Geti v3.1.0

🚀 Geti™ v3.1.0 - new models, improved pipeline reliability and much more!

Geti™ 3.1 builds on the same lightweight edge-native architecture of Geti™ 3.0, extending it with several new SOTA models while also improving dataset navigation, inference pipeline monitoring, quantization and other core features.

This release also extends the documentation and it updates some key dependencies (including PyTorch and OpenVINO™) for better performance, stability and security.

To upgrade from an existing Geti™ 3.0 deployment to 3.1, follow the upgrade guide.

🔷 Highlights

• Broader model zoo - enabled the new EdgeCrafter (ECDet) architecture and more RF-DETR variants, giving users more flexibility to choose the best model for their accuracy and speed requirements.
• Filter datasets by subset - the Dataset page and the annotator now offer the possibility to filter media by a specific subset (training, validation or testing).
• Client-provided video files as pipeline sources - upload a video straight from your local filesystem when configuring a source for the inference pipeline, without needing the file to be available in the server beforehand.
• Pipeline observability – it is now easier to monitor the status of the inference pipeline, by means of a health indicator (with fine-grained status info reported for each component) and a one-click 'Test connection' button for sources and sinks.
• Increased UI responsiveness – thanks to HTTP/2, concurrent API requests are handled more efficiently, improving the general responsiveness of the application.

🧠 Models and training

• EdgeCrafter for object detection - the new ECDet (a.k.a. EdgeCrafter) model is now available and fully supported for training, quantization and inference in Geti™. ECDet is available in different size variants: S, M, L, and X.
• New size variants of RF-DETR were added for object detection (N) and instance segmentation (N, 2XL), in addition to the ones already present since 3.0.
• Tiling - All detection and instance segmentation models now support tiling for training and inference, closing the gap for some models that didn't already support it (e.g. D-FINE, RF-DETR, ...)
• Iteration limit for accuracy-aware quantization - a new parameter is available to optionally cap the number of iterations during accuracy-aware quantization, preventing it from running indefinitely.

🗂️ Projects, datasets and annotation

• Project filtering - projects can be searched by name and filtered by task type.
• Pinned active project - any project with a running inference pipeline is now pinned to the top of the list and tagged as "Active" badge for better visibility.
• Annotator media filtering - all media filters are now also available inside the annotator, making it much faster to navigate between images and videos without leaving the annotator.
• Subset filtering - a new filter is available to select media assigned to a specific subset (training, validation, testing or unassigned).

🚀 Deployment and inference pipeline

• Video upload for source creation - it is now possible to create a new source of type 'Video file' by uploading a video from your local filesystem. Previously, the video needed to be already available in the server beforehand, which is not a practical setup for many users.
• Test connection - sources and sinks can be validated before use through the new 'Test connection', with an inline connection-status badge reflecting the result.
• Pipeline health monitor - a new indicator in the 'Pipeline' page aggregates status information about the source, model and sink in real-time, providing informative diagnostic info when something goes wrong in one of these components.

⚙️ Platform, API, and infrastructure

• Upgrade mechanism - in-place migration from a 3.x release to a newer version is now fully supported, automated and documented for all installation modes: Docker, MSIX, Tauri desktop app and scripts (install.sh / install.ps1). All projects and models are preserved, all data will be migrated automatically to the newer version as necessary. As an extra safety mechanism, the database and the application files are backed up before migration, and an informative error message should be displayed in case of problems.
• HTTP/2 REST API - the communication between the UI and the server now leverages HTTP/2, which is more efficient to handle concurrent requests and it enables more flexibility (notably, it allows having more than 7 jobs queued at once). The necessary TLS certificates are generated automatically on installation, regardless of the chosen mode. Note that the URL to access the UI now starts with https:// instead of http://.
• Faster startup - several imports were optimized, noticeably reducing the time to launch the application.
• Pretrained weights provisioning - for some model architectures, it is now possible to automatically download the pre-trained weights from a mirror location if the primary one becomes unavailable, for better reliability. Moreover, for users who need to install Geti in an air-gapped environment, a technical path is available and documented.
• Several dependencies were updated, most notably torch (2.12.1), openvino (2026.3.0) and rfdetr (1.8.0).

