Functions — github.com/cbfinn/maml

CodeHub MCPcopy Create free account

hub / github.com/cbfinn/maml / functions

Functions19 in github.com/cbfinn/maml

⨍ Functions 19 ◇ Types & classes 2

↓ 4 callersFunctionconv_block

Perform, conv, batch norm, nonlinearity, and max pool

↓ 3 callersFunctionnormalize

(inp, activation, reuse, scope)

↓ 2 callersMethodconstruct_model

(self, input_tensors=None, prefix='metatrain_')

↓ 2 callersMethodmake_data_tensor

(self, train=True)

data_generator.py:82

↓ 1 callersFunctionget_images

(paths, labels, nb_samples=None, shuffle=True)

↓ 1 callersFunctionmain

↓ 1 callersFunctiontest

(model, saver, sess, exp_string, data_generator, test_num_updates=None)

↓ 1 callersFunctiontrain

(model, saver, sess, exp_string, data_generator, resume_itr=0)

Function_MaxPoolGradGrad

special_grads.py:7

Args: num_samples_per_class: num samples to generate per class in one batch batch_size: size of meta batch size (e.g.

data_generator.py:17

must call construct_model() after initializing MAML!

Methodconstruct_conv_weights

Methodconstruct_fc_weights

Methodforward_conv

(self, inp, weights, reuse=False, scope='')

Methodforward_fc

(self, inp, weights, reuse=False)

Methodgenerate_sinusoid_batch

(self, train=True, input_idx=None)

data_generator.py:162

Methodtask_metalearn

Perform gradient descent for one task in the meta-batch.