Changelog

Unreleased (latest)

Contributors: Blaise Gauvin St-Denis (@bstdenis), Trevor James Smith (@Zeitsperre), Nazim Azeli (@Nazim-crim), Francis Charette-Migneault (@fmigneault)

Changes

• Add support for Python 3.14. (PR/65).

• Add netcdf_utils module. (PR/47).

• Add timeseries module. (PR/47).

• Add rdps_to_hrdps module. (PR/47).

• Add DatasetWithSplits, DatasetWithSave and DatasetFromNetCDFSave classes to data_loader_utils module. (PR/47).

• Add dataset_manager module. (PR/47).

• Add ml_loops module. (PR/47).

• Add NeuralNetworksManager class to network_manager module. (PR/47).

• Add LinearReLU class to neural_networks_basic module. (PR/47).

• Add ml_sample_plot module. (PR/47).

• Add ml_training_plot module. (PR/47).

• Add logging_utils module. (PR/47).

• Add memory_utils module. (PR/47).

• Add utils module. (PR/47).

  • Add save_to_netcdf function to save inference results to NetCDF files.

• Add override_config_paths function to io_utils.py to allow overriding config values at runtime. (PR/47).

• Update downscaling_inference_rdps_to_hrdps.py CLI to support the following arguments for overriding config values. (PR/47).

  • --preprocess_batch: Path to the preprocessed batch file (overrides path_preprocessed_batch in config).

  • --path_models: Path to the directory containing the pre-trained model (overrides path_models in config).

  • --path_output: Path to the directory where inference results will be saved (overrides path_output in config).

• Add examples/deploy folder with example files for model deployment using Common Workflow Language and Weaver. (PR/47).

  • Add unet.cwl Common Workflow Language file for model inference.

  • Add execute_unet_cwl_schema.yml for deployment using weaver.

• Add docker folder with Dockerfile for resoterre and model inference. (PR/47).

  • Add Dockerfile.base for resoterre package

  • Add Dockerfile.inference for model inference using preprocess data.

• Add model/ entry to .gitignore to avoid model checkpoint inclusion from docker/README.md procedure.

• Add notebooks/unetToMLM.ipynb to describe the UNet downscaling model by generating a STAC Item and Collection validated with the STAC MLM and Datacube extensions. (PR/21).

• Add UNet option to go to a linear layer at the bottom for 1D inputs. (PR/23).

• Add UNet option to use inputs in the last layer for static features. (PR/23).

• Add config_utils module. (PR/23).

• Add variables module. (PR/23).

• Add hrdps_variables module. (PR/23).

• Add rdps_variables module. (PR/23).

• Add rdps_to_hrdps_workflow module. (PR/23).

• Add io_utils module. (PR/23).

• Add runner_unet module. (PR/23).

• Add snakemake_utils module. (PR/23).

Fixes

• Fix typing issues with data_loader_kwargs. (PR/64).

• Fix broken types-PyYAML dependency in environment-dev.yml. (PR/64).

• Set GitHub workflows to support Python version 3.13. (PR/22).

• Set GitHub workflows to support Python version 3.14. (PR/65).

Internal changes

• Updated the cookiecutter template. (PR/24):

  • Replaced tox.ini with new tox.toml spec

  • Enabled the labelling workflow

  • Updated pyproject.toml to use PEP 639

  • Added a CITATION.cff file

  • Replaced the python-coveralls dependency (abandoned) for the coverallsapp/github-action

  • Updated the CODE_OF_CONDUCT.md to Contributor Covenant v3.0

• Updated the cookiecutter template. (PR/65):

  • Updated GitHub Workflows to use more secure actions/connections

  • Replaced pre-commit with prek. Updates pre-commit hooks

  • GenAI contributions guidelines have been added to project files

  • dependency-groups are now used for managing development/testing/documentation dependencies (PEP 735)

  • General development/documentation dependency updates.

  • Makefile recipes now handle some dependency management commands

  • tox.toml now uses dependency-groups and Makefile

  • pytest has been updated to v9.0. Now uses TOML-based configuration spec

  • Linting tools now include vulture, deptry, yamllint, and codespell

v0.1.2 (2025-08-14)

Contributors: Blaise Gauvin St-Denis (@bstdenis)

Changes

• Add DenseUNet class to neural_networks_unet module. (PR/11)

• Add DenseUNetConfig class to neural_networks_unet module. (PR/11)

• Refactor handling of initialization functions in neural network modules. (PR/11)

• Add data_loader_utils module. (PR/11)

v0.1.1 (2025-07-29)

Contributors: Blaise Gauvin St-Denis (@bstdenis), Trevor James Smith (@Zeitsperre).

Changes

• Add network_manager module. (PR/8).

  • nb_of_parameters function to count the number of parameters in a network.

• Add neural_networks_basic module. (PR/8).

  • ModuleWithInitTracker and ModuleInitFnTracker classes to track module initialization functions.

  • SEBlock class for Squeeze-and-Excitation blocks.

• Add neural_networks_unet module. (PR/8).

  • UNet class for U-Net architecture.

• First release of resoterre on PyPI.