您好,我用自己数据集训练得到的yolov3-tiny做加速,float32位精度可以转成功,但是转float16时报错,
WARNING: Half2 support requested on hardware without native FP16 support, performance will be negatively affected.
ERROR: ../builder/cudnnBuilderWeightConverters.cpp (310) - Misc Error in transformWeightsIfFP: 1 (Weights are outside of fp16 range. A possible fix is to retrain the model with regularization to bring the magnitude of the weights down.)
这是什么原因呢？
运行的代码为：
./buildEngine -i model/yolov3-tiny-highway-704.onnx -o model/yolov3-tiny-hoghway-704-fp16.engine -b 1 -m 1
CaoWGG
commented
4 years ago
您好,我用自己数据集训练得到的yolov3-tiny做加速,float32位精度可以转成功,但是转float16时报错, WARNING: Half2 support requested on hardware without native FP16 support, performance will be negatively affected. ERROR: ../builder/cudnnBuilderWeightConverters.cpp (310) - Misc Error in transformWeightsIfFP: 1 (Weights are outside of fp16 range. A possible fix is to retrain the model with regularization to bring the magnitude of the weights down.) 这是什么原因呢？ 运行的代码为： ./buildEngine -i model/yolov3-tiny-highway-704.onnx -o model/yolov3-tiny-hoghway-704-fp16.engine -b 1 -m 1