Mercurial > pylearn
changeset 785:12f587e37ee3
* fixed test_stacker.py
* required updating stacker by: removing kits, reworking so that it work with
"new" version of BinRegressor, creating new methods instead of using .duplicate
| author | desjagui@atchoum.iro.umontreal.ca |
|---|---|
| date | Mon, 29 Jun 2009 11:55:29 -0400 |
| parents | ba65e95d1221 |
| children | 0eb53b967ee7 |
| files | pylearn/algorithms/stacker.py pylearn/algorithms/tests/test_stacker.py |
| diffstat | 2 files changed, 19 insertions(+), 15 deletions(-) [+] |
line wrap: on
line diff
--- a/pylearn/algorithms/stacker.py Mon Jun 29 09:49:28 2009 -0400 +++ b/pylearn/algorithms/stacker.py Mon Jun 29 11:55:29 2009 -0400 @@ -25,7 +25,7 @@ for i, (submodule, outname) in enumerate(submodules): layer = submodule(current, regularize = regularize) layers.append(layer) - current = layer[outname] + current = getattr(layer, outname) self.layers = layers self.input = self.layers[0].input @@ -35,16 +35,14 @@ local_update = [] global_update = [] to_update = [] - all_kits = [] for layer, (submodule, outname) in zip(layers, submodules): u = layer.update u.resolve_all() to_update += u.updates.keys() - all_kits += u.kits # the input is the whole deep model's input instead of the layer's own # input (which is previous_layer[outname]) inputs = [self.input] + u.inputs[1:] - method = theano.Method(inputs, u.outputs, u.updates, u.kits) + method = theano.Method(inputs, u.outputs, u.updates) local_update.append(method) global_update.append( theano.Method(inputs, @@ -52,9 +50,8 @@ # we update the params of the previous layers too but wrt # this layer's cost dict((param, param - layer.lr * T.grad(layer.cost, param)) - for param in to_update), - list(all_kits))) - representation.append(theano.Method(self.input, layer[outname])) + for param in to_update))) + representation.append(theano.Method(self.input, getattr(layer,outname))) # @todo: Add diagnostics # self.diagnose_from_input = Method([self.input], self.layers[0].diagnose.outputs + self.layers[1].diagnose.outputs ... @@ -64,12 +61,16 @@ self.representation = representation self.update = self.global_update[-1] self.compute = theano.Method(self.input, self.output) + + # takes method from last layer (usually ll.classify), copies it to self., + # while converting its input to deal with the global "model" input ll = self.layers[-1] - for name, method in ll.components_map(): + for name, method in ll.__dict__['local_attr'].iteritems(): if isinstance(method, theano.Method) and not hasattr(self, name): - m = method.dup() - m.resolve_all() - m.inputs = [self.input if x is ll.input else x for x in m.inputs] + if not isinstance(method.inputs, (list,dict)): + method.inputs = [method.inputs] + inputs = [self.input if x is ll.input else x for x in method.inputs] + m = theano.Method(inputs, method.outputs, method.updates) setattr(self, name, m) def _instance_initialize(self, obj, nunits = None, lr = 0.01, seed = None, **kwargs):
--- a/pylearn/algorithms/tests/test_stacker.py Mon Jun 29 09:49:28 2009 -0400 +++ b/pylearn/algorithms/tests/test_stacker.py Mon Jun 29 11:55:29 2009 -0400 @@ -5,12 +5,15 @@ import numpy import time +class StackBinRegressor(models_reg.BinRegressor): + def __init__(self, input = None, target = None, regularize = True): + super(StackBinRegressor, self).__init__(input, target, regularize) + self.build_extensions() def test_train(mode = theano.Mode('c|py', 'fast_run')): - - reg = models_stacker.Stacker([(models_reg.BinRegressor, 'output'), - (models_reg.BinRegressor, 'output')], - regularize = False) + reg = models_stacker.Stacker([(StackBinRegressor, 'output'), + (StackBinRegressor, 'output')], + regularize = False) #print reg.global_update[1].pretty(mode = mode.excluding('inplace')) model = reg.make([100, 200, 1],