reference/generated/tensordict.prototype.fx.TDGraphModule

Run in Google Colab

Colab

Download Notebook

Notebook

View on GitHub

GitHub

TDGraphModule

class tensordict.prototype.fx.TDGraphModule(graph_module: GraphModule, out_keys: list[NestedKey])

Bases: Module

The module that symbolic_trace() returns: an FX graph that reads and writes a tensordict.

A call reads the in_keys of the traced module from the input tensordict, runs the graph, and writes the outputs at out_keys, into tensordict_out if given, else into the input. The graph and code attributes are those of the underlying torch.fx.GraphModule.

This class is a prototype: its API can change without a deprecation period.

Parameters:

• graph_module (GraphModule) – the traced graph.

• out_keys (list of NestedKey) – the keys where the outputs are written.

forward(tensordict: TensorDictBase, tensordict_out: TensorDictBase | None = None, **kwargs) → TensorDictBase

Define the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.