add support for type_as/_to_copy#77971
add support for type_as/_to_copy#77971eellison wants to merge 15 commits intogh/eellison/294/basefrom
Conversation
[ghstack-poisoned]
🔗 Helpful links
✅ No Failures (0 Pending)As of commit 88a6bc9 (more details on the Dr. CI page): Expand to see more💚 💚 Looks good so far! There are no failures yet. 💚 💚 This comment was automatically generated by Dr. CI (expand for details).Please report bugs/suggestions to the (internal) Dr. CI Users group. |
Without this I would get strange `invalid arguments` in CUDA call. [ghstack-poisoned]
Without this I would get strange `invalid arguments` in CUDA call. [ghstack-poisoned]
Without this I would get strange `invalid arguments` in CUDA call. [ghstack-poisoned]
Without this I would get strange `invalid arguments` in CUDA call. [ghstack-poisoned]
| # Run the original computation | ||
|
|
||
| # _to_copy fails when run with FakeTensors to cuda device | ||
| # TODO: debug |
There was a problem hiding this comment.
You're talking about the meta implementation here right? It makes sense, since we can't actually produce the cuda data (but for fake tensor that's OK, because the fake cuda device doesn't have data anyway)
Without this I would get strange `invalid arguments` in CUDA call. [ghstack-poisoned]
Without this I would get strange `invalid arguments` in CUDA call. [ghstack-poisoned]
Without this I would get strange `invalid arguments` in CUDA call. [ghstack-poisoned]
Without this I would get strange `invalid arguments` in CUDA call. [ghstack-poisoned]
|
@eellison has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator. |
Without this I would get strange `invalid arguments` in CUDA call. Differential Revision: [D36618463](https://our.internmc.facebook.com/intern/diff/D36618463) [ghstack-poisoned]
|
@eellison has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator. |
Without this I would get strange `invalid arguments` in CUDA call. Differential Revision: [D36618463](https://our.internmc.facebook.com/intern/diff/D36618463) [ghstack-poisoned]
Without this I would get strange `invalid arguments` in CUDA call. Differential Revision: [D36618463](https://our.internmc.facebook.com/intern/diff/D36618463) [ghstack-poisoned]
Without this I would get strange `invalid arguments` in CUDA call. Differential Revision: [D36618463](https://our.internmc.facebook.com/intern/diff/D36618463) [ghstack-poisoned]
Without this I would get strange `invalid arguments` in CUDA call. Differential Revision: [D36618463](https://our.internmc.facebook.com/intern/diff/D36618463) [ghstack-poisoned]
Without this I would get strange `invalid arguments` in CUDA call. [ghstack-poisoned]
|
@pytorchbot merge this please |
|
Hey @eellison. |
Summary: Pull Request resolved: #77971 Approved by: https://github.com/ezyang Test Plan: contbuild & OSS CI, see https://hud.pytorch.org/commit/pytorch/pytorch/4c18f362a9d88e664fa1ced1ac77e1c6ccac960e Reviewed By: seemethere Differential Revision: D36784772 Pulled By: seemethere fbshipit-source-id: 76e3be49c6a723ed7094436adf89a01df065c909
Stack from ghstack (oldest at bottom):
_make_subclass#77970Without this I would get strange
invalid argumentsin CUDA call.