Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

TODOs before 0.15 release #7217

Closed
41 of 49 tasks
NicolasHug opened this issue Feb 10, 2023 · 1 comment
Closed
41 of 49 tasks

TODOs before 0.15 release #7217

NicolasHug opened this issue Feb 10, 2023 · 1 comment

Comments

@NicolasHug
Copy link
Member

NicolasHug commented Feb 10, 2023

This is a meta-issue to keep track of all the things we need to do before the branch cut / release. Core branch cut is planned for Monday 13 15th.

This superseeds #7092 and includes stuff unrelated to transformsV2. @pmeier @vfdev-5 please feel free to directly edit this issue with updates, and add anything I may have missed

Before branch cut - Friday 17th

Transforms V2 Stuff

Other stuff

If time allows

After branch cut, before release

Just after release

  • Port training references to V2? Would it require releasing new models?
@justinchuby
Copy link
Contributor

Linking #7056 for visibility

vfdev-5 added a commit to vfdev-5/vision that referenced this issue May 4, 2023
Description:
- Now that pytorch/pytorch#90771 is merged, let Resize() rely on interpolate()'s native uint8 handling instead of converting to and from float.

  - uint8 input is not casted to f32 for nearest mode and bilinear mode if the latter has AVX2.

Context: pytorch#7217

Benchmarks:
```
[----------- Resize cpu torch.uint8 InterpolationMode.NEAREST -----------]
                         |  resize v2  |  resize stable  |  resize nightly
1 threads: ---------------------------------------------------------------
      (3, 400, 400)      |      457    |        461      |        480
      (16, 3, 400, 400)  |     6870    |       6850      |      10100

Times are in microseconds (us).

[---------- Resize cpu torch.uint8 InterpolationMode.BILINEAR -----------]
                         |  resize v2  |  resize stable  |  resize nightly
1 threads: ---------------------------------------------------------------
      (3, 400, 400)      |      326    |        329      |        844
      (16, 3, 400, 400)  |     4380    |       4390      |      14800

Times are in microseconds (us).
```

[Source](https://gist.github.com/vfdev-5/a2e30ed50b5996807c9b09d5d33d8bc2)
vfdev-5 added a commit to vfdev-5/vision that referenced this issue May 9, 2023
Description:
- Now that pytorch/pytorch#90771 is merged, let Resize() rely on interpolate()'s native uint8 handling instead of converting to and from float.

  - uint8 input is not casted to f32 for nearest mode and bilinear mode if the latter has AVX2.

Context: pytorch#7217

Benchmarks:
```
[----------- Resize cpu torch.uint8 InterpolationMode.NEAREST -----------]
                         |  resize v2  |  resize stable  |  resize nightly
1 threads: ---------------------------------------------------------------
      (3, 400, 400)      |      457    |        461      |        480
      (16, 3, 400, 400)  |     6870    |       6850      |      10100

Times are in microseconds (us).

[---------- Resize cpu torch.uint8 InterpolationMode.BILINEAR -----------]
                         |  resize v2  |  resize stable  |  resize nightly
1 threads: ---------------------------------------------------------------
      (3, 400, 400)      |      326    |        329      |        844
      (16, 3, 400, 400)  |     4380    |       4390      |      14800

Times are in microseconds (us).
```

[Source](https://gist.github.com/vfdev-5/a2e30ed50b5996807c9b09d5d33d8bc2)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants