convnet_benchmarks Namespace Reference

Module caffe2.experiments.python.convnet_benchmarks. More...

Functions
def	MLP (order)
def	AlexNet (order)
def	OverFeat (order)
def	VGGA (order)
def	net_DAG_Builder (model)
def	Inception (order)
def	AddInput (model, batch_size, db, db_type)
def	AddParameterUpdate (model)
def	Benchmark (model_gen, arg)
def	GetArgumentParser ()
def	MLP (order, cudnn_ws, mkl)
def	ResNet50 (order, cudnn_ws, mkl)
def	AlexNet (order, cudnn_ws, mkl)
def	OverFeat (order, cudnn_ws, mkl)
def	VGGA (order, cudnn_ws, mkl)
def	Inception (order, cudnn_ws, mkl)

Variables
	args = GetArgumentParser().parse_args()
dictionary	model_map
	extra_args

Detailed Description

Module caffe2.experiments.python.convnet_benchmarks.

Module caffe2.python.convnet_benchmarks.

Function Documentation

def convnet_benchmarks.AddInput	(		model,
			batch_size,
			db,
			db_type
	)

Adds the data input part.

Definition at line 514 of file convnet_benchmarks.py.

def convnet_benchmarks.AddParameterUpdate

(

model

)

Simple plain SGD update -- not tuned to actually train the models 

Definition at line 527 of file convnet_benchmarks.py.

Variable Documentation

dictionary convnet_benchmarks.model_map

Initial value:

= {
        'AlexNet': AlexNet,
        'OverFeat': OverFeat,
        'VGGA': VGGA,
        'Inception': Inception,
        'MLP': MLP,
    }

Definition at line 678 of file convnet_benchmarks.py.