Caffe2 - Python API
A deep learning, cross platform ML framework
Public Member Functions | Public Attributes | List of all members
caffe2.python.layers.adaptive_weight.AdaptiveWeight Class Reference
Inheritance diagram for caffe2.python.layers.adaptive_weight.AdaptiveWeight:

Public Member Functions

def __init__ (self, model, input_record, name="adaptive_weight", optimizer=None, weights=None, enable_diagnose=False, estimation_method="log_std", pos_optim_method="log_barrier", reg_lambda=0.1, kwargs)
def concat_data (self, net)
def log_std_init (self)
def log_std_weight (self, x, net, weight)
def log_std_reg (self, net, reg)
def inv_var_init (self)
def inv_var_weight (self, x, net, weight)
def inv_var_reg (self, net, reg)
def add_ops (self, net)
- Public Member Functions inherited from caffe2.python.layers.layers.ModelLayer
def __init__ (self, model, prefix, input_record, predict_input_record_fields=None, tags=None, kwargs)
def get_type (self)
def predict_input_record (self)
def input_record (self)
def predict_output_schema (self)
def predict_output_schema (self, output_schema)
def output_schema (self)
def output_schema (self, output_schema)
def get_parameters (self)
def get_fp16_compatible_parameters (self)
def get_memory_usage (self)
def add_init_params (self, init_net)
def create_param (self, param_name, shape, initializer, optimizer, ps_param=None, regularizer=None)
def get_next_blob_reference (self, name)
def add_operators (self, net, init_net=None, context=InstantiationContext.TRAINING)
def add_ops (self, net)
def add_eval_ops (self, net)
def add_train_ops (self, net)
def add_ops_to_accumulate_pred (self, net)
def add_param_copy_operators (self, net)
def export_output_for_metrics (self)
def export_params_for_metrics (self)

Public Attributes

- Public Attributes inherited from caffe2.python.layers.layers.ModelLayer

Detailed Description

Definition at line 16 of file

Member Function Documentation

def caffe2.python.layers.adaptive_weight.AdaptiveWeight.inv_var_init (   self)
k = 1 / variance
per task objective:
min 1 / 2 * k  X - 1 / 2 * log k

Definition at line 107 of file

def caffe2.python.layers.adaptive_weight.AdaptiveWeight.log_std_init (   self)
mu = 2 log sigma, sigma = standard variance
per task objective:
min 1 / 2 / e^mu X + mu / 2

Definition at line 76 of file

def caffe2.python.layers.adaptive_weight.AdaptiveWeight.log_std_weight (   self,
min 1 / 2 / e^mu X + mu / 2

Definition at line 94 of file

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