A highly compressed, spliced variant of Google's lightweight Gemma 4 E2B-it (Effective 2 Billion, also known as E4B with 4.65 Billion active parameters) model.
Model Description
This repository contains a spliced mixed-precision GGUF binary of the Gemma 4 E2B-it architecture. By surgically removing redundant layers, this model is optimized for hyper-fast, whisper-silent execution on low-memory edge platforms and commodity Apple Silicon hardware.
Format: GGUF (Q4_K_M)
Size: 3.45 GB
Target Platforms: Apple Silicon MacBooks (M1/M2/M3/M4) and standard CPU/GPU local runtimes.
Local Quickstart
Run the model natively via standard llama-cli:
llama-cli -ngl 100 -m ./google_gemma-4-E2B-it-Q4_K_M.gguf -p "The beauty of edge-first AI lies in" -n 128
