My report:
HuixiangDou: Overcoming Group Chat Scenarios with LLM-based Technical Assistance arxiv2401.08772
My favorate projects:
- llama onnx format and single demo without
torch
- how to optimize GEMM,armv7/aarch64/aarch64-int8/cuda/cuda-int4/vulkan all supported
- ML solution for long-tailed demands, MegFlow is implemented with Rust and Python
-
A free open service for onnx/trt/ncnn/openvino model zoo and online model conversion and testing