Add NaViT: Native Resolution Vision Transformer with Patch n' Pack (#9011)

Adds NaViT (monai.networks.nets.NaViT), a Vision Transformer that
removes the fixed-resolution constraint of standard ViT by packing
multiple variable-size images into a single sequence per batch element.

Key features:
- Patch n' Pack: multiple images concatenated into one sequence per
group, with a per-image attention mask preventing cross-image attention
- Factorised positional embeddings: separate learnable tables per
spatial axis, allowing generalisation to unseen resolutions
- Token dropout: configurable fraction of patch tokens dropped during
training (float or callable)
- Attention pooling: learned query attends over each image's tokens to
produce a fixed-size per-image representation
- QK normalisation: RMS normalisation on queries and keys (ViT-22B
style)
- 2D and 3D support: works for (C, H, W) and (C, H, W, D) inputs

Changes:
- monai/networks/nets/navit.py: new NaViT implementation
- monai/networks/nets/__init__.py: export NaViT
- tests/networks/nets/test_navit.py: 24 unit tests covering shape,
variable resolutions, token dropout, auto-grouping, gradient flow, ill
arguments, and forward validation
- docs/source/networks.rst: autoclass entry
- docs/source/whatsnew_1_5_2.md: feature description
- CHANGELOG.md: entry under Unreleased

Fixes # .

### Description

A few sentences describing the changes proposed in this pull request.

### Types of changes
<!--- Put an `x` in all the boxes that apply, and remove the not
applicable items -->
- [x] Non-breaking change (fix or new feature that would not break
existing functionality).
- [ ] Breaking change (fix or new feature that would cause existing
functionality to change).
- [x] New tests added to cover the changes.
- [ ] Integration tests passed locally by running `./runtests.sh -f -u
--net --coverage`.
- [x] Quick tests passed locally by running `./runtests.sh --quick
--unittests --disttests`.
- [x] In-line docstrings updated.
- [x] Documentation updated, tested `make html` command in the `docs/`
folder.

---------

Signed-off-by: Vikash Gupta <write2vikash@gmail.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Eric Kerfoot <17726042+ericspod@users.noreply.github.com>

Author: 3 people

CHANGELOG.md

@@ -5,6 +5,7 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 
 ## [Unreleased]
 ### Added
+* `NaViT` (`monai.networks.nets.NaViT`): Native Resolution Vision Transformer with Patch n' Pack, supporting variable-resolution 2D and 3D inputs. Implements factorized positional embeddings, token dropout, attention pooling, and QK normalization, based on ["Patch n' Pack: NaViT, a Vision Transformer for any Aspect Ratio and Resolution"](https://arxiv.org/abs/2307.06304).
 * `HyenaMixer`, `HyenaTransformerBlock`, and `DepthwiseFFTConv{2,3}d` in `monai.networks.blocks`: subquadratic O(N log N) alternatives to windowed self-attention, backed by the HyenaND operator from the optional `nvsubquadratic` package.
 * `HyenaNDUNETR` (`monai.networks.nets.HyenaNDUNETR`): thin `SwinUNETR` subclass with a `get_variant(name)` classmethod for the three Hyena variants (`HHHH`, `HAHA`, `HHAA`) from the NeurIPS 2026 paper "Native Multi-Dimensional Subquadratic Operators via Input Dependent Long Convolutions" (paper id 26539).
 * `SwinUNETR.use_hyena` and `SwinUNETR.hyena_stages` kwargs to thread HyenaND blocks through any subset of Swin stages. Default `use_hyena=False` preserves bit-identical forward behavior of the existing code path.

docs/source/networks.rst

@@ -662,6 +662,11 @@ Nets
 .. autoclass:: VarAutoEncoder
   :members:
 
+`NaViT`
+~~~~~~~
+.. autoclass:: NaViT
+  :members:
+
 `ViT`
 ~~~~~
 .. autoclass:: ViT

monai/networks/nets/__init__.py

@@ -75,6 +75,7 @@
     MedNextSmall,
 )
 from .milmodel import MILModel
+from .navit import NaViT
 from .netadapter import NetAdapter
 from .patchgan_discriminator import MultiScalePatchDiscriminator, PatchDiscriminator
 from .quicknat import Quicknat