releasing 1.6.2 [rebase & merge]

[Borda]

🚀 Added

• RFDETR.predict(shape=...) — pass an explicit (height, width) tuple to run inference at a non-square resolution, matching the resolution used when exporting the model. Both dimensions must be positive integers divisible by 14. (feat(predict): add shape parameter for non-square inference #866)

detections = model.predict("image.jpg", shape=(480, 640))

🌱 Changed

• ModelConfig.device and RFDETR.train(device=...) now accept torch.device objects and indexed device strings ("cuda:0", "cuda:1"). Existing string values ("cpu", "cuda") are unchanged. RFDETR.train() warns when a valid but unmapped device type is passed to PyTorch Lightning auto-detection. (Accept torch.device and indexed device strings in ModelConfig and RFDETR.train() #872)

from rfdetr import RFDETRSmall
from torch import device

model = RFDETRSmall(...)

model.train(..., device=device("cuda:1"))
model.train(..., device="cuda:0")

🔧 Fixed

• Fixed ONNX export ignoring an explicit patch_size argument: export() and predict() now resolve patch_size from model_config by default, validate it strictly (must be a positive integer, not bool), and enforce that (H, W) dimensions are divisible by patch_size × num_windows. (fix: ONNX export shape ignored; validate divisibility by patch_size #876)
• Fixed ONNX export for models traced with dynamic batch dimensions — torch.full is now used for Python-int spatial dims to avoid H_.expand(N_) tracer failures. (Fix ONNX export for dynamic batch dimensions #871)

---

🏆 Contributors

Welcome to our new contributors, and thank you to everyone who helped with this release:

• zhaoshuo (@zhaoshuo1223) — ONNX export shape validation and patch_size fixes
• Sven Goluza (@svengoluza) — ONNX export dynamic batch fix
• Jirka Borovec (@Borda) (LinkedIn) — shape inference, torch.device support, release coordination

---

Full changelog: 1.6.1...1.6.2

[codecov]

Codecov Report

❌ Patch coverage is 98.03922% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 76%. Comparing base (bddf21f) to head (ce998f7).
⚠️ Report is 6 commits behind head on release/stable.

Additional details and impacted files

@@              Coverage Diff              @@
##           release/stable   #884   +/-   ##
=============================================
  Coverage              76%    76%           
=============================================
  Files                  92     92           
  Lines                7158   7241   +83     
=============================================
+ Hits                 5406   5489   +83     
  Misses               1752   1752           

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[Copilot]

Pull request overview

Release prep for rfdetr==1.6.2, expanding inference/export shape handling, improving device selection for training, and fixing ONNX export edge-cases (dynamic batch + patch/window divisibility).

Changes:

• Add/enable RFDETR.predict(shape=(H, W)) with strict shape validation and correct non-square resizing behavior.
• Accept torch.device and indexed device strings (e.g. "cuda:1") in ModelConfig.device and map RFDETR.train(device=...) to PTL accelerator/devices.
• Fix ONNX export correctness: dynamic-batch dynamic_axes, patch/window divisibility validation, and tracing robustness for dynamic batch dimensions.

Reviewed changes

Copilot reviewed 14 out of 14 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
src/rfdetr/detr.py	Adds shared shape/patch-size validation helpers; implements predict(shape=...), patch/window divisibility checks, dynamic batch export, and improved device-to-PTL mapping in train().
src/rfdetr/config.py	Normalizes device via a Pydantic validator to support torch.device and indexed device strings.
src/rfdetr/export/main.py	Fixes rectangular Resize((H, W)) and forwards dynamic_axes when dynamic_batch is set.
src/rfdetr/models/transformer.py	Adjusts spatial shape handling to better support ONNX tracing and avoid dynamic-batch tracer failures.
src/rfdetr/models/backbone/projector.py	Fixes LayerNorm to use self.normalized_shape for traceability.
tests/models/test_predict.py	Adds regression coverage for predict(shape=...) resizing and for patch/window-based shape validation.
tests/models/test_validate_shape_dims.py	New focused unit tests for _validate_shape_dims and _resolve_patch_size.
tests/models/test_export.py	Adds tests for dynamic batch export forwarding and patch/window shape validation; adds regression for rectangular infer-image creation.
tests/models/test_transformer.py	Regression test ensuring gen_encoder_output_proposals handles Python-int spatial shapes when masks=None.
tests/models/test_config.py	Adds tests for ModelConfig.device accepting indexed strings and torch.device.
tests/training/test_detr_shim.py	Updates training shim tests to reflect new device forwarding/mapping behavior and warnings/errors.
CHANGELOG.md	Adds 1.6.2 release notes.
pyproject.toml	Bumps package version to 1.6.2.
docs/tutorials/index.md	Adds LitServe deployment tutorial link to tutorials index.