Unable to locate llama3.2 model and tokenizer files in the ios demo app

[hietalajulius]

Following this tutorial: https://github.com/pytorch/executorch/blob/main/examples/demo-apps/apple_ios/LLaMA/docs/delegates/xnnpack_README.md (on main, commit 92d1d1e410b11945869472e88a2247305921989a, iPhone 12 Pro, iOS 17.6.1)

When I drag and drop the model and tokenizer files in finder, they are not visible in the file navigator when trying to select them in the app:

(llama3_2.pte and tokenizer.bin inside the directories)

Any ideas what I'm doing wrong?

update: airdropping the files from my mac makes them visible and selectable in the app but it still says "Select model..." when both model and tokenizer are seemingly set.

[iseeyuan]

@shoumikhin , could you help look at this?

[a8nova]

I am also facing a similar issue

[WuhanMonkey]

cc. @Riandy

[Riandy]

Try dropping them "outside" the models and tokenizers folder, basically direct child of the "iLLAMA" app in your finder.

Something like iLLaMA

• <your_model_file.pte>
• <your_tokenizer_file.bin>

[hietalajulius]

Try dropping them "outside" the models and tokenizers folder, basically direct child of the "iLLAMA" app in your finder.

Something like iLLaMA

• <your_model_file.pte>
• <your_tokenizer_file.bin>

Looks like the same thing with this setup

[shoumikhin]

@hietalajulius try removing the app from the device, deploying&launching the app again, then copy the files, wait till copying finishes and proceed with the app UI to pick them.

[hietalajulius]

@hietalajulius try removing the app from the device, deploying&launching the app again, then copy the files, wait till copying finishes and proceed with the app UI to pick them.

Thanks, that worked! I was convinced I did it already but apparently not :D