Training Extensions Versions Save

New features

• Add zero-shot visual prompting (https://github.com/openvinotoolkit/training_extensions/pull/2616, https://github.com/openvinotoolkit/training_extensions/pull/2706, https://github.com/openvinotoolkit/training_extensions/pull/2753)
• Add support for the training and validation on the XPU devices (https://github.com/openvinotoolkit/training_extensions/pull/3058)

Enhancements

• Upgrade OpenVINO to 2023.3 (https://github.com/openvinotoolkit/training_extensions/pull/2879)
• Automate performance benchmark (https://github.com/openvinotoolkit/training_extensions/pull/2742)
• Bump ONNX version to 1.16.0 to resolve CVE-2022-25882 (https://github.com/openvinotoolkit/training_extensions/pull/3205)

What's Changed- Full Version

• Enable PTQ for visual prompting model by @sungchul2 in https://github.com/openvinotoolkit/training_extensions/pull/2519
• Update reg test for Swin-T model by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2527
• [OTX] remove dead code by @Harmjandevries in https://github.com/openvinotoolkit/training_extensions/pull/2528
• Fix the incorrect progress information when AdaptiveRepeatDataHook is enabled by @sungmanc in https://github.com/openvinotoolkit/training_extensions/pull/2510
• Migrate anomaly exportable code to modelAPI by @ashwinvaidya17 in https://github.com/openvinotoolkit/training_extensions/pull/2432
• Fix nncf optimization for anomaly task by @ashwinvaidya17 in https://github.com/openvinotoolkit/training_extensions/pull/2537
• Enable deit testing by @negvet in https://github.com/openvinotoolkit/training_extensions/pull/2536
• Remove unused HPO weight while HPO is running by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2545
• Add option to attach host path to container's cache by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2546
• Enable resuming with weights.pth saved by visual prompting by @sungchul2 in https://github.com/openvinotoolkit/training_extensions/pull/2538
• Upgrade MAPI to v 0.1.6 by @sovrasov in https://github.com/openvinotoolkit/training_extensions/pull/2529
• Add proxy env setting to ci script by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2561
• Mergeback 1.5.0rc1 by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2562
• Add light backbones for visual prompting (#1) by @sungchul2 in https://github.com/openvinotoolkit/training_extensions/pull/2544
• Update for CI by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2566
• Update OV to 2023.1 by @sovrasov in https://github.com/openvinotoolkit/training_extensions/pull/2505
• Fix and update Semi-SL Object Detection algorithms by @kprokofi in https://github.com/openvinotoolkit/training_extensions/pull/2554
• Apply resize cache to non-default models & recipes by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2576
• Update det e2e to skip unsupported train type by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2588
• Add missing import pathlib for cls e2e testing by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2610
• Unify logger usage by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2612
• Mergeback 1.5.0 to develop by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2633
• Mergeback 1.5.0 to develop by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2642
• Revert "Mergeback 1.5.0 to develop" by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2645
• Mergeback/1.5.0 by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2646
• Add a tool to help conduct experiments by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2651
• Revert inference batch size to 1 for instance segmentation by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2648
• Remove unnecessary log while building a model by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2658
• Fix a minor bug of experiment.py by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2662
• Not check avg_time_per_image during test by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2665
• Mergeback releases/1.5.0 to develop by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2663
• Update docs for enabling sphinx.ext.autosummary by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2654
• Update PTQ docs by @GalyaZalesskaya in https://github.com/openvinotoolkit/training_extensions/pull/2672
• Update regression tests for develop by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2652
• Exclude py37 target config for cibuildwheel by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2673
• Add --dryrun option to tools/experiment.py by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2674
• Update OTX explain CLI arguments by @GalyaZalesskaya in https://github.com/openvinotoolkit/training_extensions/pull/2671
• Fix e2e tests for explain by @GalyaZalesskaya in https://github.com/openvinotoolkit/training_extensions/pull/2681
• Add README.md for experiment.py by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2688
• Fix typo in reg test cmd by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2691
• Select more proper model weight file according to commands run just before by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2696
• Add visual prompting zero-shot learning (learn & infer) by @sungchul2 in https://github.com/openvinotoolkit/training_extensions/pull/2616
• Fix unsupported dtype in ov graph constant converter by @sovrasov in https://github.com/openvinotoolkit/training_extensions/pull/2676
• Skip failure TC with adding issue number ref. by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2717
• Fix visual prompting e2e test by @sungchul2 in https://github.com/openvinotoolkit/training_extensions/pull/2719
