Pedro Cuenca
Pedro Cuenca
@nafrinafri @tommy9 We have tested on several M1, M2 and Intel computers, but it's hard to get access to systems with 4 GB of VRAM that can also run Ventura...
That's awesome, @tommy9! I'll prepare a test branch for you to try :)
Interesting, thanks @gary149! I could reproduce the "Reveal in finder..." bug (it doesn't open Finder if the model currently selected is not downloaded), but I could not reproduce the other...
> Testflight (version 1.1) There was [a bug in that version](https://github.com/huggingface/swift-coreml-diffusers/commit/d69e554b46257a0cffdba1d440679e5d1831c171) related to downloading, but I'm not sure it would exhibit the behavior you've seen. I'll create a TestFlight build...
@gary149 a new TestFlight build (1.1.1) is up, could you try to reproduce there? (And Reveal in Finder should work consistently).
Hi @jo32, sorry for the delay in getting back to you. Are those results consistent across runs? I've seen some erratic behaviour on iPhone myself – sometimes it's very fast...
Thanks for the report, I'll try to reproduce!
@3DTOPO That'd be very cool indeed! Let's see if someone in the community takes up the challenge to convert the model to Core ML!
It works on my 2020 iMac (16 GB AMD Radeon Pro 5700 XT), using the `ORIGINAL` attention implementation instead of `SPLIT_EINSUM`.
@oscarbg I believe the main focus of this repo is about Apple Silicon on both macOS and iOS/iPadOS. Happy to share a data point taken on my iMac: iMac Retina...