docs: add initial Jupyter Book documentation scaffold

docs/_config.yml

@@ -0,0 +1,12 @@
+title: neural-lam Documentation
+author: mllam contributors
+
+execute:
+  execute_notebooks: off
+
+repository:
+  url: https://github.com/mllam/neural-lam
+
+html:
+  use_repository_button: true
+  use_issues_button: true

docs/_toc.yml

@@ -0,0 +1,6 @@
+format: jb-book
+root: index
+
+chapters:
+  - file: quickstart
+  - file: create_reduced_meps_dataset

docs/index.md

@@ -0,0 +1,13 @@
+# neural-lam Documentation
+
+Welcome to the official documentation for neural-lam.
+
+This documentation provides:
+
+- A Quickstart guide for new users
+- Dataset preparation notebooks
+- API reference (auto-generated from docstrings)
+
+## Getting Started
+
+Start with the [Quickstart Guide](quickstart.md).

docs/quickstart.md

@@ -0,0 +1,50 @@
+# Quickstart Guide
+
+This guide demonstrates how to use neural-lam for computing evaluation metrics.
+
+## Installation
+
+Clone the repository and install in editable mode:
+
+```bash
+git clone https://github.com/mllam/neural-lam.git
+cd neural-lam
+pip install -e .
+```
+
+## Minimal Example
+
+```python
+import torch
+from neural_lam.metrics import mse, get_metric
+
+# Dummy predictions
+pred = torch.randn(2, 10, 3)
+target = torch.randn(2, 10, 3)
+pred_std = torch.ones_like(pred)
+
+# Compute MSE
+loss = mse(pred, target, pred_std)
+print("MSE:", loss)
+```
+
+## Using Registered Metrics
+
+You can also retrieve metrics dynamically:
+
+```python
+metric_fn = get_metric("wmse")
+value = metric_fn(pred, target, pred_std)
+print("Weighted MSE:", value)
+```
+
+## Available Metrics
+
+Currently registered metrics:
+
+- mse
+- wmse
+- mae
+- wmae
+- nll
+- crps_gauss