What it takes to run Gemma 4 E4B on an iPhone

1 pointsposted 4 hours ago
by kehmka

2 Comments

mugul

4 hours ago

Great job, very interesting! However I wonder: since you are targetting iOS, why only restrict yourself to generalist frameworks such as llama.cpp and LiteRT-LM? MLX Swift looks like a better fit, and the model catalog on HF's MLX community is pretty huge and diverse.