Skip to content
This repository has been archived by the owner on Oct 31, 2023. It is now read-only.

Fix dispatch breakage #555

Merged
merged 1 commit into from
Mar 11, 2019
Merged

Fix dispatch breakage #555

merged 1 commit into from
Mar 11, 2019

Conversation

vishwakftw
Copy link
Contributor

@fmassa This fixes the dispatch breakage introduced by pytorch/pytorch@3aeb780.

Closes pytorch/pytorch#17861

@facebook-github-bot
Copy link

Thank you for your pull request. We require contributors to sign our Contributor License Agreement, and yours has expired.

Before we can review or merge your code, we need you to email [email protected] with your details so we can update your status.

@fmassa
Copy link
Contributor

fmassa commented Mar 11, 2019

Thanks!

But I suppose that we need now to ask users to use PyTorch from source, or else it will break for those who are using PyTorch 1.0, or 1.0.1, right?

@vishwakftw
Copy link
Contributor Author

The installation instructions suggested to use the nightly build, which is why I sent in a PR.

@fmassa
Copy link
Contributor

fmassa commented Mar 11, 2019

That used to be the case, but we have since #510 changed it to install from a pinned version.

Something should be changed, here are a few possibilities:

  • change the install instructions to install from a nightly, and add a big warning there saying that this repo doesn't work with 1.0 or 1.0.1
  • revert this patch and keep this repo only working on a pinned version of PyTorch
  • find a better solution

@fmassa
Copy link
Contributor

fmassa commented Mar 11, 2019

As it turns out, #510 had a bugged install anyway, so I'll be reverting parts of it, and in the meantime I'll also make it work only on a nightly version

@vishwakftw
Copy link
Contributor Author

Oh I'm sorry I didn't see the installation snippet. The first line in INSTALL.md seems to be misleading.

@fmassa
Copy link
Contributor

fmassa commented Mar 11, 2019

@vishwakftw I'll fix the installation instructions, don't worry, they are bugged now anyway :-D

@vishwakftw
Copy link
Contributor Author

Thank you, and sorry about the confusion.

@fmassa
Copy link
Contributor

fmassa commented Mar 11, 2019

No worries, thanks for fixing this issue so quickly!

@kyamagu kyamagu mentioned this pull request Mar 20, 2019
miguelvr pushed a commit to miguelvr/maskrcnn-benchmark that referenced this pull request Mar 26, 2019
fmassa pushed a commit that referenced this pull request Mar 26, 2019
* Merge branch 'master' of /home/braincreator/projects/maskrcnn-benchmark with conflicts.

* rolls back the breaking AT dispatch changes (#555)

* revert accidental docker changes

* revert accidental docker changes (2)
eskjorg pushed a commit to eskjorg/maskrcnn-benchmark that referenced this pull request Mar 27, 2019
* Merge branch 'master' of /home/braincreator/projects/maskrcnn-benchmark with conflicts.

* rolls back the breaking AT dispatch changes (facebookresearch#555)

* revert accidental docker changes

* revert accidental docker changes (2)
@drcege
Copy link

drcege commented Jun 8, 2019

@vishwakftw I'll fix the installation instructions, don't worry, they are bugged now anyway :-D

@fmassa
The first line of Installation/Requirements (PyTorch 1.0 from a nightly release) is still misleading.
The nightly release is now PyTorch 1.2 (am I right?). And README also said the project is using PyTorch 1.0.

By the way, are Dockerfiles up to date?

Lyears pushed a commit to Lyears/maskrcnn-benchmark that referenced this pull request Jun 28, 2020
Lyears pushed a commit to Lyears/maskrcnn-benchmark that referenced this pull request Jun 28, 2020
* Merge branch 'master' of /home/braincreator/projects/maskrcnn-benchmark with conflicts.

* rolls back the breaking AT dispatch changes (facebookresearch#555)

* revert accidental docker changes

* revert accidental docker changes (2)
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

could not convert ‘Double’ from ‘c10::ScalarType’ to ‘<type error>’
6 participants