• Remove duplicated variable combination in experiment.py by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2713
• Enhance detection & instance segmentation experiment by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2710
• Upgrade NNCF & OpenVINO by @sovrasov in https://github.com/openvinotoolkit/training_extensions/pull/2656
• Fix multilabel classification class index by @sungmanc in https://github.com/openvinotoolkit/training_extensions/pull/2736
• Refine parsing final score of training in experiment.py by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2738
• Make mean teacher algorithm consider distributed training by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2729
• Add visual prompting zero-shot learning (export, IR inference) by @sungchul2 in https://github.com/openvinotoolkit/training_extensions/pull/2706
• Automate performance benchmark by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2742
• Update tools/expreiment.py by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2751
• Split accuracy & speed benchmark github workflows by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2763
• Fix a bug that error is raised when train set size is greater than minimumof batch size in HPO by exactly 1 by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2760
• Fix a bug that a process tracking resource usage doesn't exit when main process raises an error by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2765
• Support multiple experiments in single recipe for tools/experiment.py by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2757
• Enable perf benchmark result logging to mlflow server by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2768
• Bump datumaro version to 1.6.0rc1 by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2784
• Update perf logging by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2785
• Update perf logging workflow to get branch+sha from gh context by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2791
• Add visual prompting zero-shot learning (optimize, documentation, bug fixes) by @sungchul2 in https://github.com/openvinotoolkit/training_extensions/pull/2753
• Check performance benchmark result with reference by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2821
• Mergeback releases/1.5.0 to develop by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2830
• Create OSSF scorecard workflow by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2831
• Fix ossf/scorecard-action version by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2832
• Update perf benchmark reference by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2843
• Set default wf permission to read-all by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2882
• Remedy token permission issue by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2888
• Add progress callback interface to HPO by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2889
• Restrict configurable parameters to avoid unreasonable cost for SaaS trial by @harimkang in https://github.com/openvinotoolkit/training_extensions/pull/2891
• Fix more token-permission issues - part3 by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2893
• Resolve pinned-dependency issues on publish_internal workflow by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2907
• Forward unittest workloads to AWS by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2887
• Resolve pinned dependency issues on workflows by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2909
• Fix pinned-dependency issues - part2 by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2911
• Add pinning dependencies by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2916
• Update pip install cmd to use hashes by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2919
• Fix HPO progress callback bug by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2908
• Fix pinned-dependencies issues by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2929
• Remove unused test files by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2930
• Update weekly workflow to run perf tests by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2920
• Adjust permission of documentation workflows from pages to contents for writing by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2933
• Add dependabot.yml for auto updating dependencies by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2935
• Bump actions/checkout from 3.6.0 to 4.1.1 by @dependabot in https://github.com/openvinotoolkit/training_extensions/pull/2949
• Bump github/codeql-action from 2.24.0 to 3.24.3 by @dependabot in https://github.com/openvinotoolkit/training_extensions/pull/2941
• Bump actions/download-artifact from 3.0.2 to 4.1.2 by @dependabot in https://github.com/openvinotoolkit/training_extensions/pull/2945
• Bump pypa/gh-action-pypi-publish from 1.7.1 to 1.8.11 by @dependabot in https://github.com/openvinotoolkit/training_extensions/pull/2947
• Update usage of save_model by @sovrasov in https://github.com/openvinotoolkit/training_extensions/pull/2932
• Provide iter_time, data_time from anomaly task by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2956
• Add ossf-scorecard badge to readme by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2959
• Add codeql workflow by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2963
• Promote numpy version by @sovrasov in https://github.com/openvinotoolkit/training_extensions/pull/2958
• Update OV by @sovrasov in https://github.com/openvinotoolkit/training_extensions/pull/2879
• Update visual prompting task (finetuning & zero-shot learning) by @sungchul2 in https://github.com/openvinotoolkit/training_extensions/pull/2892
