Pinned Loading
Repositories
Showing 10 of 144 repositories
- laya-axera Public Forked from mizorewww/laya-mlx
Native MLX runtime for Laya typed decision models — 7–14 ms short decisions on M3 Max. No text generation, PyTorch, or cloud API.
- QAT.PaddleOCR Public
This project is based on Baidu's PaddleOCR and AXERA's self-developed Tongyuan mixed-precision NPU, implementing quantization-aware training (QAT) for PaddleOCR models and efficient deployment on the NPU side.
- laya.axera Public
Laya typed-decision models for AX8850/NPU3: ONNX export, task-aware calibration, Pulsar2 build, and board validation.
- Multimodal_RAG-AX650N Public
-
-
- QAT.Ultralytics.YOLOv5 Public
-
- ax-video-sdk Public
基于 AXERA 的 SDK 实现,简单易懂的C++现代编解码库,支持axcl/ax650/ax620e(ax630c/ax620q/ax620qp),支持jpeg编解码,支持264/265编解码