pnnx convert interpolate nearest-exact

[nihui]

No description provided.

[tencent-adm]

Thank you for your submission, we really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[github-actions]

The binary size change of libncnn.so (bytes)

architecture	base size	pr size	difference
x86_64	15208240	15208240	0 😘
armhf	6210540	6210540	0 😘
aarch64	9523592	9523592	0 😘

[Copilot]

Pull Request Overview

This pull request adds support for the nearest-exact interpolation mode in the PNNX converter. The nearest-exact mode provides an alternative nearest-neighbor interpolation algorithm that differs from the standard nearest mode in how it calculates pixel indices.

Key changes:

• Added nearest-exact mode handling in PNNX conversion passes for both F.interpolate and nn.Upsample
• Updated test files with version-gated test cases for PyTorch 1.11+ and 2.9+ (ONNX)
• Modified float/double formatting in utils.cpp to use %g format specifier

Reviewed Changes

Copilot reviewed 9 out of 9 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
tools/pnnx/tests/test_nn_Upsample.py	Added version-gated test cases for nearest-exact mode in 1D, 2D, and 3D upsampling
tools/pnnx/tests/test_F_interpolate.py	Added version-gated test cases for nearest-exact mode in functional interpolation
tools/pnnx/tests/onnx/test_nn_Upsample.py	Added version-gated ONNX test cases for nearest-exact mode with PyTorch 2.9+
tools/pnnx/tests/onnx/test_F_interpolate.py	Added version-gated ONNX test cases for nearest-exact mode with PyTorch 2.9+
tools/pnnx/src/utils.cpp	Changed float/double formatting from fixed precision to %g format
tools/pnnx/src/pass_ncnn/nn_Upsample.cpp	Added handling for nearest-exact mode with stderr warnings
tools/pnnx/src/pass_ncnn/F_interpolate.cpp	Added handling for nearest-exact mode with stderr warnings
tools/pnnx/src/pass_level2/F_interpolate.cpp	Added graph rewriter passes for nearest-exact interpolation and ONNX conversions
tools/pnnx/src/pass_level1/nn_Upsample.cpp	Added detection and parameter setting for nearest-exact upsampling operations

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.