Thanks @jwfromm and you're definitely right. This is the fastest way to integrate TVM functions into TensorFlow if we can not convert the whole model to TVM. This may be meaningful for TensorFlow developers if they can to try TVM and leverage the sub-graph optimizaition from TVM.
Actually this project is the TensorFlow custom op with TVM runtime. We originally develop in the standalone project https://github.com/tobegit3hub/tftvm . Since it depends on TVM and TensorFlow to compile, it is okay to be one of the TVM contrib libraries or maintain in the independent project. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/apache/incubator-tvm/issues/4464#issuecomment-563093299