• Update ote-ci container image to ote-ci:11.7.1.1-devel-ubuntu20.04 by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2976
• Bump onnx from 1.13.0 to 1.15.0 in /requirements by @dependabot in https://github.com/openvinotoolkit/training_extensions/pull/2986
• Update pymongo requirement from ==4.5.* to ==4.6.* in /requirements by @dependabot in https://github.com/openvinotoolkit/training_extensions/pull/2988
• Bump actions/upload-artifact from 3.1.3 to 4.3.1 by @dependabot in https://github.com/openvinotoolkit/training_extensions/pull/2979
• Bump mlflow version to 2.10.2 to omit XSS issue by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/3000
• Update codeql workflow by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/3001
• Remove yapf dependency by @dependabot in https://github.com/openvinotoolkit/training_extensions/pull/2987
• Apply changes to omit artifact upload issue by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2999
• Loose codecov failure checking threshold 0 to 1% by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2995
• Fix weekly workflow issue by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2974
• Update coverage requirement from ==7.3.* to ==7.4.* in /requirements by @dependabot in https://github.com/openvinotoolkit/training_extensions/pull/2984
• Bump actions/setup-python from 4.7.1 to 5.0.0 by @dependabot in https://github.com/openvinotoolkit/training_extensions/pull/2981
• Bump ubuntu from a4fab18 to bb1c416 in /docker by @dependabot in https://github.com/openvinotoolkit/training_extensions/pull/2983
• Add push event to code scanning workflow by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/3016
• Add memory usage logging to unittests by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/3015
• Hofix weekly workflow by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/3017
• Bump github/codeql-action from 3.24.3 to 3.24.5 by @dependabot in https://github.com/openvinotoolkit/training_extensions/pull/2978
• Bump actions/setup-python from 4.7.1 to 5.0.0 by @dependabot in https://github.com/openvinotoolkit/training_extensions/pull/3022
• Bump actions/stale from 4.1.1 to 9.0.0 by @dependabot in https://github.com/openvinotoolkit/training_extensions/pull/2982
• Bump ossf/scorecard-action from 2.1.2 to 2.3.1 by @dependabot in https://github.com/openvinotoolkit/training_extensions/pull/3020
• Update doc for CPU install and MMCV troubleshooting by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/3033
• Revert MRCNN resize to the state from 1.4 [backported from 1.5] by @sovrasov in https://github.com/openvinotoolkit/training_extensions/pull/3030
• Bump twine from 4.0.2 to 5.0.0 in /requirements by @dependabot in https://github.com/openvinotoolkit/training_extensions/pull/2985
• Bump pydata-sphinx-theme from 0.12.0 to 0.15.2 in /requirements by @dependabot in https://github.com/openvinotoolkit/training_extensions/pull/3025
• Update myst-parser requirement from ==1.0.* to ==2.0.* in /requirements by @dependabot in https://github.com/openvinotoolkit/training_extensions/pull/3024
• Bump pypa/gh-action-pypi-publish from 1.8.11 to 1.8.12 by @dependabot in https://github.com/openvinotoolkit/training_extensions/pull/3021
• Bump svenstaro/upload-release-action from 2.7.0 to 2.9.0 by @dependabot in https://github.com/openvinotoolkit/training_extensions/pull/2980
• Bump sphinx version to 6.2.1 by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/3042
• Fix daily tests by @sovrasov in https://github.com/openvinotoolkit/training_extensions/pull/3050
• Update visual prompting on 1.x by @sungchul2 in https://github.com/openvinotoolkit/training_extensions/pull/3038
• Refine perf benchmark v1 by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/3061
• Add support of the Intel dGPUs devices by @kprokofi in https://github.com/openvinotoolkit/training_extensions/pull/3058
• Add xfails in vpm e2e by @sungchul2 in https://github.com/openvinotoolkit/training_extensions/pull/3062
• Fix a bug that error is raised when training a model after HPO on multi XPU environment by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/3081
• Update perf benchmark refererence v1 by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/3086
• Redirect unittest workflow to local runners by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/3088
• Add the documentation for installing OTX on XPU devices by @kprokofi in https://github.com/openvinotoolkit/training_extensions/pull/3080
• Fix bug on visual prompting by @sungchul2 in https://github.com/openvinotoolkit/training_extensions/pull/3093
• Update benchmark summary v1 by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/3100
• Update pytorch-lightning version requirement for XPU by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/3145
• Run perf benchmark on large-disk host by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/3157
• Make HPO raise an error once HPO trial exits abnormally by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/3190
• Remove internal publish workflow by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/3192
• Add use ellipse shapes to omz config by @jaegukhyun in https://github.com/openvinotoolkit/training_extensions/pull/3207
• Bump onnx version to 1.16.0 by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/3205
• Change logger in otx.api to default logger by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/3228
• Update tpp file by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/3237
• Mergeback 1.5.2 to 1.6.0 by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/3214
• Adaptive batch size supports XPU by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/3256

