I have a pre-trained mobilenet and I want to put it into vta. But some errors occurred:
raise NNVMError(py_str(_LIB.NNGetLastError()))
nnvm._base.NNVMError: Failed loading Op stage0_conv0_fwd of type Convolution: [22:28:04] /home/varinic/git_project/tvm/nnvm/src/core/op.cc:74: Check failed: op != nullptr: Operator Convolution is not registered
It seemed the operators have to be registered in NNVM. Obviously the naming of the operators of the pre-trained module is arbitrary. So how can I solve this problem
More generally, how can I put an arbitrary neural network on vta ?
By the way, is there any open source implementation of mobilenet mentioned in the ‘A Hardware-Software Blueprint for Flexible Deep Learning Specialization’ ?