# metadata_logger

Python API: `luxonis_train.callbacks.metadata_logger`

Logs chosen config values as hyperparameters and saves them to `metadata.yaml`.

## Classes

### MetadataLogger

Callback that logs chosen config values as hyperparameters.

When a fit starts, the callback reads each key of `hyperparams` from the config. It logs the values with the logger of the model
and saves them to `metadata.yaml` in the save directory of the model.

The callback is in the `CALLBACKS` registry, so a config can add it:

```yaml
trainer:
  callbacks:
    - name: MetadataLogger
      params:
        hyperparams: ["trainer.epochs", "trainer.batch_size"]
```

#### Methods

##### init

```python
def __init__(hyperparams: list[str]):
```

Initialize the callback.

Parameters

 * `hyperparams` (`list[str]`): The config keys to log. A key separates its levels with dots, for example `"trainer.epochs"`. A
   level of a list is an integer index, for example `"model.nodes.0.name"`.

##### on_fit_start

```python
def on_fit_start(_: pl.Trainer, pl_module: lxt.LuxonisLightningModule):
```

Log and save the chosen config values.

Lightning calls this hook at the start of a fit. The hook reads each key of `hyperparams` with `cfg.get`. A key that is not in the
config gives `None`. `cfg.get` raises `ValueError` for a level of a list that is not an integer. The hook passes the dictionary of
keys and values to `log_hyperparams` of `pl_module.logger`. Then it writes the dictionary with `yaml.safe_dump` to `metadata.yaml`
in `pl_module.save_dir` and replaces an existing file. `yaml.safe_dump` raises `yaml.representer.RepresenterError` for a value
that is not a plain YAML type, for example a nested config section.

Parameters

 * `_` (`pl.Trainer`): The trainer. Unused.
 * `pl_module` (`lxt.LuxonisLightningModule`): The model. It gives the config, the logger, and the save directory.