π Hangzhou, China | π€ On-Device AI & LLM Enthusiast | ποΈ MNN Core Developer
Inference Engine Expert at Alibaba (Taotian Group), core architect of MNN. Making LLMs run fast on edge devices.
- π§ MNN - Blazing fast, lightweight deep learning inference engine (13k+ stars)
- π¬ mnn-llm - LLM deployment on mobile devices
- π onnx-llm - ONNX-based LLM inference
- π¦ llm-export - Export LLM models to ONNX format for cross-platform deployment
- π jinja.cpp - Single-header C++11 Jinja2 engine for LLM chat templates
- π€ tokenizer.cpp - Lightweight C++ library for LLM tokenization, HuggingFace compatible
- π€ mnn-asr - MNN-based Automatic Speech Recognition demo
- π mnn-tts - MNN-based Text-to-Speech demo
- π― mnn-yolo - MNN-based YOLO object detection demo
- βοΈ mnn-segment-anything - MNN-based Segment Anything demo
- π± mnn-mobilenet - MNN-based MobileNet classification demo
- π¨ mnn-stable-diffusion - MNN-based Stable Diffusion image generation
- πΌοΈ clawdicons - Icon collection
- π§ͺ llm-lab - LLM experiments and research notes
π Tech Blog | β Follow me on GitHub