Ability to run litertlm models like Gemma 4 E4B

[Jakarrrg]

Would it be possible to load litertlm models like Gemma 4, which would run on my GPU (Google pixel 6 pro/GrapheneOS) and run much faster than gguf models?
I have used other apps like "https://github.com/jegly/Box" that can do it.

[Siddhesh2377]

Hey @Jakarrrg Yaa, I can work on this

[VarshaWanjari0]

@Siddhesh2377 You can add support for image classification models as need same backed as litert

[Siddhesh2377]

yaa, once litert gets in we can try out more new things !