 8 Elite, which is the successor of 8gen3. While newer generations generally are faster, it may not be true for every single model....

@htwang14 Thanks for filing this! This is indeed a bug. Gray ops can legitimately happen when the ops are fused away and never end up running individually at the lowest...

We are still tracking this internally, so even if this gets marked as stale and closes, we are still working on this.