🐞 Bug fixes

• Resolved an issue causing training to sometimes fail or hang when tiling is enabled.
• Resolved an issue where trained models would be exported to OpenVINO IR with a static batch dimension of 1, causing problems on batch inference.
• Resolved various issues where Intel® XPU devices would occasionally not be properly recognized by the platform.
• Resolved two issues causing D-FINE models to crash or output poor quality predictions on some Intel® XPU devices.
• Resolved an issue where Mask-RCNN EfficientNet-B2 would crash on inference on some Intel® XPU devices.
• Resolved several issues and race conditions in the inference pipeline that caused instability and silent failures.
• Resolved some issues with the Auto-segmentation annotation tool (SAM), that would often timeout on Tauri / MSIX.
• Resolved various issues with the download of pre-trained weights, which is now more robust thanks to a retry mechanism and mirrors.
• Resolved an issue that prevented the import of some datasets with 16-bit images.
• Resolved an issue that caused errors when importing datasets with out-of-bounds coordinates; the app now sanitizes these values.
• Resolved a visualization issue with intensity mapping configurable parameters, where the displayed decimal precision was insufficient for certain use cases.
• Resolved a formatting issue with training logs.

Download this release

Windows Installer (run Geti as a native Windows application, with prebuilt images for Intel® XPU, NVIDIA® CUDA, and CPU-only environments):
Download CPU-only version installer.
Download Intel® XPU version installer.
Download Nvidia® CUDA version installer

Docker Images

ghcr.io/open-edge-platform/geti-cpu:3.1.0
ghcr.io/open-edge-platform/geti-xpu:3.1.0
ghcr.io/open-edge-platform/geti-cuda:3.1.0

Refer to the installation and upgrade documentation for more details, the guides provide instructions for advanced deployment scenarios too.

---

Geti Library (getitune) v0.2.0

getitune is the low-code transfer-learning library that powers Geti™ training, export, optimization, and inference. This release focuses on widening the model zoo, generalizing the Ultralytics backend to new vision tasks, and making the training and export contracts more predictable.

The getitune library is available as a Pypi package (pip install getitune) or it can be installed from source.

New models

• EdgeCrafter detection model - a new powerful model called EdgeCrafter (a.k.a. ECDet) is now available for object detection, in different sizes (S / M / L / X). This modern model architecture is built on top of a ViT backbone, a RT-DETR-based encoder and a DETR-style decoder. The addition of ECDet to getitune enables users to train this model, quantize it and use it for inference on Intel® XPU.
• More RF-DETR recipes - more size variants of RF-DETR were enabled for object detection (N) and instance segmentation (X, 2XL)

API changes

• Simplified engine creation - engine factory create_engine(...) now accepts either a recipe path or a plain model name (for example create_engine(model="yolox_s", data="path/to/data/root")). It is now unified across all our supported backends.
• Configurable pretrained weights - when instantiating any model (derived from LightningModel), the new parameters pretrained and pretrained_weights allow to optionally supply a checkpoint file to initialize the weights.
• **Ite...

Geti v3.0.0

🚀 Geti™ v3.0.0 — AI development moves to the edge

Geti 3.0 marks a fundamental achievement in the evolution of vision AI development. Geti is now a lightweight application that runs directly on edge devices and AI PC hardware, while still remaining compatible with traditional cloud and cluster‑based infrastructure.
With Geti v3 you can fine‑tune models and run inference directly on edge and client hardware — including Intel® Panther Lake and Arc™ Battlemage (B‑series) GPUs — keeping your data, training, and deployment on the same machine.

The result is a faster, simpler, and edge-native way to go from data to a deployed model: on a laptop, a workstation, or an edge device.