Full Changelog: https://github.com/openvinotoolkit/training_extensions/compare/1.5.2...1.6.0

Bug fixes

• Fix label order for h-cls (https://github.com/openvinotoolkit/training_extensions/pull/2921)
• Revert MRCNN resize to state from 1.4 (https://github.com/openvinotoolkit/training_extensions/pull/2922)
• Bump up MAPI version to 0.1.8 -> 0.1.9 (https://github.com/openvinotoolkit/training_extensions/pull/2923)
• Revert polygon clipping code (https://github.com/openvinotoolkit/training_extensions/pull/2926)
• Fix default memcache size to 100MB (https://github.com/openvinotoolkit/training_extensions/pull/2960, https://github.com/openvinotoolkit/training_extensions/pull/2990)
• Fix wrong domain in tiling rotated detection (https://github.com/openvinotoolkit/training_extensions/pull/3141)

Full Changelog: https://github.com/openvinotoolkit/training_extensions/compare/1.5.0...1.5.2

Bug fixes

• 🐞 Filter invalid polygon shapes (https://github.com/openvinotoolkit/training_extensions/pull/2795)
• 🐞 Set reverse_input_channels to True in OpenVINO models (https://github.com/openvinotoolkit/training_extensions/pull/2848)
• Remove dependency of protobuf (https://github.com/openvinotoolkit/training_extensions/pull/2851)
• 🐞 Fix label_to_idx for hierarchical classification (https://github.com/openvinotoolkit/training_extensions/pull/2906)

Full Changelog: https://github.com/openvinotoolkit/training_extensions/compare/1.4.4...1.4.5

New features

• Enable configurable confidence threshold for otx eval and export (https://github.com/openvinotoolkit/training_extensions/pull/2388)
• Add YOLOX variants as new object detector models (https://github.com/openvinotoolkit/training_extensions/pull/2402)
• Enable FeatureVectorHook to support action tasks (https://github.com/openvinotoolkit/training_extensions/pull/2408)
• Add ONNX metadata to detection, instance segmantation, and segmentation models (https://github.com/openvinotoolkit/training_extensions/pull/2418)
• Add a new feature to configure input size (https://github.com/openvinotoolkit/training_extensions/pull/2420)
• Introduce the OTXSampler and AdaptiveRepeatDataHook to achieve faster training at the small data regime (https://github.com/openvinotoolkit/training_extensions/pull/2428)
• Add a new object detector Lite-DINO (https://github.com/openvinotoolkit/training_extensions/pull/2457)
• Add Semi-SL Mean Teacher algorithm for Instance Segmentation task (https://github.com/openvinotoolkit/training_extensions/pull/2444)
• Official supports for YOLOX-X, YOLOX-L, YOLOX-S, ResNeXt101-ATSS (https://github.com/openvinotoolkit/training_extensions/pull/2485)
• Add new argument to track resource usage in train command (https://github.com/openvinotoolkit/training_extensions/pull/2500)
• Add Self-SL for semantic segmentation of SegNext families (https://github.com/openvinotoolkit/training_extensions/pull/2215)
• Adapt input size automatically based on dataset statistics (https://github.com/openvinotoolkit/training_extensions/pull/2499)

Enhancements

• Refine input data in-memory caching (https://github.com/openvinotoolkit/training_extensions/pull/2416)
• Adapt timeout value of initialization for distributed training (https://github.com/openvinotoolkit/training_extensions/pull/2422)
• Optimize data loading by merging load & resize operations w/ caching support for cls/det/iseg/sseg (https://github.com/openvinotoolkit/training_extensions/pull/2438, https://github.com/openvinotoolkit/training_extensions/pull/2453, https://github.com/openvinotoolkit/training_extensions/pull/2460)
• Support torch==2.0.1 (https://github.com/openvinotoolkit/training_extensions/pull/2465)
• Set "Auto" as default input size mode (https://github.com/openvinotoolkit/training_extensions/pull/2515)

