You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Caffe2 C++ API is undergoing changes since we are currently working on merging the caffe2 and pytorch backend. There might be breaking changes, but most of them should be easy to fix. For most updated version of caffe2::Tensor, please refer to: https://github.com/pytorch/pytorch/blob/master/caffe2/core/tensor.h
Note that this includes pure C++ Detectron inference and included new C++ operators we had to write. We'll PR back to pytorch whenever time allows and if there's interest.
@jerryzh168 I know it is a bit irrelevant, just asking for some info and an opinion:
Is caffe2 going to be deprecated in favour of ATen, Autograd?
For us who often have to use C++ should we keep working on caffe2 or switch to the above?
Is pytorch using Aten and Autograd in the background? I assume so, and I've familiarised my self with pytorch, so I'd prefer to have a similar approach when using C++
Many thanks and much appreciate any advice/opinion :-)
@alexge233 sorry for the late reply, we have merged TensorImpl in caffe2 and pytorch, right now caffe2::Tensor and aten::Tensor are thin wrappers around TensorImpl. To answer your question
There won't be distinction of caffe2/pytorch any more, they are the same thing, you can use existing caffe2 abstractions to write operators or use ATen abstractions, but they are being merged, for new operators probably better to follow ATen I think
It's the same thing, old caffe2 stuff should still be working, but if you want to write new operators, maybe better to put them in ATen folder
Activity
[-]Caffe2 API is ongoing changes[/-][+]Caffe2 API is undergoing changes[/+]rilesdg3 commentedon Nov 8, 2018
I forked this project and updated parts of it to reflect the Caffe2 API changes as of 10 22 2018. I was able to get the rnn tutorial to run.
All the necessary files to run only the rnn example can be found here
beniz commentedon Nov 8, 2018
For full C++ Caffe2 code working with pytorch/caffe2 1.0 release and the non templated tensors, see https://github.com/jolibrain/deepdetect/tree/master/src/backends/caffe2 and jolibrain/deepdetect#510
Note that this includes pure C++ Detectron inference and included new C++ operators we had to write. We'll PR back to pytorch whenever time allows and if there's interest.
jerryzh168 commentedon Nov 8, 2018
@rilesdg3 that's great! probably talk to author to merge?
jerryzh168 commentedon Nov 8, 2018
we have more API changes coming(e.g. method renaming), not just the template..
beniz commentedon Nov 9, 2018
Is there a public list of forthcoming changes ?
jerryzh168 commentedon Nov 9, 2018
@beniz I think there will be announcement when the two Tensors are merged.
rilesdg3 commentedon Nov 12, 2018
@jerryzh168 will do, I will keep an eye open for the changes
BIGBALLON commentedon Dec 21, 2018
Hi, @jerryzh168 .
Is there any new process about the merge of two tensors?
jerryzh168 commentedon Dec 21, 2018
We're still removing tech debts in our codebase.
alexge233 commentedon Mar 22, 2019
@jerryzh168 I know it is a bit irrelevant, just asking for some info and an opinion:
ATen,Autograd?AtenandAutogradin the background? I assume so, and I've familiarised my self with pytorch, so I'd prefer to have a similar approach when using C++Many thanks and much appreciate any advice/opinion :-)
jerryzh168 commentedon Oct 9, 2019
@alexge233 sorry for the late reply, we have merged
TensorImplin caffe2 and pytorch, right nowcaffe2::Tensorandaten::Tensorare thin wrappers aroundTensorImpl. To answer your questionATenI think