model_helper.ModelHelperBase Class Reference

Inheritance diagram for model_helper.ModelHelperBase:

Public Member Functions
def	__init__ (self, name=None, init_params=True, allow_not_known_ops=True, skip_sparse_optim=False, param_model=None)
def	get_name (self)
def	add_param (self, param, key=None, shape=None, length=None)
def	param_info (self, grad_type=None, id=None)
def	GetParams (self, namescope=None, top_scope=False)
def	Proto (self)
def	InitProto (self)
def	RunAllOnGPU (self, args, kwargs)
def	CreateDB (self, blob_out, db, db_type, kwargs)
def	AddGradientOperators (self, args, kwargs)
def	get_param_to_grad (self, params)
def	GetOptimizationPairs (self, params=None)
def	GetComputedParams (self, namescope=None)
def	GetAllParams (self, namescope=None)
def	TensorProtosDBInput (self, unused_blob_in, blob_out, batch_size, db, db_type, kwargs)
def	AddOperator (self, op_type, inputs, parameters, args, kwargs)
def	GetDevices (self)
def	__getattr__ (self, op_type)

Public Attributes
	name
	net
	param_init_net
	param_to_grad
	params
	computed_params
	gradient_ops_added
	init_params
	allow_not_known_ops
	skip_sparse_optim
	weights
	biases
	grad_map

Detailed Description

A helper model so we can write models more easily, without having to
manually define parameter initializations and operators separately.
In order to add support for specific operators, inherit from this class
and add corresponding methods. Operator representing methods should
take care of adding their parameters to params

Definition at line 54 of file model_helper.py.

Member Function Documentation

__getattr__()

def model_helper.ModelHelperBase.__getattr__	(		self,
			op_type
	)

Catch-all for all other operators, mostly those without params.

Definition at line 284 of file model_helper.py.

AddOperator()

def model_helper.ModelHelperBase.AddOperator	(		self,
			op_type,
			inputs,
			parameters,
			args,
			kwargs
	)

Adds an operator to a model. Use parameters list
to specify which operator inputs are model parameters to be
optimized.

Example of usage:

model.SparseLengthsSum(
     [embedding, indices, lengths],
     parameters=[embedding],
)

Here embedding is a parameter to be optimized while indices
and lengths are not.

Definition at line 253 of file model_helper.py.

get_param_to_grad()

def model_helper.ModelHelperBase.get_param_to_grad	(		self,
			params
	)

Given a list of parameters returns a dict from a parameter
to a corresponding gradient

Definition at line 186 of file model_helper.py.

GetComputedParams()

def model_helper.ModelHelperBase.GetComputedParams	(		self,
			namescope = None
	)

Returns the computed params in current namescope. 'Computed params'
are such parameters that are not optimized via gradient descent but are
directly computed from data, such as the running mean and variance
of Spatial Batch Normalization.

Definition at line 220 of file model_helper.py.

GetOptimizationPairs()

def model_helper.ModelHelperBase.GetOptimizationPairs	(		self,
			params = None
	)

Returns a map for param => grad.
If params is not specified, all parameters will be considered.

Definition at line 202 of file model_helper.py.

GetParams()

def model_helper.ModelHelperBase.GetParams	(		self,
			namescope = None,
			top_scope = False
	)

Returns the params in current namescope

Definition at line 142 of file model_helper.py.

TensorProtosDBInput()

def model_helper.ModelHelperBase.TensorProtosDBInput	(		self,
			unused_blob_in,
			blob_out,
			batch_size,
			db,
			db_type,
			kwargs
	)

TensorProtosDBInput.

Definition at line 244 of file model_helper.py.

---

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

• caffe2/python/model_helper.py