Bug fixes

• Fix F1 auto-threshold to choose best largest confidence (https://github.com/openvinotoolkit/training_extensions/pull/2371)
• Fix a performance drop while training EfficientNetV2 with multi-GPU (https://github.com/openvinotoolkit/training_extensions/pull/2398)
• Fix bug that auto adaptive batch size raises an error if CUDA isn't available (https://github.com/openvinotoolkit/training_extensions/pull/2410)
• Fix sampler degradation issue (https://github.com/openvinotoolkit/training_extensions/pull/2482)
• Fix a bug that fp16 key error is raised when training without GPU (https://github.com/openvinotoolkit/training_extensions/pull/2501)
• Fix bug that auto batch size doesn't consider distributed training (https://github.com/openvinotoolkit/training_extensions/pull/2533)
• Fix auto input size mismatch in eval & export (https://github.com/openvinotoolkit/training_extensions/pull/2530)
• Fix the CustomNonLinearClsHead when the batch_size is set to 1 (https://github.com/openvinotoolkit/training_extensions/pull/2571)
• Fix IBLoss enablement with DeiT-Tiny when class incremental training (https://github.com/openvinotoolkit/training_extensions/pull/2594)
• Fix XAI algorithm for Detection (https://github.com/openvinotoolkit/training_extensions/pull/2617)

Full Changelog: https://github.com/openvinotoolkit/training_extensions/compare/1.4.4...1.5.0

What's Changed

• Update ModelAPI configuration by @sovrasov in https://github.com/openvinotoolkit/training_extensions/pull/2564
• Add Anomaly modelAPI changes to releases/1.4.0 by @ashwinvaidya17 in https://github.com/openvinotoolkit/training_extensions/pull/2563
• Update for 1.4.4rc1 by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2572
• Bump datumaro version to 1.5.1rc3 by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2587
• Add label ids to anomaly OpenVINO model xml by @ashwinvaidya17 in https://github.com/openvinotoolkit/training_extensions/pull/2590
• Fix DeiT-Tiny regression test for release/1.4.0 by @wonjuleee in https://github.com/openvinotoolkit/training_extensions/pull/2595
• Fix mmcls bug not wrapping model in DataParallel on CPUs by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2601
• Fix h-label loss normalization issue w/ exclusive label group of singe label by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2604
• Boost up Image numpy accessing speed through PIL by @wonjuleee in https://github.com/openvinotoolkit/training_extensions/pull/2586
• Fix division by zero in class incremental learning for classification by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2606
• Fix XAI algorithm for Detection by @GalyaZalesskaya in https://github.com/openvinotoolkit/training_extensions/pull/2609
• Tighten dependency constraint only adapting latest patches by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2607
• Add metadata to optimized model by @ashwinvaidya17 in https://github.com/openvinotoolkit/training_extensions/pull/2618
• Update dependency constraint by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2622
• Fix h-label bug of missing parent labels in output by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2626
• Update publish workflow by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2625
• Update publish workflow for tag checking by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2632
• Update release note and readme by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2637
• Make max_num_detections configurable by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2647
• Fix CPU training issue on non-CUDA system by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2655
• Update MAPI version by @sovrasov in https://github.com/openvinotoolkit/training_extensions/pull/2730
• Update dependency for exportable code by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2732

Full Changelog: https://github.com/openvinotoolkit/training_extensions/compare/1.4.3...1.4.4

What's Changed

• Update ModelAPI configuration by @sovrasov in https://github.com/openvinotoolkit/training_extensions/pull/2564
• Add Anomaly modelAPI changes to releases/1.4.0 by @ashwinvaidya17 in https://github.com/openvinotoolkit/training_extensions/pull/2563
• Update for 1.4.4rc1 by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2572
• Bump datumaro version to 1.5.1rc3 by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2587

Full Changelog: https://github.com/openvinotoolkit/training_extensions/compare/1.4.3...1.4.4rc1

