Skip to content

Latest commit

 

History

History
264 lines (174 loc) · 20.1 KB

CHANGELOG.md

File metadata and controls

264 lines (174 loc) · 20.1 KB

Changelog

v0.16.0 (2024-06-13)

Feature

Fix

  • Localizeexperiment with location config (5edff9f)

v0.15.0 (2024-05-15)

Feature

  • Allow freeze of model weights in training (1fee972)

v0.14.1 (2024-04-16)

Fix

  • Test pipeline pendulum dependency (299426a)
  • Update poetry in Github pipelines to 1.7.1 (2d0dd2e)
  • Load non-parq files with DFLoader load_df (b693710)
  • Add original model id to eval experiment info (494c101)

v0.14.0 (2024-02-08)

Feature

  • Update altair to 5.2.0 (#113) (fbb72d4)
  • Update tensorflow (>=2.13,<=2.15) and dagster (~1.6.0). Remove support for python 3.8 (#106) (9c3f6af)

Fix

  • Simplify ExperimentManager using exp_dict (9a54dc9)
  • Query experiments based on storage handler name not mutable id (1e94add)
  • Changed tf-macos version to <=2.14 (5516531)

v0.13.0 (2024-01-16)

Feature

  • Update tensorflow (>=2.13,<=2.15) and dagster (~1.6.0). Remove support for python 3.8 (#106) (71d82cc)

Fix

  • ci: Set Python version to 3.11 due to an interpretation error of the Github actions (29bafbf)
  • Changed tf-macos version to <=2.14 (a84c6c4)

v0.12.0 (2024-01-04)

Feature

  • Dependencies: Allow Python 3.11 and adapt dependency versions (a60a420)

Fix

  • Dependencies: Set tensorflow-io-gcs-filesystem to ~0.33.0 (59a059f)

Documentation

  • Adjust python version information in documentation (5ea5e93)

v0.11.0 (2023-12-20)

Feature

  • Add contribution guidelines to the project (#103) (b9ea55a)
  • MLflow: Allow logging of nested tuple or int metrics in the TensorGraphAnalyzer (daea28a)
  • Change model custom objects and callbacks refactored (c6f9fd3)
  • Write and read json options in experiment context (#90) (1aa8bd5)

Fix

  • Rename tensorflow to keras (f38536f)
  • Pillow vulnerability and other + formatting (#89) (9c62f3b)
  • Reload finished experiments in dashboard (#87) (aafe59a)
  • Generate mkdocs graphs with multiple dependencies (#86) (61488b2)
  • Replace lambda class_extractor in DirClsDataInfoListing with a private function because lambdas are not pickable (#85) (98f25df)
  • Remove job flag to prevent multiple mlflow runs (#82) (31b3875)
  • Remove job flag to prevent multiple mlflow runs (8c8b78f)

Documentation

  • contribution: Add information about how conventional and non-conventional commits are handled in pull requests (ec12c19)
  • DevEntSetup: Refactor hint for other than conda (bc5acfe)
  • DevEnvSetup: Add hint for pipenv or other comparable solutions. Add recommendation for Apple Silicon (2c732f5)
  • Update contact information in CODE_OF_CONDUCT (9780565)
  • Add contribution guidelines and community standards into documentation (b209c4b)
  • Optimize coding standards (3810dcb)
  • Add Commit Guidelines (c404b38)
  • Add Developer Environment Setup Guide (94d5e83)
  • Standardize headings (50805f7)
  • Add coding standards and contribution guidelines (3a64c22)

v0.10.0 (2023-12-12)

Feature

v0.9.1 (2023-12-05)

Fix

v0.9.0 (2023-11-28)

Feature

  • Write and read json options in experiment context (#90) (77d3dc9)

Fix

  • Pillow vulnerability and other + formatting (#89) (3d3e6af)

v0.8.3 (2023-11-08)

Fix

  • Reload finished experiments in dashboard (#87) (39764f5)
  • Generate mkdocs graphs with multiple dependencies (#86) (ac56427)
  • Replace lambda class_extractor in DirClsDataInfoListing with a private function because lambdas are not pickable (#85) (def7d1a)

v0.8.2 (2023-10-09)

Fix

  • Add installation info to main documentation (#84) (10e60db)
  • Add installation info to main documentation (b2f8c10)

v0.8.1 (2023-09-26)

Fix

  • Analysis now have credentials for datasets (#81) (aeb6e63)
  • Analysis now have credentials for datasets (87f7214)

v0.8.0 (2023-09-21)