🔷 Major features and improvements

A whole new installation experience:

• Native Windows app — run Geti directly on Windows via a simple MSIX installer.

• Single Docker image — run Geti from one container on Linux or Windows, while maintaining compatibility with Kubernetes deployments.

State‑of‑the‑art model architectures

• RF‑DETR S / M / L (Roboflow) — a real‑time DETR transformer on a DINOv2 backbone, SOTA accuracy/latency on COCO, for detection and instance segmentation.

• DINOv3 DETR S / M / L — Meta's DINOv3 self‑supervised backbone combined with DEIMv2 + DETR for high‑quality detection with less labeled data.

A new Ultralytics integration introduces YOLO26 — the latest NMS‑free, edge‑optimized models (Nano / Small / Medium) for object detection and instance segmentation. The integration covers the full model lifecycle: training, inference, quantization, and OpenVINO™ model export.

Dramatically lower hardware requirements: Geti v3 no longer depends on Kubernetes as the default architecture, the overall footprint is significantly reduced requiring fewer CPU resources and less memory (the minimum recommended configuration is 8 CPU threads, 16 GB RAM, and 40 GB of free disk space)[VJ1.1][YV1.2]. Smaller models can be trained on CPU with modest memory usage, while a GPU is recommended for larger models.

Apache 2.0 license: Geti 3.0 code is now under the permissive Apache 2.0 license, simplifying adoption, integration, and redistribution.

More features:

• Dataset versioning - track how datasets evolve as images and annotations are added, with the ability to explore revisions, link models to a specific dataset state, and train on any prior revision.

• Model versioning / lineage - track model lineage with links to parent revisions and weights, view exact training hyperparameters, and fine-tune from any previous model version.

• Configurable training devices - select from available hardware (CPU, GPU), including specific GPUs in multi-device setups.

• 16‑bit media support - train and infer on 16-bit single-channel images and video with configurable intensity range preprocessing (not supported for YOLOX‑S/M/L).

• Decoupled edge-ready pipelines - build flexible inference pipelines where data sources (e.g. cameras, files, streams) and output targets can be swapped or reconfigured without retraining models, enabling easier integration and faster iteration across different deployment scenarios.

• Pipeline export for edge deployment - package complete pipelines, including model, preprocessing, and configuration, into OpenVINO™‑optimized bundles ready for deployment on edge and Intel®‑based hardware.

• Integrated deployment & inference — a new Inference page builds custom pipelines (source → model → sink) to deploy models inside Geti and monitor predictions in real time. Sources include USB/IP cameras and video files; optional sinks include MQTT, webhook, and folder.

• Improved evaluation metrics - expanded training-time and evaluation metrics, including mAP at multiple thresholds for detection and segmentation tasks, providing deeper insight into model performance.

• Label filtering for dataset import/export - selectively include or exclude labels during dataset export and import, automatically filtering out associated annotations.

• Native dataset format (Geti) - export and import datasets in a Geti‑optimized format, designed for improved performance on large datasets while preserving full metadata.

• Manual subset assignment - explicitly assign images and video frames to training, validation, or test subsets in the annotator, with safeguards to prevent data leakage and automatic assignment for unassigned media.

• Real‑time training logs - monitor training logs live during job execution and access them later via the UI for improved observability and troubleshooting.

• Demo scripts for inference - exported ONNX and OpenVINO™ model packages include example scripts to run inference on sample images, simplifying validation and onboarding.

• Bulk annotation for classification - assign one or more labels to multiple images or video frames at once, significantly speeding up annotation workflows.

• Accuracy‑aware quantization - define acceptable accuracy drop during INT8 model optimization, enabling better trade‑offs between inference performance and accuracy.

Some features from earlier versions are no longer included in v3 or have been reworked:

If you rely on any of the capabilities below or are exploring alternatives, we encourage you to join the conversation in our GitHub Discussions. Your feedback helps shape the roadmap, and the community may also share practical workarounds or migration strategies.

