Fix links in docs. Add --upgrade to mkdocs deploy requirements install.

.github/workflows/docs-publish.yml

@@ -24,5 +24,5 @@ jobs:
           path: .cache
           restore-keys: |
             mkdocs-material-
-      - run: pip install mkdocs-material mkdocs-autorefs mkdocstrings mkdocs-gen-files mkdocs-literate-nav mkdocs-section-index mkdocstrings-python
+      - run: pip install --upgrade mkdocs-material mkdocs-autorefs mkdocstrings mkdocs-gen-files mkdocs-literate-nav mkdocs-section-index mkdocstrings-python
       - run: mkdocs gh-deploy --force

docs/how-to/adapters.md

@@ -164,4 +164,4 @@ Adapters are a cornerstone of Lighter's flexibility and extensibility. By using
 
 With adapters, you adapt Lighter to your specific research needs and build complex, yet well-organized and maintainable deep learning experiment configurations.
 
-Next, explore the [Writers](05_using_and_extending_writers.md) to learn how to save model predictions and outputs to files. For a further read on Lighter's design principles and why we designed adapters, check out the [Design section](../design/overview.md).
+Next, explore the [Writers](writers.md) to learn how to save model predictions and outputs to files. For a further read on Lighter's design principles and why we designed adapters, check out the [Design section](../design/overview.md).

docs/how-to/configure.md

@@ -54,7 +54,7 @@ In this example, we define a simple linear model, a cross-entropy loss, and an A
 In addition to mandatory `trainer` and `system` sections, you can include the following optional sections: 
 
 *   **`_requires_`**: Evaluated before the rest of the config. Useful for importing modules used by Python expressions in the config as explained in [Evaluating Python Expressions](#evaluating-python-expressions).
-*   **`project`**: Path to your project directory. Used to import custom modules. For more details, see [Custom Project Modules](../how-to/01_custom_project_modules.md).
+*   **`project`**: Path to your project directory. Used to import custom modules. For more details, see [Custom Project Modules](../how-to/project_module.md).
 *   **`vars`**: Store variables for use in other parts of the config. Useful to avoid repetition and easily update values. See [Referencing Other Components](#referencing-other-components).
 *   **`args`**: Arguments to pass to the the stage of the experiment being run.
 

docs/how-to/freezers.md

@@ -120,4 +120,4 @@ Example: 2 `Freezer` callbacks for gradual unfreezing - initial backbone freeze,
 
 Lighter `Freezer` callback: flexible, fine-grained model layer training control via `config.yaml`. Optimize training, performance, pre-trained knowledge.
 
-Next: [Custom Inferer How-To guide](07_using_inferers.md), [How-To guides](../how-to/01_custom_project_modules.md), [Design section](../design/01_overview.md), [Tutorials section](../tutorials/01_configuration_basics.md).
+Next: [Inferers](inferers.md).

docs/index.md

@@ -175,7 +175,7 @@ See how training a model on CIFAR-10 differs between Lighter and PyTorch Lightni
     ---
 
     Learn about Lighter's advanced features with practical guides
-    [:octicons-arrow-right-24: Learn More](how-to/custom_project_modules.md)
+    [:octicons-arrow-right-24: Learn More](how-to/project_module.md)
 
 -   :material-lightbulb:{ .lg .middle }  __Design__