Revert " try use eval instead of function"

This reverts commit 171baab.

Author: chaosconst

code/DBN.py

@@ -12,7 +12,7 @@
 import theano.tensor as T
 from theano.tensor.shared_randomstreams import RandomStreams
 
-from logistic_sgd_kaggle import LogisticRegression, load_data
+from logistic_sgd import LogisticRegression, load_data
 from mlp import HiddenLayer
 from rbm import RBM
 

code/convolutional_mlp_kaggle.py

@@ -34,7 +34,7 @@
 from theano.tensor.signal import downsample
 from theano.tensor.nnet import conv
 
-from logistic_sgd_kaggle import LogisticRegression, load_data
+from logistic_sgd import LogisticRegression, load_data
 from mlp import HiddenLayer
 
 
@@ -196,6 +196,10 @@ def evaluate_lenet5(learning_rate=0.1, n_epochs=200,
                 x: test_set_x[index * batch_size: (index + 1) * batch_size],
                 y: test_set_y[index * batch_size: (index + 1) * batch_size]})
 
+    predict_model = theano.function([index], layer3.predict(),
+             givens={
+                x: predict_set_x[index * batch_size: (index + 1) * batch_size]})
+
     validate_model = theano.function([index], layer3.errors(y),
             givens={
                 x: valid_set_x[index * batch_size: (index + 1) * batch_size],
@@ -282,8 +286,8 @@ def evaluate_lenet5(learning_rate=0.1, n_epochs=200,
                     test_losses = [test_model(i) for i in xrange(n_test_batches)]
                     test_score = numpy.mean(test_losses)
 
-                    predict_res_array = [layer3.y_pred.eval({input:predict_set_x[i * batch_size: (i + 1) * batch_size]})
-                    
+                    predict_res_array = [predict_model(i) for i in xrange(n_predict_batches)]
+                    print predict_res_array;
                     f = open("predict_res","w+");
                     for y_pred_item_array in predict_res_array:
                       for y_pred_item in y_pred_item_array:

code/logistic_sgd_kaggle.py

@@ -121,9 +121,11 @@ def negative_log_likelihood(self, y):
         # LP[n-1,y[n-1]]] and T.mean(LP[T.arange(y.shape[0]),y]) is
         # the mean (across minibatch examples) of the elements in v,
         # i.e., the mean log-likelihood across the minibatch.
+        return -T.mean(T.log(self.p_y_given_x)[T.arange(y.shape[0]), y])
 
-        return -T.mean(T.log(self.p_y_given_x)[T.arange(y.shape[0]), y]) + 0.0001*T.sum(self.W **2) 
-
+    def predict(self):
+        return T.mul(self.y_pred,1)
+        
     def errors(self, y):
         """Return a float representing the number of errors in the minibatch
         over the total number of examples of the minibatch ; zero one
@@ -192,7 +194,7 @@ def load_data(dataset):
     test_set_size = 1000;
     predict_set_size = 28000;
 
-    debug = "true";
+    debug = "false";
     if debug == "true":
       train_set_size = 3600;
       valid_set_size = 500;
@@ -283,7 +285,7 @@ def shared_dataset(data_xy, borrow=True):
 
 def sgd_optimization_mnist(learning_rate=0.13, n_epochs=1000,
                            dataset='../data/mnist.pkl.gz',
-                           batch_size=500):
+                           batch_size=600):
     """
     Demonstrate stochastic gradient descent optimization of a log-linear
     model
@@ -342,6 +344,11 @@ def sgd_optimization_mnist(learning_rate=0.13, n_epochs=1000,
                 x: test_set_x[index * batch_size: (index + 1) * batch_size],
                 y: test_set_y[index * batch_size: (index + 1) * batch_size]})
 
+    release_output = theano.function(inputs=[index],
+            outputs=classifier.release_output(),
+            givens={
+                x: predict_set_x[index:]})
+
     validate_model = theano.function(inputs=[index],
             outputs=classifier.errors(y),
             givens={
@@ -421,11 +428,9 @@ def sgd_optimization_mnist(learning_rate=0.13, n_epochs=1000,
                     test_losses = [test_model(i)
                                    for i in xrange(n_test_batches)]
                     test_score = numpy.mean(test_losses)
-                    
-                    y_pred_show = classifier.y_pred.eval({input:predict_set_x})
 
-                    y_pred_show = [classifier.y_pred.eval({input:predict_set_x[index * batch_size: (index + 1) * batch_size]})
-                                   for index in xrange(n_predict_batches)]
+                    y_pred_show = [release_output(i)
+                                   for i in xrange(n_predict_batches)]
 
                     print y_pred_show;
                     f = open("predict_res","w+");