optima2005 opened a new pull request #4400: implement conv3d op URL: https://github.com/apache/incubator-tvm/pull/4400 This is a start attempt to implement #4009 The implemation in this PR is a very basic version of conv3d( NCDHW layout only). I am proposing this version to confirm that I am in the correct direction. And if so, I think I can go on or others can keep working it from this on. I can run through the below testing on a x86 cpu and a nvidia gpu server, both linux OS. TVM_FFI=ctypes python -m pytest -v tests/python/relay/test_op_level2.py -k test_conv3d_run TVM_FFI=ctypes python -m pytest -v tests/python/contrib/test_cudnn.py -k test_conv3d
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
