Add interpolate_like function to oneflow.nn.functional

[ShawnXuan]

This PR adds a new utility function interpolate_like to the oneflow.nn.functional module. It allows resizing an input tensor to match the spatial dimensions of a reference tensor using interpolation.

Changes:

• Added the interpolate_like function in interpolate.py.
• Exposed interpolate_like via functional/init.py.

[Copilot]

Pull Request Overview

This PR adds a new interpolate_like utility to resize an input tensor to match the spatial dimensions of a reference tensor and exposes it in the functional API.

• Implements interpolate_like in interpolate.py, wrapping the existing Interpolate module.
• Updates functional/__init__.py to export interpolate_like.

Reviewed Changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
python/oneflow/nn/modules/interpolate.py	Added interpolate_like function leveraging Interpolate(...).forward
python/oneflow/nn/functional/init.py	Imported and exposed interpolate_like in the functional namespace

Comments suppressed due to low confidence (1)

python/oneflow/nn/modules/interpolate.py:313

• No unit tests were added for interpolate_like. Consider adding tests covering different input ranks (3D, 4D, 5D) and modes to ensure correct behavior across all supported cases.

def interpolate_like(

[github-actions]

Code got formatted by CI. Please request CI again if you still want to have this PR merged. If the PR is from a forked repo, please download the patch files from the GitHub Actions web page and apply them locally.

[github-actions]

View latest API docs preview at: https://oneflow-staging.oss-cn-beijing.aliyuncs.com/docs/Oneflow-Inc/oneflow/pr/10644/

[github-actions]

Speed stats:

GPU Name: NVIDIA GeForce RTX 3080 Ti 

❌ OneFlow resnet50 time: 43.8ms (= 4375.1ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 57.4ms (= 5742.6ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.31 (= 57.4ms / 43.8ms)

OneFlow resnet50 time: 26.3ms (= 2632.5ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 37.5ms (= 3749.5ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.42 (= 37.5ms / 26.3ms)

OneFlow resnet50 time: 18.6ms (= 3714.7ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 36.6ms (= 7318.1ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.97 (= 36.6ms / 18.6ms)

OneFlow resnet50 time: 17.4ms (= 3477.3ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 29.9ms (= 5978.5ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.72 (= 29.9ms / 17.4ms)

OneFlow resnet50 time: 17.5ms (= 3490.4ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 29.7ms (= 5944.2ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.70 (= 29.7ms / 17.5ms)

OneFlow swin dataloader time: 0.199s (= 39.753s / 200, num_workers=1)
PyTorch swin dataloader time: 0.128s (= 25.686s / 200, num_workers=1)
Relative speed: 0.646 (= 0.128s / 0.199s)

OneFlow swin dataloader time: 0.055s (= 10.976s / 200, num_workers=4)
PyTorch swin dataloader time: 0.033s (= 6.563s / 200, num_workers=4)
Relative speed: 0.598 (= 0.033s / 0.055s)

OneFlow swin dataloader time: 0.031s (= 6.125s / 200, num_workers=8)
PyTorch swin dataloader time: 0.016s (= 3.288s / 200, num_workers=8)
Relative speed: 0.537 (= 0.016s / 0.031s)

❌ OneFlow resnet50 time: 49.2ms (= 4919.7ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 66.0ms (= 6595.4ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.34 (= 66.0ms / 49.2ms)

OneFlow resnet50 time: 37.4ms (= 3742.7ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 47.1ms (= 4713.0ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.26 (= 47.1ms / 37.4ms)

OneFlow resnet50 time: 27.7ms (= 5547.1ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 40.2ms (= 8035.0ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.45 (= 40.2ms / 27.7ms)

OneFlow resnet50 time: 25.3ms (= 5051.9ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 38.9ms (= 7783.4ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.54 (= 38.9ms / 25.3ms)

OneFlow resnet50 time: 24.8ms (= 4951.2ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 35.8ms (= 7164.6ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.45 (= 35.8ms / 24.8ms)

[github-actions]

View latest API docs preview at: https://oneflow-staging.oss-cn-beijing.aliyuncs.com/docs/Oneflow-Inc/oneflow/pr/10644/

[github-actions]

Speed stats:

GPU Name: NVIDIA GeForce RTX 3080 Ti 

❌ OneFlow resnet50 time: 43.4ms (= 4342.8ms / 100, input_shape=[16, 3, 224, 224])
PyTorch resnet50 time: 58.0ms (= 5799.6ms / 100, input_shape=[16, 3, 224, 224])
✔️ Relative speed: 1.34 (= 58.0ms / 43.4ms)

OneFlow resnet50 time: 26.5ms (= 2649.4ms / 100, input_shape=[8, 3, 224, 224])
PyTorch resnet50 time: 37.8ms (= 3779.0ms / 100, input_shape=[8, 3, 224, 224])
✔️ Relative speed: 1.43 (= 37.8ms / 26.5ms)

OneFlow resnet50 time: 19.2ms (= 3835.5ms / 200, input_shape=[4, 3, 224, 224])
PyTorch resnet50 time: 36.1ms (= 7223.0ms / 200, input_shape=[4, 3, 224, 224])
✔️ Relative speed: 1.88 (= 36.1ms / 19.2ms)

OneFlow resnet50 time: 17.7ms (= 3534.3ms / 200, input_shape=[2, 3, 224, 224])
PyTorch resnet50 time: 30.6ms (= 6118.1ms / 200, input_shape=[2, 3, 224, 224])
✔️ Relative speed: 1.73 (= 30.6ms / 17.7ms)

OneFlow resnet50 time: 17.3ms (= 3466.2ms / 200, input_shape=[1, 3, 224, 224])
PyTorch resnet50 time: 31.5ms (= 6298.9ms / 200, input_shape=[1, 3, 224, 224])
✔️ Relative speed: 1.82 (= 31.5ms / 17.3ms)

OneFlow swin dataloader time: 0.201s (= 40.298s / 200, num_workers=1)
PyTorch swin dataloader time: 0.136s (= 27.172s / 200, num_workers=1)
Relative speed: 0.674 (= 0.136s / 0.201s)

OneFlow swin dataloader time: 0.055s (= 11.035s / 200, num_workers=4)
PyTorch swin dataloader time: 0.034s (= 6.766s / 200, num_workers=4)
Relative speed: 0.613 (= 0.034s / 0.055s)

OneFlow swin dataloader time: 0.031s (= 6.159s / 200, num_workers=8)
PyTorch swin dataloader time: 0.017s (= 3.415s / 200, num_workers=8)
Relative speed: 0.554 (= 0.017s / 0.031s)

❌ OneFlow resnet50 time: 49.2ms (= 4920.8ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 68.0ms (= 6804.8ms / 100, input_shape=[16, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.38 (= 68.0ms / 49.2ms)

OneFlow resnet50 time: 37.2ms (= 3722.7ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 45.8ms (= 4577.4ms / 100, input_shape=[8, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.23 (= 45.8ms / 37.2ms)

OneFlow resnet50 time: 28.5ms (= 5708.0ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 42.3ms (= 8459.5ms / 200, input_shape=[4, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.48 (= 42.3ms / 28.5ms)

OneFlow resnet50 time: 25.4ms (= 5075.5ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 40.2ms (= 8049.3ms / 200, input_shape=[2, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.59 (= 40.2ms / 25.4ms)

OneFlow resnet50 time: 25.6ms (= 5116.3ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
PyTorch resnet50 time: 38.1ms (= 7620.0ms / 200, input_shape=[1, 3, 224, 224], ddp, world size=2)
✔️ Relative speed: 1.49 (= 38.1ms / 25.6ms)