> ## Documentation Index
> Fetch the complete documentation index at: https://nixtlaverse.nixtla.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Callbacks

> Utility functions use in the predict step.

##

### `SaveFeatures`

```python theme={null}
SaveFeatures()
```

Saves the features in every timestamp.

#### `SaveFeatures.get_features`

```python theme={null}
get_features(with_step=False)
```

Retrieves the input features for every timestep

**Parameters:**

| Name | Type | Description | Default |
| - | - | - | - |
| `with_step` | `bool` | Add a column indicating the step. Defaults to False. | `False` |

**Returns:**

| Type | Description |
| - | - |
| `pandas or polars DataFrame` | DataFrame with input features |


## Related topics

- [Predict callbacks](https://nixtlaverse.nixtla.io/mlforecast/docs/how-to-guides/predict_callbacks.html.md)
- [LightGBMCV](https://nixtlaverse.nixtla.io/mlforecast/lgb_cv.html.md)
- [Save and Load Models](https://nixtlaverse.nixtla.io/neuralforecast/docs/capabilities/save_load_models.html.md)
- [Distributed Forecast](https://nixtlaverse.nixtla.io/mlforecast/distributed.forecast.html.md)
- [Core | MLForecast](https://nixtlaverse.nixtla.io/mlforecast/core.html.md)
