search

Zhen-Dong / HAWQ

Quantization library for PyTorch. Support low-precision and mixed-precision quantization, with hardware implementation through TVM.
MIT License
410 stars 83 forks source link

TensorRT model #16

Open Cat11438 opened 3 years ago

Cat11438 commented 3 years ago

Does HAWQ supports conversion to TensorRT models? Thanks

zachzzc commented 3 years ago

No, we don't have a tensorRT model now. Since we separate the training part in Pytorch and inference part in TVM, it is possible to build a tensorRT model to read the trained parameters and use it. We would really appreciate it if you want to try it out.