MCPcopy Create free account
hub / github.com/DeepRec-AI/DeepRec / conv

Function conv

tensorflow/compiler/tf2xla/python/xla.py:228–272  ·  view source on GitHub ↗

Wraps the XLA ConvGeneralDilated operator. ConvGeneralDilated is the most general form of XLA convolution and is documented at https://www.tensorflow.org/performance/xla/operation_semantics#conv_convolution Args: lhs: the input tensor rhs: the kernel tensor window_strides: the

(lhs,
         rhs,
         window_strides,
         padding,
         lhs_dilation,
         rhs_dilation,
         dimension_numbers,
         feature_group_count=1,
         precision_config=None,
         name=None)

Source from the content-addressed store, hash-verified

226
227
228def conv(lhs,
229 rhs,
230 window_strides,
231 padding,
232 lhs_dilation,
233 rhs_dilation,
234 dimension_numbers,
235 feature_group_count=1,
236 precision_config=None,
237 name=None):
238 """Wraps the XLA ConvGeneralDilated operator.
239
240 ConvGeneralDilated is the most general form of XLA convolution and is
241 documented at
242 https://www.tensorflow.org/performance/xla/operation_semantics#conv_convolution
243
244 Args:
245 lhs: the input tensor
246 rhs: the kernel tensor
247 window_strides: the inter-window strides
248 padding: the padding to apply at the start and end of each input dimensions
249 lhs_dilation: dilation to apply between input elements
250 rhs_dilation: dilation to apply between kernel elements
251 dimension_numbers: a `ConvolutionDimensionNumbers` proto.
252 feature_group_count: number of feature groups for grouped convolution.
253 precision_config: a `xla.PrecisionConfig` proto.
254 name: an optional name for the operator
255
256 Returns:
257 A tensor representing the output of the convolution.
258 """
259 precision_config_proto = ""
260 if precision_config:
261 precision_config_proto = precision_config.SerializeToString()
262 return gen_xla_ops.xla_conv(
263 lhs,
264 rhs,
265 window_strides=window_strides,
266 padding=padding,
267 lhs_dilation=lhs_dilation,
268 rhs_dilation=rhs_dilation,
269 feature_group_count=feature_group_count,
270 dimension_numbers=dimension_numbers.SerializeToString(),
271 precision_config=precision_config_proto,
272 name=name)
273
274
275convert_element_type = math_ops.cast

Callers 5

test_regularizationMethod · 0.85
modelMethod · 0.85
_simple_metagraphFunction · 0.85
modelMethod · 0.85

Calls 1

SerializeToStringMethod · 0.45

Tested by 5

test_regularizationMethod · 0.68
modelMethod · 0.68
_simple_metagraphFunction · 0.68
modelMethod · 0.68