polygraphy.backend.tf.loader.UseTfTrt Class Reference

Inheritance diagram for polygraphy.backend.tf.loader.UseTfTrt:

Collaboration diagram for polygraphy.backend.tf.loader.UseTfTrt:

Public Member Functions
def	__init__ (self, graph, max_workspace_size=None, fp16=None, int8=None, max_batch_size=None, is_dynamic_op=False, minimum_segment_size=None)
def	__call__ (self)

Public Attributes
	max_workspace_size
	fp16
	int8
	max_batch_size
	is_dynamic_op
	minimum_segment_size

Private Attributes
	_graph

Constructor & Destructor Documentation

__init__()

def polygraphy.backend.tf.loader.UseTfTrt.__init__	(		self,
			graph,
			max_workspace_size = None,
			fp16 = None,
			int8 = None,
			max_batch_size = None,
			is_dynamic_op = False,
			minimum_segment_size = None
	)

Functor that optimizes a TensorFlow model using TF-TRT.

Args:
    graph (Callable() -> Tuple[tf.Graph, Sequence[str]]):
    A callable that can supply a tuple containing a TensorFlow graph and output names.
    max_workspace_size (int): The maximum workspace size.
    fp16 (bool): Whether to run in FP16 mode.
    max_batch_size (int): The maximum batch size.

Member Function Documentation

__call__()

def polygraphy.backend.tf.loader.UseTfTrt.__call__

(

self

)

Optimizes a TensorFlow model using TF-TRT.

Returns:
    Tuple[tf.Graph, Sequence[str]]: The TensorFlow graph, and the names of its outputs.

Member Data Documentation

_graph

polygraphy.backend.tf.loader.UseTfTrt._graph

private

max_workspace_size

polygraphy.backend.tf.loader.UseTfTrt.max_workspace_size

fp16

polygraphy.backend.tf.loader.UseTfTrt.fp16

int8

polygraphy.backend.tf.loader.UseTfTrt.int8

max_batch_size

polygraphy.backend.tf.loader.UseTfTrt.max_batch_size

is_dynamic_op

polygraphy.backend.tf.loader.UseTfTrt.is_dynamic_op

minimum_segment_size

polygraphy.backend.tf.loader.UseTfTrt.minimum_segment_size

---

The documentation for this class was generated from the following file:

• backend/tf/loader.py