Int8 with TRT

baidu "caffe int8 "

 

https://blog.csdn.net/qq_33431368/article/details/85029041 利用量化工具caffe-int8-convert-tools实现caffemodel量化(double32->int8)

https://note.youdao.com/ynoteshare1/index.html?id=829ba6cabfde990e2832b048a4f492b3&type=note#/ (good)  基于tensorRT方案的INT8量化实现

http://on-demand.gputechconf.com/gtc/2017/presentation/s7310-8-bit-inference-with-tensorrt.pdf nvidia pdf   8-bit inference with tensorrt

 

https://cloud.tencent.com/developer/article/1435233 CNN模型 INT8 量化实现方式（一）

https://blog.csdn.net/u014644466/article/details/83278954 基于Caffe-Int8-Convert-Tools进行caffe模型转int8量化日常记录

https://github.com/lyk125/caffe-int8-convert-tools  Caffe-Int8-Convert-Tools

 

https://github.com/BUG1989/caffe-int8-convert-tools 

https://github.com/Tencent/ncnn/pull/749

 

 

"mobile inference"

https://blog.csdn.net/weixin_34038293/article/details/86752375

 

"TRT sorce code"

https://blog.csdn.net/bingbob/article/details/92802444

https://devblogs.nvidia.com/speed-up-inference-tensorrt/   How to Speed Up Deep Learning Inference Using TensorRT