Hi, I have seen your new paper " A Hardware-Software Blueprint for Flexible Deep Learning Specialization"
(Sorry, I can’t upload the image)
Fig. 8 of the evaluation part shows you have succeeded to run many models like ResNet-34, ResNet-50, DCGAN.
So I searched the github repo used for this paper.
Finally I found tmoreau89’s forked repo
But I can’t found VTA FPGA bitstream v0.0.4 and have no idea how to convert nnvm API code to relay API and can’t find what relay APIs are supported
Actually, I want to execute InceptionV3 with VTA.
So my question is when is the release date that supports Relay for VTA
and includes bitpacking/quantization API for Relay
Or can you provide a guideline or some hints to run InceptionV3 model in VTA?
I’ve already tried to build InceptionV3 for VTA but failed…
I couldn’t find int8 quantized InceptionV3 model and
didn’t know how to bitpack nnvm/relay of InceptionV3
I am very grateful for your dedication.
and wish for the success of TVM.