Skip to content

TensorRT-API构建方式是否支持对pytorch_quatization量化训练导出Q/DQ-onnx转换int8 engine? #7

@songkq

Description

@songkq

@DataXujing Hi,请问是否尝试过使用TensorRT-API方式从量化训练导出的Q/DQ-onnx模型构建trt_int8 engine?
我尝试直接使用trtexec对Q/DQ-onnx进行转换,nsight分析发现推理过程transpose+reshape耗时占用高,想结合api的方式优化这部分性能,可否给一些建议?多谢~

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions