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

# EarlyStopping

> Callback that monitors a metric and signals early stopping when it stops improving.

This is a dummy implementation that tracks state but doesn't actually stop
the optimization (that's handled by CallbackManager). The callback sets
a `_should_stop` flag when patience is exceeded.

## Parameters

<ResponseField name="monitor" type="str" required>
  Metric name to monitor (e.g., "system\_energy", "best\_system\_energy").
</ResponseField>

<ResponseField name="patience" type="int" required>
  Number of steps with no improvement before stopping. Ideally, this should be set as a multiple of the number of Minimizer steps necessary to mutate the entire sequence (for the largest State).
</ResponseField>

<ResponseField name="min_delta" type="float" default="0.0">
  Minimum change in the monitored metric to be considered an improvement.
</ResponseField>

<ResponseField name="mode" type="str" default="'min'">
  Whether lower ("min") or higher ("max") values are better.
</ResponseField>

## Methods

### on\_optimization\_start

Reset tracking state at the start of optimization.

**Parameters**

<ResponseField name="context" type="CallbackContext" required />

### on\_step\_end

Check if monitored metric has improved and update tracking state.

**Parameters**

<ResponseField name="context" type="CallbackContext" required />

## Example

```python theme={null}
early_stop = EarlyStopping(
    monitor="best_system_energy",
    patience=100,
    mode="min"
)
```