eugene.models.load_config¶

eugene.models.load_config(config_path, **kwargs)¶

Instantiate a module or architecture from a config file

This function is used to instantiate a module or architecture from a config file. The config file must be a YAML file with parameters from the module or architecture as well as the name of the module or architecture. For example, to instantiate a CNN within a SequenceModule, the config file might look like this:

```yaml module: SequenceModule model:

model_name: simple_cnn arch_name: CNN arch:

input_len: 100 output_dim: 1 conv_kwargs:

input_channels: 4 conv_channels: [32] conv_kernels: [13] conv_strides: [1] pool_kernels: [2] pool_strides: [2] dropout_rates: 0.3 batchnorm: True activations: relu

dense_kwargs:
hidden_dims: [64] dropout_rates: 0.2 batchnorm: True

task: regression loss_fxn: mse optimizer: adam optimizer_lr: 0.001 ```

The module parameter is the name of the LightningModule to instantiate in eugene.models. The arch_name parameter is the name of the architecture to instantiate in eugene.models.zoo. The arch parameter contains all the arguments for the CNN class in eugene.models.zoo._basic_models The conv_kwargs and dense_kwargs are Conv1DTower and DenseBlock respectively in eugene.models.base The parameters task, loss_fxn, optimizer, and optimizer_lr are all arguments for SequenceModule.

If a “module” parameter is not passed in, this function assumes that we just want to instantiate an architecture. For example, to instantiate a CNN, the config file might look like this:

```yaml model:

model_name: simple_cnn arch_name: CNN arch:

input_len: 100 output_dim: 1 conv_kwargs:

input_channels: 4 conv_channels: [32] conv_kernels: [13] conv_strides: [1] pool_kernels: [2] pool_strides: [2] dropout_rates: 0.3 batchnorm: True activations: relu

dense_kwargs:
hidden_dims: [64] dropout_rates: 0.2 batchnorm: True

```

where we have removed the module, task, loss_fxn, optimizer, and optimizer_lr parameters. This will return an instance of the CNN class in eugene.models.zoo._basic_models as an nn.Module.

Parameters:

config_path (str or PathLike) – Path to a YAML config file
**kwargs – Additional keyword arguments to pass to the module or architecture

Return type:

Union[SequenceModule, ProfileModule, nn.Module]