What's Changed

• POT->PTQ renaming in tests by @sovrasov in https://github.com/openvinotoolkit/training_extensions/pull/2319
• Update the version of the openvino badge by @sungmanc in https://github.com/openvinotoolkit/training_extensions/pull/2329
• Tiling doc for develop by @eugene123tw in https://github.com/openvinotoolkit/training_extensions/pull/2334
• Relax reference saliency values check by @negvet in https://github.com/openvinotoolkit/training_extensions/pull/2338
• Call otx.cli before the TaskType is imported by @harimkang in https://github.com/openvinotoolkit/training_extensions/pull/2342
• added fix-cpu numbers option by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2340
• Fix tiling unittest and random max select logic by @eugene123tw in https://github.com/openvinotoolkit/training_extensions/pull/2336
• Update incorrect license header by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2352
• Fixing XAI tests by @negvet in https://github.com/openvinotoolkit/training_extensions/pull/2357
• Change IR scale factor to 1.0 if tile size is too big by @eugene123tw in https://github.com/openvinotoolkit/training_extensions/pull/2337
• Add model category attributes to model template by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2362
• Fix F1 auto-threshold to choose best largest confidence by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2371
• Add model type to deit ptq config by @sovrasov in https://github.com/openvinotoolkit/training_extensions/pull/2379
• Add description & remove unused configurable parameters by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2382
• Add onnx metadata to classification models by @sovrasov in https://github.com/openvinotoolkit/training_extensions/pull/2378
• Update exportable code README (to develop) by @GalyaZalesskaya in https://github.com/openvinotoolkit/training_extensions/pull/2389
• Rename SAMImageClassifier -> CustomImageClassifier by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2384
• Merge back 1.4.0 to develop by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2387
• Update version string of the OTX by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2393
• [develop] XAI: Return saliency maps for Mask RCNN IR async infer by @GalyaZalesskaya in https://github.com/openvinotoolkit/training_extensions/pull/2391
• Enable deterministic for anomalib and visual prompting by @sungchul2 in https://github.com/openvinotoolkit/training_extensions/pull/2361
• Fix a performance drop while training EfficientNetV2 with multi-GPU by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2398
• Change POT to PTQ in test code by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2401
• Enable configurable confidence threshold for ov export and inference by @jaegukhyun in https://github.com/openvinotoolkit/training_extensions/pull/2388
• Fix e2e by @jaegukhyun in https://github.com/openvinotoolkit/training_extensions/pull/2404
• Add YOLOX variants to OTX by @jaegukhyun in https://github.com/openvinotoolkit/training_extensions/pull/2402
• Enable FeatureVectorHook for action tasks by @jaegukhyun in https://github.com/openvinotoolkit/training_extensions/pull/2408
• Update Anomaly OpenVINO export for ModelAPI by @ashwinvaidya17 in https://github.com/openvinotoolkit/training_extensions/pull/2400
• Fix bug that auto adaptive batch size raises an error if CUDA isn't available by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2410
• Check sal maps for async infer for M-RCNN by @negvet in https://github.com/openvinotoolkit/training_extensions/pull/2396
• Remove data cfg in detection training recipe by @jaegukhyun in https://github.com/openvinotoolkit/training_extensions/pull/2409
• Switch to MAPI in classification and detection inference tasks by @sovrasov in https://github.com/openvinotoolkit/training_extensions/pull/2390
• Mergeback 1.4.1 to develop by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2413
• Enable ViTReciproCAM for VisionTransformer backbone (cls task) by @negvet in https://github.com/openvinotoolkit/training_extensions/pull/2403
• Add onnx meta for sseg, iseg, and detection models by @sovrasov in https://github.com/openvinotoolkit/training_extensions/pull/2418
• Reduce amount of SSeg PTQ e2e tests by @sovrasov in https://github.com/openvinotoolkit/training_extensions/pull/2414
• Refactor instance segmentation and tiling inference by @sovrasov in https://github.com/openvinotoolkit/training_extensions/pull/2415
• Refine input data in-memory caching by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2416
• Configurable input size by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2420
• Adpat timeout value of torch.dist.init_process_group depending on elapsed time by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2422
• Refactor semantic segmentation inference to use MAPI by @sovrasov in https://github.com/openvinotoolkit/training_extensions/pull/2421
• Update FQ refs for iseg models by @sovrasov in https://github.com/openvinotoolkit/training_extensions/pull/2426
• Limit range of PTQ parameters by @sovrasov in https://github.com/openvinotoolkit/training_extensions/pull/2425
• Refactor configurer - Phase1 by @jaegukhyun in https://github.com/openvinotoolkit/training_extensions/pull/2424
• Upgrade ModelAPI by @sovrasov in https://github.com/openvinotoolkit/training_extensions/pull/2429
• Move SegNext models from experimental to custom models by @kprokofi in https://github.com/openvinotoolkit/training_extensions/pull/2312
• Refactor configurer - Phase 2 by @jaegukhyun in https://github.com/openvinotoolkit/training_extensions/pull/2430
• Introduce AdaptiveRepeatDataHook for the classification by @sungmanc in https://github.com/openvinotoolkit/training_extensions/pull/2428
• Write a docs and cover additional case about configurable input size by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2433
• Fix E2E, Regression test failures and edit the inproper log message by @sungmanc in https://github.com/openvinotoolkit/training_extensions/pull/2436
• Enable multi-gpus for weekly by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2364
• Add MaskRCNN SwinTransformer-T by @eugene123tw in https://github.com/openvinotoolkit/training_extensions/pull/2423
• Fix i-seg e2e test code by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2446
• Stabilize multi GPU training by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2443
• Merge load / resize / cache to optimize data loading efficiency for classification by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2438
• Refactor configurer - Phase 3 by @jaegukhyun in https://github.com/openvinotoolkit/training_extensions/pull/2442
