TF 2.0 is out! Get hands-on practice at TF World, Oct 28-31. Use code TF20 for 20% off select passes. Register now

Module: tfp.experimental.auto_batching

View source on GitHub

TensorFlow Probability auto-batching package.

Modules

allocation_strategy module: Live variable analysis.

dsl module: Python-embedded DSL frontend for authoring autobatchable IR programs.

frontend module: AutoGraph-based auto-batching frontend.

instructions module: Instruction language for auto-batching virtual machine.

liveness module: Live variable analysis.

lowering module: Lowering the full IR to stack machine instructions.

numpy_backend module: Numpy backend for auto-batching VM.

stack_optimization module: Optimizing stack usage (pushes and pops).

stackless module: A stackless auto-batching VM.

tf_backend module: TensorFlow (graph) backend for auto-batching VM.

type_inference module: Type inference pass on functional control flow graph.

virtual_machine module: The auto-batching VM itself.

xla module: XLA utilities.

Classes

class Context: Context object for auto-batching multiple Python functions together.

class NumpyBackend: Implements the Numpy backend ops for a PC auto-batching VM.

class TensorFlowBackend: Implements the TF backend ops for a PC auto-batching VM.

class TensorType: TensorType(dtype, shape)

class Type: Type(tensors,)

Functions

truthy(...): Normalizes Tensor ranks for use in if conditions.