fc Namespace Reference

Module caffe2.python.helpers.fc. More...

Functions
def	FC (model, args, kwargs)
def	PackedFC (model, args, kwargs)
def	FC_Decomp (model, blob_in, blob_out, dim_in, dim_out, rank_approx=5, weight_init=None, bias_init=None, kwargs)
def	FC_Prune (model, blob_in, blob_out, dim_in, dim_out, weight_init=None, bias_init=None, mask_init=None, threshold=0.00001, need_compress_rate=False, comp_lb=0.05, kwargs)
def	FC_Sparse (model, blob_in, blob_out, w_csr, iw, jw, bias, kwargs)

Detailed Description

Module caffe2.python.helpers.fc.

Module caffe2.python.layers.fc.

Module caffe2.python.helpers.pooling.

Function Documentation

FC_Decomp()

def fc.FC_Decomp	(		model,
			blob_in,
			blob_out,
			dim_in,
			dim_out,
			rank_approx = 5,
			weight_init = None,
			bias_init = None,
			kwargs
	)

FC_Decomp version
Here we assume that the rank of original input is bigger than 5.

Definition at line 60 of file fc.py.

FC_Prune()

def fc.FC_Prune	(		model,
			blob_in,
			blob_out,
			dim_in,
			dim_out,
			weight_init = None,
			bias_init = None,
			mask_init = None,
			threshold = 0.00001,
			need_compress_rate = False,
			comp_lb = 0.05,
			kwargs
	)

FC_Prune version
Runnable so far. Great!:)

Definition at line 95 of file fc.py.

FC_Sparse()

def fc.FC_Sparse	(		model,
			blob_in,
			blob_out,
			w_csr,
			iw,
			jw,
			bias,
			kwargs
	)

FC_Sparse: Only takes in alocated weights

Definition at line 178 of file fc.py.