• Merge load / resize / cache to optimize data loading efficiency for detection & instance segmentation by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2453
• Align resume batch size with train batch size in i-seg e2e test by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2458
• Merge load / resize / cache to optimize data loading efficiency for semantic segmentation by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2460
• Apply load+resize cache to validation pipelines by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2463
• Add a new object detector, Lite-DINO by @jaegukhyun in https://github.com/openvinotoolkit/training_extensions/pull/2457
• Bump datumaro version by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2466
• Set tox version constraint by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2471
• Add utilization of the unlabeled subset in PTQ optimization by @kprokofi in https://github.com/openvinotoolkit/training_extensions/pull/2468
• Remove unnecessary workflows by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2474
• Add unit tests for DINO and Lite-DINO by @jaegukhyun in https://github.com/openvinotoolkit/training_extensions/pull/2475
• Mergeback 1.4.2rc2 by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2464
• Added quantization configs for SegNext T S B by @nikita-savelyevv in https://github.com/openvinotoolkit/training_extensions/pull/2477
• Update regression criteria for anomaly task by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2476
• Fix some issues on cls regression tests by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2478
• Adjust PTQ fq target by @chuneuny-emily in https://github.com/openvinotoolkit/training_extensions/pull/2480
• Add Semi-SL Instance Segmentation by @kprokofi in https://github.com/openvinotoolkit/training_extensions/pull/2444
• Fix some bugs for reg tests by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2481
• Fix sampler degradation issue by @sungmanc in https://github.com/openvinotoolkit/training_extensions/pull/2482
• Make MaskRCNN-SwinT from experimental to production by @eugene123tw in https://github.com/openvinotoolkit/training_extensions/pull/2454
• Remove mpa legacy by @jaegukhyun in https://github.com/openvinotoolkit/training_extensions/pull/2486
• Fix some regression test issues for segmentation by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2488
• Re-organization of configurer by @jaegukhyun in https://github.com/openvinotoolkit/training_extensions/pull/2484
• Model revamp for Geti - Object Detection by @jaegukhyun in https://github.com/openvinotoolkit/training_extensions/pull/2485
• Fix e2e tests for xai_sanity_classification by @GalyaZalesskaya in https://github.com/openvinotoolkit/training_extensions/pull/2490
• Fix object detection e2e tests by @jaegukhyun in https://github.com/openvinotoolkit/training_extensions/pull/2492
• Add Semi-SL Instance Segmentation Documentation by @kprokofi in https://github.com/openvinotoolkit/training_extensions/pull/2487
• Update unit tests for XAI by @GalyaZalesskaya in https://github.com/openvinotoolkit/training_extensions/pull/2491
• Update .pre-commit-config.yaml by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2494
• Align batch size in instance segmentation task for Semi-SL to avoid OOM error by @kprokofi in https://github.com/openvinotoolkit/training_extensions/pull/2493
• Fix unit tests for XAI by @GalyaZalesskaya in https://github.com/openvinotoolkit/training_extensions/pull/2495
• Update inst-seg end2end test fake quantization by @eugene123tw in https://github.com/openvinotoolkit/training_extensions/pull/2496
• Add new argument to track CPU&GPU utilization and memory usage by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2500
• Fix a bug that fp16 key error is raised when training without GPU by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2501
• Move configuration of configs in task.py to configurer.py by @jaegukhyun in https://github.com/openvinotoolkit/training_extensions/pull/2498
• Fix reg summary bug by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2503
• Set mem-cache-1G as a default training by @sungmanc in https://github.com/openvinotoolkit/training_extensions/pull/2489
• Add Self-SL configs for SegNext by @sungchul2 in https://github.com/openvinotoolkit/training_extensions/pull/2215
• Update Inst Seg Model Revamp by @eugene123tw in https://github.com/openvinotoolkit/training_extensions/pull/2506
• Add human right disclaimer in README.md by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2508
• Fix e2e failed case of visual prompting task by @sungchul2 in https://github.com/openvinotoolkit/training_extensions/pull/2509
• Support torch==2.0.1 by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2465
• Mergeback 1.4.2rc5 by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2511
• Adapt input size automatically based on dataset statistics by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2499
• Fix rotated detection by @sungmanc in https://github.com/openvinotoolkit/training_extensions/pull/2512
• Adjust FQ criteria in anomaly/instance seg tasks by @chuneuny-emily in https://github.com/openvinotoolkit/training_extensions/pull/2517
• Set "Auto" as default input size mode by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2515
• Correct artifact csv name by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2516
• Fix E2E tests for detection by @sungmanc in https://github.com/openvinotoolkit/training_extensions/pull/2520
• Add skips to E2E detection tests by @sungmanc in https://github.com/openvinotoolkit/training_extensions/pull/2521
• Delete mem cache handler after training is done by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2535
• Fix bug that auto batch size doesn't consider distributed training by @eunwoosh in https://github.com/openvinotoolkit/training_extensions/pull/2533
• Apply fix progress hook to release 1.5.0 by @sungmanc in https://github.com/openvinotoolkit/training_extensions/pull/2539
• Mirror Anomaly ModelAPI changes by @ashwinvaidya17 in https://github.com/openvinotoolkit/training_extensions/pull/2531
• Re-introduce adaptive training by @jaegukhyun in https://github.com/openvinotoolkit/training_extensions/pull/2543
• Fix auto input size mismatch in eval & export by @goodsong81 in https://github.com/openvinotoolkit/training_extensions/pull/2530
• Update ref. fq number for anomaly e2e2 by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2547
• Skip e2e det tests by issue2548 by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2550
• Add skip to chained TC for issue #2548 by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2552

