hydraflow 0.16.0__tar.gz → 0.16.1__tar.gz

{hydraflow-0.16.0 → hydraflow-0.16.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: hydraflow
-Version: 0.16.0
+Version: 0.16.1
 Summary: HydraFlow seamlessly integrates Hydra and MLflow to streamline ML experiment management, combining Hydra's configuration management with MLflow's tracking capabilities.
 Project-URL: Documentation, https://daizutabi.github.io/hydraflow/
 Project-URL: Source, https://github.com/daizutabi/hydraflow

{hydraflow-0.16.0 → hydraflow-0.16.1}/docs/part1-applications/main-decorator.md

@@ -242,6 +242,36 @@ This option is particularly useful when:
 - You're iterating on experiments with command-line variations
 - Your configuration contains volatile or automatically generated values
 
+### Dynamic Configuration Updates (`update`)
+
+Modify or enhance the configuration after it has been loaded by Hydra
+but before the run starts:
+
+```python
+def update_config(cfg: Config) -> Config:
+    # Calculate derived values or add runtime information
+    if cfg.width > 0 and cfg.height > 0:
+        cfg.area = cfg.width * cfg.height
+    return cfg
+
+@hydraflow.main(Config, update=update_config)
+def train(run: Run, cfg: Config) -> None:
+    # Configuration has been updated with calculated area
+    print(f"Area: {cfg.area}")
+```
+
+This option is powerful when you need to:
+
+- Calculate derived parameters based on existing configuration values
+- Apply conditional logic to adjust parameters based on their relationships
+- Ensure consistency between related parameters
+- Adapt configurations to the current environment (e.g., hardware capabilities)
+
+The `update` function should accept a configuration object and
+return the same object (or None).
+Any changes made to the configuration will be saved to the run's configuration file,
+ensuring that the stored configuration accurately reflects all updates.
+
 ## Best Practices
 
 1. **Keep Configuration Classes Focused**: Break down complex configurations

{hydraflow-0.16.0 → hydraflow-0.16.1}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "hydraflow"
-version = "0.16.0"
+version = "0.16.1"
 description = "HydraFlow seamlessly integrates Hydra and MLflow to streamline ML experiment management, combining Hydra's configuration management with MLflow's tracking capabilities."
 readme = "README.md"
 license = { file = "LICENSE" }

{hydraflow-0.16.0 → hydraflow-0.16.1}/src/hydraflow/core/main.py

@@ -82,7 +82,12 @@ def main[C](
         rerun_finished: If True, allows rerunning completed runs. Defaults to
             False.
         update: A function that takes a configuration and returns a new
-            configuration. Defaults to None.
+            configuration or None. The function can modify the configuration in-place
+            and/or return it. If the function returns None, the original (potentially
+            modified) configuration is used. Changes made by this function are saved
+            to the configuration file. This is useful for adding derived parameters,
+            ensuring consistency between related values, or adding runtime information
+            to the configuration. Defaults to None.
 
     """
     import mlflow