• Anomaly detection is now a standalone product: Anomalib Studio repo.

• Task chaining, hierarchical classification, and rotated detection are no longer supported as dedicated workflows. In some cases, alternatives such as single-task workflows or instance segmentation may be used, but these are not full equivalents; migration should be evaluated accordingly.

• Semantic segmentation, Keypoint detection, XAI (Explainable AI visualizations) - these capabilities are not included in Geti v3.

• Workspaces & user management, API keys / tokens - multi-user access control and API-based authentication features are not part of Geti v3

• Separate testing datasets - explicit support for dedicated test dataset splits is no longer available; dataset handling is simplified within the unified dataset workflow. Models can now be evaluated in real time or with an image folder as part of the inference pipeline.

• Project import/export is not supported in Geti v3. Users can still import/export their annotated datasets to ensure continuity.

Documentation:

• The REST API has been redesigned — see the updated specification; model export and deployment have been simplified and streamlined, with improved performance achieved by using the Model API instead of the SDK

• Migration notes: models trained in 2.x must be retrained in Geti v3. This can be achieved by exporting the dataset from Geti 2.x and importing it to Geti v3. Projects with multiple datasets require each dataset to be exported and imported separately — please follow the migration guidance.

Download this release

Windows Installer (run Geti as a native Windows application, with prebuilt images for Intel® XPU, NVIDIA® CUDA, and CPU-only environments):
Download CPU-only version installer.
Download Intel® XPU version installer.
Download Nvidia® CUDA version installer

Docker Images
ghcr.io/open-edge-platform/geti-cpu:3.0.0
ghcr.io/open-edge-platform/geti-xpu:3.0.0
ghcr.io/open-edge-platform/geti-cuda:3.0.0

Install natively with Ultralytics YOLO26 models (the latest NMS‑free, edge‑optimized models (Nano / Small / Medium) for object detection and instance segmentation. The integration covers the full model lifecycle: training, inference, quantization, and OpenVINO™ model export.

• Linux / WSL2:

curl -fsSL https://raw.githubusercontent.com/open-edge-platform/geti/develop/install.sh | bash

• Windows (PowerShell):

irm https://raw.githubusercontent.com/open-edge-platform/geti/develop/install.ps1 | iex

Geti Library (geti-tune) Release Notes (Formerly known as OTX)

We are pleased to announce the first public release of geti-tune, a package evolved from the OTX — OpenVINO™ Training Extensions — library. The library has been renamed and enhanced to support streamlined model fine-tuning workflows.
geti-tune provides an end-to-end stack for AI model development and optimization, enabling efficient adaptation of pretrained models to custom datasets and tasks.

The library includes several key modules that enable:

• Multi-task support for classification, object detection, rotated detection, instance segme...

