[dev] C2 - Add typed configs as pydantic dataclasses & omegaconf dictconfigs by deruyter92 · Pull Request #3191 · DeepLabCut/DeepLabCut

deruyter92 · 2026-01-29T18:05:22Z

This PR is part of the WIP for migrating from dictionary configs to typed & validated configurations (see #3193 for an overview). (Credits to @arashsm79 for originally introducing these changes).

Motivation
Introduces typed, validated config classes (Pydantic + OmegaConf) and a shared ConfigMixin for project and pose configs. Replaces ad-hoc dicts with a consistent load/save/validation API and better type support.

Changes:

Core: ConfigMixin, ProjectConfig, and utils (moved from core/config.py). New core/types.py with PydanticNDArray.
Weight init: WeightInitialization is a Pydantic dataclass using ConfigMixin; snapshot_path optional; conversion_array typed as PydanticNDArray.
Pose PyTorch: New typed configs in pose_estimation_pytorch/config/ (data, inference, logger, model, pose, project, runner, training). GenSamplingConfig moved from data/generative_sampling.py to config/data.py. Inference config classes moved from runners/inference.py to config/inference.py; inference runner uses InferenceConfig.from_any().
Deps: omegaconf>=2.3.0, pydantic>=2.12.3.
Tests: tests/core/config/test_config_mixin.py for ConfigMixin.

Notes:

The typed configurations are not yet used in the pipeline, but implemented in parallel as part of the work-in-progress.
The typed configurations in this PR mimic the structure of the older dictionary configs (same fields) to facilitate easy migration. These will be updated later.

Copilot

Pull request overview

This PR introduces typed, validated configuration classes using Pydantic and OmegaConf to replace dictionary-based configurations. The changes establish a foundation for better type safety and validation while maintaining backward compatibility with existing dictionary structures.

Changes:

Adds ConfigMixin base class providing common config operations (load, save, validate, convert)
Implements typed config classes for project, data, model, training, runner, inference, and logger configurations
Migrates WeightInitialization and GenSamplingConfig to use Pydantic dataclasses
Adds omegaconf>=2.3.0 and pydantic>=2.12.3 as dependencies

Reviewed changes

Copilot reviewed 21 out of 23 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
tests/core/config/test_config_mixin.py	Comprehensive test coverage for ConfigMixin functionality
setup.py	Adds omegaconf and pydantic dependencies
deeplabcut/pose_estimation_pytorch/runners/inference.py	Removes InferenceConfig classes (moved to config module)
deeplabcut/pose_estimation_pytorch/data/generative_sampling.py	Removes GenSamplingConfig (moved to config/data.py)
deeplabcut/pose_estimation_pytorch/data/dataset.py	Updates import for GenSamplingConfig
deeplabcut/pose_estimation_pytorch/data/base.py	Updates import for GenSamplingConfig
deeplabcut/pose_estimation_pytorch/data/init.py	Updates import for GenSamplingConfig
deeplabcut/pose_estimation_pytorch/config/training.py	Adds TrainSettingsConfig dataclass
deeplabcut/pose_estimation_pytorch/config/runner.py	Adds OptimizerConfig, SchedulerConfig, SnapshotCheckpointConfig, and RunnerConfig
deeplabcut/pose_estimation_pytorch/config/project.py	Adds typed ProjectConfig matching old dictionary structure
deeplabcut/pose_estimation_pytorch/config/pose.py	Adds main PoseConfig with enums for method and network types
deeplabcut/pose_estimation_pytorch/config/model.py	Adds ModelConfig and DetectorModelConfig
deeplabcut/pose_estimation_pytorch/config/logger.py	Adds LoggerConfig classes for WandB and CSV loggers
deeplabcut/pose_estimation_pytorch/config/inference.py	Adds typed inference config classes
deeplabcut/pose_estimation_pytorch/config/data.py	Adds data config classes including GenSamplingConfig
deeplabcut/pose_estimation_pytorch/init.py	Updates import for GenSamplingConfig
deeplabcut/core/weight_init.py	Converts WeightInitialization to Pydantic dataclass with ConfigMixin
deeplabcut/core/types.py	Adds PydanticNDArray type alias for numpy array support
deeplabcut/core/config/project_config.py	Adds core ProjectConfig (duplicate of pose_estimation_pytorch version)
deeplabcut/core/config/config_mixin.py	Implements ConfigMixin base class
deeplabcut/core/config/init.py	Exports config utilities and classes

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

deeplabcut/pose_estimation_pytorch/config/project.py

deeplabcut/core/config/project_config.py

deeplabcut/pose_estimation_pytorch/config/inference.py

deeplabcut/core/config/config_mixin.py

deeplabcut/pose_estimation_pytorch/config/data.py

deeplabcut/core/config/project_config.py

deeplabcut/core/config/config_mixin.py

This mixin provides methods for: - Loading configurations from dictionaries or YAML files - Validating configuration data against pydantic models - Converting configurations to dictionaries - Pretty printing configuration data add imports from utils

- Introduced new configuration classes for inference, logging, model, pose, project, runner, and training settings. - Refactored data loading mechanisms to utilize new configuration structures. - Moved the multithreading and compilation options in inference configuration to the config module. - Typed configuration for logging. - Updated dataset loaders to accept model configurations directly or via file paths.

