chore(main): Release training 0.4.0

[DeployDuck]

🤖 Automated Release PR

This PR was created by release-please to prepare the next release. Once merged:

1. A new version tag will be created
2. A GitHub release will be published
3. The changelog will be updated

Changes to be included in the next release:

0.4.0 (2025-04-11)

⚠ BREAKING CHANGES

• models,training: temporal interpolation (#153)
• training,rollout: Rollout Schedulers (#46)
• config: Improved configuration and data structures (#79)

Features

• Add a CLI to dump the Hydra configuration files into a single YAML file. (#137) (ef1e76e)
• Add EarlyStopping Wrapper (#130) (21d06be)
• Add the possibility to train a model with a dry MLflow run ID (#164) (9849d21)
• config: Improved configuration and data structures (#79) (1f7812b)
• graphs: migrate edge builders to torch-cluster (#56) (f67da66)
• Kcrps (#182) (8bbe898)
• make colormaps configurable for groups of variable (#124) (83d72e1)
• mlflow: Make direct endpoint calls robust (#186) (77bd890)
• models,training: temporal interpolation (#153) (ea644ce)
• models: adding leaky boundings (#256) (426e860)
• training,rollout: Rollout Schedulers (#46) (08b708d)
• training: Add initial TimeLimit callback (#115) (41ff583)
• training: autoencoder 🗜️ (#172) (4684b1c)

Bug Fixes

• --longtests not available (#200) (9dfec0a)
• bugs introduced in autoencoder PR #172 (#247) (185438a)
• Checkpoint path check for multiple tasks/GPUs training (#242) (449f8bd)
• configs,schemas: hierarchical schemas (#221) (2d4a54d)
• correct config comment regarding config_validation flag (#245) (d02e0bb)
• dataset schema too defined too strictly (#143) (4792ee1)
• datashader new release test (#104) (e9c0701)
• graphs,training: skip hidden attributes (#176) (468c45a)
• lam rollout (#213) (6f78387)
• remove pydantic schemas from checkpoint pickled objects (#237) (ecb945a)
• Rename model_run_ids as trajectory_ids (#216) (e5e942d)
• training,configs: update enc-dec config (#125) (beb8c69)
• training,schema: error when using ReweightedGraphNodeAttribute (#169) (a6313ef)
• training: added weights_only=False flag to torch.load to avoid crashes in torch 2.6 (#205) (02b5117)
• training: Rework Combined Loss (#103) (b63f1aa)
• update stretched.yaml config to be consistent with schema (#195) (21255ac)
• updates to schemas to correct minor pydantic issues (#144) (4e51c17)

Documentation

• Add diagnostics to schema docs (#159) (42c5706)
• Add reference in training getting started guide for intersphinx (#220) (f1d5e1f)
• Add subheadings to schema doc page (#149) (d3c7de9)
• fix documentation to refer to anemoi datasets instead of zarr datasets (#154) (ad062b2)
• models: Docathon (#202) (5dba9d3)
• training: add ADR into docs/ (#168) (ee4da88)
• training: Docathon (#201) (e69430f)
• use new logo (#140) (c269cea)

---

Important

⚠️ Merging this PR will:

• Create a new release
• Trigger deployment pipelines
• Update package versions

Before merging:

• Ensure all tests pass
• Review the changelog carefully
• Get required approvals

Release-please documentation