Feature

  • Integrate mlflow in niceml (#79) (23aa7dc)
  • Add predictionfunction and models as folder (#78) (368c25c)

Fix

  • Return the experiment path instead of the ExperimentInfo if exp_id is not latest (#76) (2d31041)

v0.7.3 (2023-09-04)

Fix

  • Optimize file locks and remove outdated template config (#72) (dc374a6)
  • Implemented image_size.from_pil image and size (#71) (0643bf2)

v0.7.2 (2023-09-01)

Fix

  • Remove Windows from documentation (0c252aa)
  • Remove tensorflow-io-gcs-filesystem as a dependency from the project template (ce4ea6c)
  • Remove Windows as an option in niceml init. (a07cab4)
  • Safely remove tensorflow-io-gcs-filesystem from poetry-lock (31515ca)

v0.7.1 (2023-08-22)

Fix

  • Remove tensorflow-io-gcs-filesystem from dependencies (#69) (cfbadc8)

v0.7.0 (2023-08-15)

Feature

  • Improvements to DfDataset and other files to better perform regression experiments (#56) (f88abd8)
  • Join location with path objects (#63) (d73286d)
  • Dependencies: Update of dagster to 1.4.x , tensorflow, and the corresponding packages to 2.12.x (#60) (99b997e)

Fix

  • PR feedback (bd6d475)
  • PR feedback (c76a6e6)
  • How to guide one yaml was false indented (61c023c)
  • Change type of image_path in load_img_uint8 (#55) (7df85ee)

Documentation

  • Add PR feedback (84b6c6c)
  • Add how to start the pipeline via dagster-webserver (79a4db0)
  • Add how to add a custom model (4027ef8)
  • Add how to write a custom pipeline (66fe21c)
  • Add into to how to write a custom pipeline (f0ec339)
  • Add automatic generation of API documentation based on docstrings (#54) (8a16e6f)

v0.6.1 (2023-08-02)

Fix

  • Solve niceML installation issue due to pydantic and dagster version missmatsch(#57) (9303e19)
  • Update tornado to ^6.3.2 (6315889)
  • Update requests to ^2.13.0 (6559b64)
  • Update cryptrography to ^41.0.0 (b188f99)
  • Update dagster to ~1.3.13 (c842df0)
  • Set pydantic version to <2.0 (6b63ef1)

v0.6.0 (2023-07-10)

Feature

  • Add NanDataframeFilter to drop nan values of feature columns (#51) (3561200)
  • Add dagster op for dataframe normalization (#48) (fe0e8e0)
  • Add lockfile name as attribute of FileChecksumProcessor (#46) (013addf)

Fix

  • Adjust supported python versions in Getting Started docs section (4bde794)
  • Save result files from tensorgraphanalyzer at the correct place and implemented validation for that (#50) (ea3191b)
  • Remove temp directory from hydra search path. Add hydra config mapping factory (#47) (6660ec9)

v0.5.0 (2023-06-23)

Feature

  • Add abstract FileChecksumProcessor and concrete class ZippedCsvToParquetProcessor (#44) (a21d673)
  • Implementation of filelocks (#43) (9d4dacd)

Fix

v0.4.1 (2023-06-22)

Fix

  • Update the version of dagster to 1.3.9 (#34) (118643d)

v0.4.0 (2023-06-05)

Feature

  • Softmax for Semantic Segmentation (#30) (ec669b1)
  • New tensorflow-metal version (#29) (721842b)
  • New tensorflow-metal version (e629895)

v0.3.0 (2023-05-17)

Feature

  • Add albumentation for numbers dataset image augmentation (8da7b24)
  • Add conversion of cropped_numbers_images to tabular data as an op (28bfafc)

Fix

  • Adjust explanatory comments for albumentation (6f35f88)
  • Allow None as initial value for generating number images (55272e5)
  • Adjust the tests for image generation for albumentation (6db814c)
  • Improvements that for eacht subset a row is written (5982ae3)
  • Replace weak md5 hash (CWE-327) (a438cc9)
  • Make all template env variables work by default (e7b7019)
  • Make all env variables work by default (9b25b78)
  • Transfer default for experiment_uri to files (1986cd6)
  • Update dagster and add dagit (1.3.3) (66e4c9f)
  • Replace environment variable OUTPUT_PATH by EXPERIMENT_URI (d98858c)
  • Add default arguments to cli commands where reasonable (1727959)

v0.2.0 (2023-05-10)

Feature

  • Add niceml init as cli command to initialize a new niceml project (#13) (4f3090a)

Fix

  • Resize bg_images in assets for smaller wheel size (#11) (da291f4)