2.6.0

What's Changed

• Update demo requirements by @sovrasov in #4421
• Cleanup Geti task templates for anomaly task by @rajeshgangireddy in #4420
• Restore Engine by @ashwinvaidya17 in #4430
• [Cherry-pick from release/2.5] Refactor benchmark script logic (#4435) by @eugene123tw in #4442
• Provide workarounds for the XPU training by @kprokofi in #4441
• DEIM-DFine by @eugene123tw in #4446
• Update pandas requirement from <2.3 to <2.4 by @dependabot[bot] in #4427
• Bump the github-actions-dependency group with 10 updates by @dependabot[bot] in #4428
• Bump the github-actions-dependency group with 2 updates by @dependabot[bot] in #4478
• Add DoRA support for ViT classification model. by @gyuilLim in #4466
• Bump tj-actions/changed-files from 4662f28b04a04c9b3c760178bc00690771ba6fe1 to a0370f61698fcac830a08949da9fdf96ea0f3ab7 in the github-actions-dependency group by @dependabot[bot] in #4481
• Update Python version requirements and workflows to support Python 3.11 and 3.12 by @eugene123tw in #4479
• Minor updates in docs by @sovrasov in #4482
• Bump github/codeql-action from 3.29.3 to 3.29.4 in the github-actions-dependency group by @dependabot[bot] in #4483
• Disable tox auto upgrade by @sovrasov in #4484
• Bump tj-actions/changed-files from a0370f61698fcac830a08949da9fdf96ea0f3ab7 to db8d0bfea5a44e51abd5dc1454386c668ae901f9 in the github-actions-dependency group by @dependabot[bot] in #4491
• Update tox uv installer by @eugene123tw in #4503
• Add support for transforming masks and polygons in RandomAffine by @eugene123tw in #4472
• Bump the github-actions-dependency group across 1 directory with 3 updates by @dependabot[bot] in #4507
• Enable PEFT support for h-label & multi-label classification by @gyuilLim in #4509
• Merge release 2.5 branch to develop by @kprokofi in #4501
• Geti tune: fix precommit and CI workflows by @leoll2 in #4571
• Use FastAPI DI instead of Singleton pattern for services by @itallix in #4574
• apply copyright header to all .py, .ts, .tsx and .js files by @leoll2 in #4578
• Merge geti-tune to develop by @leoll2 in #4581
• Fix compatibility with old anomaly models by @kprokofi in #4584
• Add DEIM manifests + Upgrade torch by @kprokofi in #4580
• Add npm script to update api spec by @jpggvilaca in #4583
• Disable WebRTC temporarily by @jpggvilaca in #4585
• refactor: Use toast from @geti/ui by @dwesolow in #4588
• Update GH worker version by @ivanzati in #4523
• Cleanup the house by @jpggvilaca in #4591
• Bump slab from 0.4.10 to 0.4.11 in /ui/src-tauri by @dependabot[bot] in #4582
• More cleanup by @jpggvilaca in #4592
• Support output to webhook by @itallix in #4598
• Add edit bounding box tool & friends by @jpggvilaca in #4593
• 4564 - pipeline list components by @camiloHimura in #4586
• Update landing page (creation) & rename pipeline by @jpggvilaca in #4601
• WebRTC improvements by @jpggvilaca in #4603
• Add models and labels routes by @jpggvilaca in #4604
• Add documentation for PEFT (LoRA & DoRA) by @gyuilLim in #4596
• Add input source modal by @jpggvilaca in #4615
• 4558 - error handling by @camiloHimura in #4600
• Fix pre_merge action by @ivanzati in #4609
• Add legacy templates for DEIM model by @kprokofi in #4608
• Move OTX code to lib folder by @warrkan in #4614
• Replace fastrtc with aiortc-based WebRTC implementation by @itallix in #4610
• Configurable augmentations for OD, IS, KP, SemSeg tasks by @kprokofi in #4577
• Resolve merge conflict by @kprokofi in #4623
• Fix issues during shutdown by @itallix in #4622
• Update CODEOWNERS by @leoll2 in #4635
• 4613 - render annotations - p1 by @camiloHimura in #4621
• Cleanup & improve create-project page by @jpggvilaca in #4628
• Add missing zoom provider by @jpggvilaca in #4634
• GHA hardening and bugfixes by @leoll2 in #4636
• Enable mypy pre-commit hook for geti-tune backend by @warrkan in #4633
• Create unified Docker image by @warrkan in #4631
• chore: Enable React 19 by @dwesolow in #4632
• Add model latency endpoint by @A-Artemis in #4599
• Add React Compiler to Geti Tune by @jpggvilaca in #4638
• Prep before moving SAM tool by @jpggvilaca in #4640
• Add "smart tools" package by @jpggvilaca in #4645
• Add UI team to ui CODEOWNERS by @jpggvilaca in #4646
• chore(ci): enable Renovate by @AlexanderBarabanov in #4642
• fix(deps): update dependency pillow to v11.3.0 [security] by @oep-renovate[bot] in #4652
• Fix GHA labeler by @leoll2 in #4641
• Add tool selection bar by @jpggvilaca in #4655
• 4545: Create skeleton of dataset endpoints by @warrkan in #4651
• Fix Deim recipes by @kprokofi in #4657
• Fix Deim recipe to be able to run bs search by @kprokofi in #4664
• Fix input size configuration during validation for DFine model by @kprokofi in #4666
• Fix RandomAffine and RTDetr training with IoURandomCrop by @kprokofi in #4718
• Fix loading checkpoint after 1st round of training for DFine-X model by @kprokofi in #4738
• Fix keypoint annotation filter logic by @eugene123tw in #4685
• Run unit test on self-hosted runners by @kprokofi in #4791
• Update setuptools by @kprokofi in #4766
• Fix training on CPU by @kprokofi in #4788
• Fix OOM bug on XPU by @kprokofi in http...

