Integrated sknn into the cifar10 benchmark, runs in 280s on this CPU!…

… Fixed error in MLP logging.

examples/bench_cifar10.py

@@ -20,12 +20,18 @@ def load(name):
 n_feat = data_train.shape[1]
 n_targets = labels_train.max() + 1
 
-net = DBN(
-    [n_feat, n_feat / 3, n_targets],
-    epochs=50,
-    learn_rates=0.03,
-    verbose=1,
-    )
+import sys
+import logging
+logging.basicConfig(format="%(message)s", level=logging.DEBUG, stream=sys.stdout)
+
+from sknn.mlp import MultiLayerPerceptronClassifier
+net = MultiLayerPerceptronClassifier(
+    [("Rectifier", n_feat*2/3), ("Rectifier", n_feat*1/3), ("Linear", n_targets)],
+    n_iter=50,
+    n_stable=10,
+    learning_rate=0.005,
+    valid_size=0.1,
+    verbose=1)
 net.fit(data_train, labels_train)
 
 from sklearn.metrics import classification_report

sknn/mlp.py

@@ -397,7 +397,7 @@ def _fit(self, X, y, test=None):
         log.info("Training on dataset of {:,} samples with {:,} total size.".format(num_samples, data_size))
         if self.valid_set:
             X_v, _ = self.valid_set
-            log.debug("  - Test: {: <10,}  Valid: {: <4,}".format(X.shape[0], X_v.shape[0]))
+            log.debug("  - Train: {: <9,}  Valid: {: <4,}".format(X.shape[0], X_v.shape[0]))
         if self.n_iter:
             log.debug("  - Terminating loop after {} total iterations.".format(self.n_iter))
         if self.n_stable: