Downloads · 30 days
0
masked-kunsiquat/gemma-3-1b-it-litert
gemma-3-1b-it-litert is a machine learning model from masked-kunsiquat. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for litert. The card lists the license as gemma.
This repository hosts optimized variants of Gemma 3 1B Instruction Tuned for the Lattice ecosystem. These models are the backbone of the M9 (Hardware Acceleration) milestone, designed to eliminate DRAM bottlenecks on…
Downloads · 30 days
0
Access
Public
Updated Apr 20, 2026
Repo size
10.1 GB
Likes
0
Public
Click a slice to open those files.
.litertlm4.1 GB · 67%
From the Hugging Face model README
This repository hosts optimized variants of Gemma 3 1B Instruction Tuned for the Lattice ecosystem. These models are the backbone of the M9 (Hardware Acceleration) milestone, designed to eliminate DRAM bottlenecks on modern Android silicon.
| File | Tier | Target Hardware | Optimization |
|---|---|---|---|
gemma3-1b-it-elite.litertlm | Elite | Snapdragon 8 Elite (SM8750) | Adreno 830 AOT Kernels |
gemma3-1b-it-ultra.litertlm | Ultra | Snapdragon 8 Gen 3 (SM8650) | Adreno 750 AOT Kernels |
gemma3-1b-it-universal.litertlm | Universal | Generic ARM64 / Vulkan | JIT / OpenCL Fallback |
ekv1280) — Optimized for Chain-of-Thought "Diagnosis of Thought" (DoT) loops.The Lattice downloadModels Gradle task intelligently selects the asset based on the device's Build.BOARD.
elite variant for ~10s reframes.ultra variant for ~15s reframes.universal to ensure compatibility across the Android ecosystem.In accordance with the Lattice Total Air-Gap protocol, these models:
PiiShield).Optimized builds sourced and mirrored from litert-community/Gemma3-1B-IT.