Full Changelog: https://github.com/openvinotoolkit/training_extensions/compare/1.4.3...1.5.0rc1

Enhancements

• Re introduce adaptive scheduling for training (https://github.com/openvinotoolkit/training_extensions/pull/2541)

Full Changelog: https://github.com/openvinotoolkit/training_extensions/compare/1.4.2...1.4.3

What's Changed

• Re introduce adaptive scheduling for training by @jaegukhyun in https://github.com/openvinotoolkit/training_extensions/pull/2541
• Update for release 1.4.3rc1 by @yunchu in https://github.com/openvinotoolkit/training_extensions/pull/2542

Full Changelog: https://github.com/openvinotoolkit/training_extensions/compare/1.4.2...1.4.3rc1

Enhancements

• Add workaround for the incorrect meta info M-RCNN (used for XAI) (https://github.com/openvinotoolkit/training_extensions/pull/2437)
• Add model category attributes to model template (for releases/1.4.0) (https://github.com/openvinotoolkit/training_extensions/pull/2439)
• Update PTQ ignored scope for hrnet 18 mod2 (https://github.com/openvinotoolkit/training_extensions/pull/2449)
• Modified fq numbers for lite HRNET (https://github.com/openvinotoolkit/training_extensions/pull/2445)
• Prevent zero-sized saliency map in tiling if tile size is too big (https://github.com/openvinotoolkit/training_extensions/pull/2452)
• Bump datumaro version (https://github.com/openvinotoolkit/training_extensions/pull/2502)
• Add a dummy hierarchical config required by MAPI (https://github.com/openvinotoolkit/training_extensions/pull/2483)
• Set tox version constraint (https://github.com/openvinotoolkit/training_extensions/pull/2472)
• Update pot fq reference number (https://github.com/openvinotoolkit/training_extensions/pull/2456)
• Upgrade nncf version to 2.6.0 (https://github.com/openvinotoolkit/training_extensions/pull/2459)

Bug fixes

• Fix label list order for h-label classification (https://github.com/openvinotoolkit/training_extensions/pull/2440)
• Fix OpenVINO inference for legacy models (https://github.com/openvinotoolkit/training_extensions/pull/2450)
• Bug fix for albumentations (https://github.com/openvinotoolkit/training_extensions/pull/2467)

Full Changelog: https://github.com/openvinotoolkit/training_extensions/compare/1.4.1...1.4.2