Menú
Hydra configuration file structure and conventions for models, datasets, callbacks, and training
Instalar con Codex o Claude Copia este prompt, pégalo en Codex, Claude u otro asistente, y deja que revise la página de la skill y la instale por ti.
Basado en la clasificación ocupacional SOC
Git operations: branching, commit message format, pre-commit hook handling, and push policy
Full workflow: environment setup, running experiments, reusing and implementing new components (models, datasets, plmodules, callbacks), and testing
How to load a checkpoint and run evaluation (inference only, no training)
Reference code folder conventions — storing external code and documentation for AI-assisted development context
Coding style guidelines and common pitfalls for this project
Data format conventions for datasets and plmodules
| name | configs |
| description | Hydra configuration file structure and conventions for models, datasets, callbacks, and training |
All configuration files live in configs/, organized using the Hydra framework.
accelerate/ — Distributed training accelerationcallbacks/ — Lightning callbacks (e.g., model summary, checkpointing)data/ — Dataset and dataloader configurationsdatamodule/ — PyTorch Lightning datamodule configurationsexperiment/ — Experiment-specific overridesextras/ — Additional utility configurationshydra/ — Hydra framework configurationslogger/ — Logging configurations (e.g., TensorBoard, WandB)loss/ — Loss function configurationsmodel/ — Model configurationsopt/ — Optimizer configurationspaths/ — Path configurationsplmodule/ — PyTorch Lightning module configurationstrainer/ — Trainer configurationsTop-level training config files (e.g., train_nop.yaml, train_nop_rollout.yaml) specify which configs to use for a given project.
There are two types of callback config files in configs/callbacks/:
single_callback_name.yaml — Configuration for a single callback.many_callbacks_project_name.yaml — Lists all callback configs to be used in a project.For a new project, create a many_callbacks_project_name.yaml listing the individual callback configs. If you add a new callback, also create a corresponding single_callback_name.yaml.