2.5.0

What's Changed

• 🧹Refactor Classification Entity by @ashwinvaidya17 in #4239
• Refactor OTX Models by @kprokofi in #4241
• Update dev version and changelog by @kprokofi in #4281
• Promote/demote detection models by @Daankrol in #4278
• 🗑️ Remove unused transforms by @ashwinvaidya17 in #4269
• Promote DinoV2 as "Accuracy" template by @kprokofi in #4287
• Bump the pip-others group across 1 directory with 5 updates by @dependabot[bot] in #4292
• Delete Visual Prompting Task by @kprokofi in #4291
• Fix DataInputParams Serialization by @kprokofi in #4293
• Refactor Data Entity (SSeg, KP) by @kprokofi in #4282
• Hot Fix Classification by @kprokofi in #4297
• Enable TV MaskRCNN for Geti by @eugene123tw in #4296
• Align KP Detection Validation with Model API by @kprokofi in #4300
• Update Documentation by @kprokofi in #4294
• 🧹 Minor cleanup by @ashwinvaidya17 in #4299
• Promote/Demote Inst Seg Algo by @eugene123tw in #4301
• Reduce Integration Test Scope by @eugene123tw in #4303
• Improve PR comment event trigger by @eugene123tw in #4307
• Update permissions in PR comment trigger workflow by @eugene123tw in #4309
• Update PR comment trigger by @eugene123tw in #4312
• Add workflow step to update PR description with run link and status by @eugene123tw in #4314
• Enhance PR comment trigger to include last commit message and workflow result by @eugene123tw in #4316
• Refactor PR comment trigger by @eugene123tw in #4318
• Support linear classifier fine-tuning for classification models by @gyuilLim in #4298
• 🧹 Refactor detection entity by @ashwinvaidya17 in #4322
• Update integration test commands to include run-category-only flag options by @eugene123tw in #4325
• Bump rich from 13.9.4 to 14.0.0 by @dependabot[bot] in #4324
• Fix integration test commands run-category-only flag by @eugene123tw in #4329
• updated repo with mobilenet_v4 by @GauravSRC in #4321
• Update documentation and references to Open-Edge Platform by @eugene123tw in #4331
• 🧹 Refactor Instance Segmentation Entity by @ashwinvaidya17 in #4327
• 🧹 Refactor Anomaly Entities by @ashwinvaidya17 in #4328
• Add a contribution guide by @sovrasov in #4335
• Bug Fix : Convert labels for multi-label Classification into LongTensor before one_hot encoding by @rajeshgangireddy in #4336
• Bump OV and NNCF to 2025.1 by @sovrasov in #4334
• 📝 Improve Geti Integration and Streamline Pre-Merge Tests by @eugene123tw in #4330
• 🧹 Refactor PR comment trigger workflow to improve repository checkout steps by @eugene123tw in #4342
• 📝 Enhance PR comment trigger workflow to fetch PR info dynamically before checkout by @eugene123tw in #4344
• Revert "updated repo with mobilenet_v4" by @ashwinvaidya17 in #4345
• 🧹 Update CI workflow to use Ubuntu 24.04 for Code Quality and Unit Tests by @eugene123tw in #4350
• 🧹 Remove unnecessary blank line in PR comment trigger workflow by @eugene123tw in #4351
• 🧹 Update workflows and configuration to use 'uv' for package installation by @eugene123tw in #4354
