description
ADD Custom OP
Model
ONNX
ONNX-Open Neural Network Exchange
Most PPLNN supported ops are based on onnx opset 11. If you are using onnx model with different opset version, you need to convert your onnx model opset version to 11.
ONNX officially provided an opset convert tool version_converter. Its tutorials is at: Version Conversion. Please update to onnx v1.11(or above) and try version_converter:
import onnx
from onnx import version_converter
model = onnx.load("<your_path_to_onnx_model>")
converted_model = version_converter.convert_version(onnx_model, 11)
onnx.save(converted_model, "<your_save_path>")
PyTorch to ONNX
https://pytorch.org/docs/master/onnx.html
Deploy ONNX
https://onnx.ai/supported-tools.html#deployModel
Qualcomm Mobile Deploy
Snapdragon Neural Processing Engine SDK
文档信息
- 本文作者：Kiah Zhao
- 本文链接：/2023/04/09/nn_deploy_embed/
- 版权声明：自由转载-非商用-非衍生-保持署名（创意共享3.0许可证）
