rllm.nn.encoder.ResNetPreEncoder¶

class rllm.nn.encoder.ResNetPreEncoder(out_dim: int, metadata: Dict[ColType, List[Dict[str, Any]]])[source]¶

Bases: TablePreEncoder

The pre-encoder for ResNet TNN.

This encoder builds column-type-specific pre-encoders, then delegates the shared table encoding pipeline to PreEncoder.

Parameters:

out_dim (int) – The output dimensionality.
metadata (Dict[ColType, List[Dict[str, Any]]]) – Metadata for each column type, specifying the statistics and properties of the columns.

Returns:

Encoded outputs are produced when inherited forward is called.

Example

>>> from rllm.nn.encoder import ResNetPreEncoder
>>> from rllm.types import ColType
>>> metadata = {
...     ColType.CATEGORICAL: [{"num_classes": 100}],
...     ColType.NUMERICAL: [{"mean": 0.0, "std": 1.0}],
... }
>>> encoder = ResNetPreEncoder(out_dim=32, metadata=metadata)