Can Balioglu

> This is a great feature. Thank you @zzzwen! > I remember @pritamdamania did a lot of work to make sure we did not lose any gradient during backward with...

Thanks for the feedback @yifuwang! > While I'm not very familiar with the op dispatching mechanism, isn't `meta` device already similar in the way that only tensor's storage gets special...

Hi @WilliamTambellini, thanks for your feedback! CUDA Runtime offers an IPC mechanism for interprocess memory sharing. Using the `cudaIpcGetMemHandle()` and `cudaIpcOpenMemHandle()` functions you can pass a "handle" to a (non-managed)...

> One thing I realized from this comment, that we might need some APIs to store several tensors and some metadata together. For example supporting the sparse tensor coo/csr, which...

> Hi @cbalioglu we all agree that is nice feature to have. But it became hard to see the history of RFC changes and track conversations. Could you please create...

Hi @lamhoangtung, it is still on our roadmap, but we don't have an exact ETA yet. Do you have a particular use case in your mind?

Hi @c-heat16, it is in our backlog, but we don't have a date yet.

Yeah, we should have a 0.12.3 release and mark 0.12.2 as deprecated. Let me sync up internally.

Hey everyone, sorry for the delay in checking your messages! Thanks a lot for all your feedback so far! Hope the new version will address most of your concerns raised...

Hey folks, sorry for the late reply. We have been making good progress internally, but there had been some slowdowns as well which shifted our initial estimates by a few...