The 45 repos behind github.com/john-rocky/windows-zoo: ONNX, OpenVINO IR and precompiled Intel NPU blob per model, measured on one Windows laptop.
Daisuke Majima (MLBoy) PRO
mlboydaisuke
AI & ML interests
On-device AI. Core AI model zoo (60+ models) · devicemark iPhone LLM leaderboard · author of "The Art of Core AI" → github.com/john-rocky
Recent Activity
posted an update about 1 hour ago
WeVisOne-VL-4B on a Galaxy S26: Android developers can have their app box the objects a user names in a photo, with no detector to train.
https://huggingface.co/litert-community/WeVisOne-VL-4B
By WeChat Vision at Tencent, built on Qwen3-VL: detection, pointing, counting and OCR.
In the clip, an offline test app runs it on the GPU: a 1967 NASA photo, Detect "person, hat", and 3 people and 2 hats boxed 7.3 s after the Ask tap.
To run it: load the 3.5 GB file with the litertlm-android 0.19.1 library, as in the clip; the card has the resize step and the task prompts. updated a model about 1 hour ago
litert-community/LFM2.5-VL-1.6B updated a model about 1 hour ago
litert-community/LFM2.5-1.2B-InstructOrganizations
Community .aimodel bundles of Gemma 4 (E2B, E4B, 12B, 31B) for Apple Core AI on iOS 27 / macOS 27, with recipes and measured speeds.
PyTorch models exported to ExecuTorch .pte for on-device inference — XNNPACK (CPU, portable) and Core ML builds, each parity-checked.
PyTorch models converted to Core ML for on-device inference on iPhone, iPad and Mac.
Models that read a text and pick one of the options you write, each with a probability. LiteRT conversions for Android, all Apache-2.0.
-
litert-community/laya-LiteRT
Text Classification • Updated • 947 • 4 -
litert-community/Laya-Multilingual-LiteRT
Text Classification • Updated • 1.29k • 6 -
litert-community/Laya-English-LiteRT
Text Classification • Updated • 461 • 4 -
litert-community/Julia-1-LiteRT
Text Classification • Updated • 126 • 1
Models converted to Apple's Core AI format for on-device inference on iPhone, iPad and Mac — LLMs, VLMs, ASR, TTS and a vision zoo.
Models I converted to LiteRT and shipped to litert-community: LLMs, VLMs, embeddings, TTS, ASR, and a vision zoo. Measured on-device.
The 45 repos behind github.com/john-rocky/windows-zoo: ONNX, OpenVINO IR and precompiled Intel NPU blob per model, measured on one Windows laptop.
Models that read a text and pick one of the options you write, each with a probability. LiteRT conversions for Android, all Apache-2.0.
-
litert-community/laya-LiteRT
Text Classification • Updated • 947 • 4 -
litert-community/Laya-Multilingual-LiteRT
Text Classification • Updated • 1.29k • 6 -
litert-community/Laya-English-LiteRT
Text Classification • Updated • 461 • 4 -
litert-community/Julia-1-LiteRT
Text Classification • Updated • 126 • 1
Community .aimodel bundles of Gemma 4 (E2B, E4B, 12B, 31B) for Apple Core AI on iOS 27 / macOS 27, with recipes and measured speeds.
Models converted to Apple's Core AI format for on-device inference on iPhone, iPad and Mac — LLMs, VLMs, ASR, TTS and a vision zoo.
PyTorch models exported to ExecuTorch .pte for on-device inference — XNNPACK (CPU, portable) and Core ML builds, each parity-checked.
Models I converted to LiteRT and shipped to litert-community: LLMs, VLMs, embeddings, TTS, ASR, and a vision zoo. Measured on-device.
PyTorch models converted to Core ML for on-device inference on iPhone, iPad and Mac.