• Improve pytest command flags in CI workflow test env by @eugene123tw in #4352
• Update benchmark datasets for classification and segmentation by @rajeshgangireddy in #4349
• Update Readme by @kprokofi in #4348
• Enhance installation instructions for uv by @eugene123tw in #4358
• Update dependency by @eugene123tw in #4357
• Fix missing mAP score reporting for instance segmentation. by @eugene123tw in #4364
• Bump torch to 2.7 by @sovrasov in #4361
• Add bechmark datasets for anomaly task. by @djdameln in #4366
• ✨ Upgrade setup python action in pre-merge check by @ashwinvaidya17 in #4371
• 🗑️ Remove traces of visual prompting by @rajeshgangireddy in #4370
• Add a reference to MAPI by @sovrasov in #4373
• Fix degradation in detection/inst-seg by @eugene123tw in #4372
• Fix OTX Installation by @kprokofi in #4285
• ✨ Upgrade Ubuntu version for documentation workflows to 24.04 by @eugene123tw in #4375
• Fix input_size type check to accept list in OTXDataModule by @eugene123tw in #4380
• 🗑️ Remove CLI integration tests by @eugene123tw in #4378
• Fix CodeQL findings in pr_comment_trigger.yaml by @kprokofi in #4390
• Enhance benchmark and run scripts with job retry logic and failed job handling by @eugene123tw in #4381
• Fix in pr_comment_trigger.yaml by @AlexanderBarabanov in #4393
• Update benchmark datasets for Detection and Instance Segmentation by @Daankrol in #4365
• Fix anomaly benchmark by @eugene123tw in #4395
• Disable weight computation in self-attention for TransformerDecoderLayer by @eugene123tw in #4398
• 🧹 Update CODEOWNERS by @ashwinvaidya17 in #4404
• Disable weight computation in TransformerEncoderLayer by @eugene123tw in #4399
• 🧹 Refactor base train and test recipe by @ashwinvaidya17 in #4369
• Add model arch name to exported model metadata by @sovrasov in #4407
• Allow on_*_batch_end callbacks to access outputs from validation and test steps by @itrushkin in #4384
• Remove image caching during training to reduce complexity and RAM consumption by @kprokofi in #4401
• Improve dinov2 accuracy in classification tasks. by @gyuilLim in #4411
• Refactor benchmark logic by @eugene123tw in #4400
• Update module docstrings for consistency across various files by @eugene123tw in #4417
• Use cuda 12.8 by @ashwinvaidya17 in #4409
• ✨ Merge Feature-v3 to develop by @ashwinvaidya17 in https://github.com/...

2.4.6

What's Changed

• Fix label info dispatching by @kprokofi in #4443
• Prepare for 2.4.6 release by @kprokofi in #4444

Full Changelog: 2.4.5...2.4.6

2.4.5

What's Changed

• 🐞 Fix UFlow by @ashwinvaidya17 in #4431
• Fix loading/saving checkpoints in OTX by @kprokofi in #4433
• Fix otx load checkpoint by @ashwinvaidya17 in #4438

Full Changelog: 2.4.4...2.4.5

2.4.4

[2.4.4]

Bug fixes

• Fix torch.load() to be able to load all OTX custom snapshots
  (#4392)

Full Changelog: 2.4.3...2.4.4

2.4.3

Enhancements

• Bump torch to 2.7.0

2.4.2

What's Changed

Bug fixes

• Fix specific torchmetrics dependency to 1.6.0

2.4.1

What's Changed

Bug fixes

• Update Datumaro from 1.10.0rc0 to 1.10.0