conv Namespace Reference

Module caffe2.python.helpers.conv. More...

Functions
def	ConvNd (model, blob_in, blob_out, dim_in, dim_out, kernel, weight_init=None, bias_init=None, group=1, transform_inputs=None, order="NCHW", kwargs)
def	Conv (model, blob_in, blob_out, dim_in, dim_out, kernel, weight_init=None, bias_init=None, group=1, transform_inputs=None, kwargs)
def	ConvTranspose (model, blob_in, blob_out, dim_in, dim_out, kernel, weight_init=None, bias_init=None, use_cudnn=False, order="NCHW", cudnn_exhaustive_search=False, ws_nbytes_limit=None, kwargs)
def	GroupConv (model, blob_in, blob_out, dim_in, dim_out, kernel, weight_init=None, bias_init=None, group=1, kwargs)
def	GroupConv_Deprecated (model, blob_in, blob_out, dim_in, dim_out, kernel, weight_init=None, bias_init=None, group=1, use_cudnn=False, order="NCHW", cudnn_exhaustive_search=False, ws_nbytes_limit=None, kwargs)

Detailed Description

Module caffe2.python.helpers.conv.

Function Documentation

Conv()

def conv.Conv	(		model,
			blob_in,
			blob_out,
			dim_in,
			dim_out,
			kernel,
			weight_init = None,
			bias_init = None,
			group = 1,
			transform_inputs = None,
			kwargs
	)

2-dimensional convolution.

Definition at line 116 of file conv.py.

ConvNd()

def conv.ConvNd	(		model,
			blob_in,
			blob_out,
			dim_in,
			dim_out,
			kernel,
			weight_init = None,
			bias_init = None,
			group = 1,
			transform_inputs = None,
			order = "NCHW",
			kwargs
	)

N-dimensional convolution for inputs with NCHW storage order.

Definition at line 106 of file conv.py.

ConvTranspose()

def conv.ConvTranspose	(		model,
			blob_in,
			blob_out,
			dim_in,
			dim_out,
			kernel,
			weight_init = None,
			bias_init = None,
			use_cudnn = False,
			order = "NCHW",
			cudnn_exhaustive_search = False,
			ws_nbytes_limit = None,
			kwargs
	)

ConvTranspose.

Definition at line 127 of file conv.py.

GroupConv()

def conv.GroupConv	(		model,
			blob_in,
			blob_out,
			dim_in,
			dim_out,
			kernel,
			weight_init = None,
			bias_init = None,
			group = 1,
			kwargs
	)

Group Convolution.

This is essentially the same as Conv with a group argument passed in.
We specialize this for backward interface compatibility.

Definition at line 183 of file conv.py.

GroupConv_Deprecated()

def conv.GroupConv_Deprecated	(		model,
			blob_in,
			blob_out,
			dim_in,
			dim_out,
			kernel,
			weight_init = None,
			bias_init = None,
			group = 1,
			use_cudnn = False,
			order = "NCHW",
			cudnn_exhaustive_search = False,
			ws_nbytes_limit = None,
			kwargs
	)

GroupConvolution's deprecated interface.

This is used to simulate a group convolution via split and concat. You
should always use the new group convolution in your new code.

Definition at line 207 of file conv.py.