(The fields are kept identical to old multianimal project configs for now) move ProjectConfig to deeplabcut/core/config

…correct TypeAdapter usage)

The return value should be the dictionary, not the instantiated transforms

deruyter92 mentioned this pull request Jan 29, 2026

[WIP] migration to structured configuration classes #3172

Closed

deruyter92 force-pushed the jaap/add_typed_configs branch 3 times, most recently from 59b4623 to 5c950a5 Compare January 30, 2026 13:05

deruyter92 mentioned this pull request Jan 30, 2026

[WIP] Migration to structured configurations #3193

Open

17 tasks

deruyter92 force-pushed the jaap/add_typed_configs branch from 5c950a5 to d58be73 Compare January 30, 2026 14:02

deruyter92 mentioned this pull request Jan 30, 2026

[dev] C3 - Replace configuration dictionaries with DictConfigs. #3194

Merged

deruyter92 added WORK IN PROGRESS! developers are currently working on this feature... stay tuned. enhancement New feature or request labels Feb 2, 2026

deruyter92 marked this pull request as ready for review February 2, 2026 10:33

deruyter92 requested review from C-Achard and Copilot February 2, 2026 10:34

Copilot AI reviewed Feb 2, 2026

View reviewed changes

deruyter92 force-pushed the jaap/add_typed_configs branch 2 times, most recently from e290a0c to 8b5d13f Compare February 2, 2026 12:32

deruyter92 changed the title ~~Add typed configs as pydantic dataclasses & omegaconf dictconfigs~~ [dev] C2 - Add typed configs as pydantic dataclasses & omegaconf dictconfigs Feb 3, 2026

deruyter92 added this to the Structured configs milestone Feb 3, 2026

C-Achard reviewed Feb 3, 2026

View reviewed changes

deeplabcut/core/config/project_config.py Show resolved Hide resolved

C-Achard reviewed Feb 3, 2026

View reviewed changes

deeplabcut/core/config/config_mixin.py Show resolved Hide resolved

C-Achard approved these changes Feb 3, 2026

View reviewed changes

Base automatically changed from jaap/centralize_project_configs to feat/structured_configs February 4, 2026 08:40

deruyter92 and others added 9 commits February 4, 2026 09:45

add pytorch base config mixin

7d0c41b

This mixin provides methods for: - Loading configurations from dictionaries or YAML files - Validating configuration data against pydantic models - Converting configurations to dictionaries - Pretty printing configuration data add imports from utils

add tests for ConfigMixin

6f86d10

Simplify changes (only add typed configs)

206d86a

Add typed ProjectConfig as pydantic dataclass

cd630ea

(The fields are kept identical to old multianimal project configs for now) move ProjectConfig to deeplabcut/core/config

add pytorch configs as pydantic dataclasses

686449f

copilot suggestions (return transforms, default MultthreadingConfig, …

9217b6d

…correct TypeAdapter usage)

fix transforms return value at validation

73f880a

The return value should be the dictionary, not the instantiated transforms

add comments in yaml using CommentedMap

120aedc

deruyter92 force-pushed the jaap/add_typed_configs branch 2 times, most recently from e00941b to 120aedc Compare February 4, 2026 08:57

deruyter92 mentioned this pull request Feb 4, 2026

[WIP] Final migration to configuration version 1: structured and validated configs #3198

Draft

deruyter92 merged commit defdd32 into feat/structured_configs Feb 4, 2026
1 check failed

deruyter92 deleted the jaap/add_typed_configs branch February 4, 2026 09:09

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[dev] C2 - Add typed configs as pydantic dataclasses & omegaconf dictconfigs#3191

[dev] C2 - Add typed configs as pydantic dataclasses & omegaconf dictconfigs#3191
deruyter92 merged 9 commits intofeat/structured_configsfrom
jaap/add_typed_configs

deruyter92 commented Jan 29, 2026 •

edited

Loading

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Uh oh!

Conversation

deruyter92 commented Jan 29, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

deruyter92 commented Jan 29, 2026 •

edited

Loading