Update for common tests

Author: vene

polylearn/adagrad_fast.cpp

@@ -236,4 +236,5 @@ def _fast_fm_adagrad(self,
                     sync(&P[s, j, o], &P_last_seen[s, j, o],
                          P_grad_norms[s, j, o], learning_rate,
                          beta, t)
+    return it
 

polylearn/adagrad_fast.pyx

@@ -195,4 +195,4 @@ def _cd_direct_ho(double[:, :, ::1] P not None,
             converged = True
             break
 
-    return converged
+    return converged, it

polylearn/cd_direct_fast.cpp

@@ -177,4 +177,4 @@ def _cd_lifted(double[:, :, ::1] U,
             converged = True
             break
 
-    return converged
+    return converged, it

polylearn/cd_direct_fast.pyx

@@ -130,34 +130,27 @@ def fit(self, X, y):
 
             y_pred = self._get_output(X)
 
-            converged = _cd_direct_ho(self.P_, self.w_, dataset, X_col_norms,
-                                      y, y_pred, self.lams_, self.degree,
-                                      alpha, beta, self.fit_linear,
-                                      self.fit_lower == 'explicit', loss_obj,
-                                      self.max_iter, self.tol, self.verbose)
+            converged, self.n_iter_ = _cd_direct_ho(
+                self.P_, self.w_, dataset, X_col_norms, y, y_pred,
+                self.lams_, self.degree, alpha, beta, self.fit_linear,
+                self.fit_lower == 'explicit', loss_obj, self.max_iter,
+                self.tol, self.verbose)
             if not converged:
                 warnings.warn("Objective did not converge. Increase max_iter.")
 
         elif self.solver == 'adagrad':
-            # if self.fit_lower == 'explicit' and self.degree > 2:
-            #     raise NotImplementedError("Adagrad solver currently doesn't "
-            #                               "support `fit_lower='explicit'`.")
 
             if self.init_lambdas != 'ones':
                 raise NotImplementedError("Adagrad solver currently doesn't "
                                           "support `init_lambdas != 'ones'`.")
 
             dataset = get_dataset(X, order="c")
-            # P = np.transpose(self.P_, [1, 2, 0])
-            # P = np.asfortranarray(P)
-            # print(P.shape, P.flags)
-            #
+
             self.P_ = np.asfortranarray(np.transpose(self.P_, [1, 2, 0]))
-            _fast_fm_adagrad(self, self.w_, self.P_, dataset, y,
-                             self.degree, alpha, beta, self.fit_linear,
-                             self.fit_lower == 'explicit', loss_obj,
-                             self.max_iter, self.learning_rate,
-                             self.callback, self.n_calls)
+            self.n_iter_ = _fast_fm_adagrad(
+                self, self.w_, self.P_, dataset, y, self.degree, alpha,
+                beta, self.fit_linear, self.fit_lower == 'explicit', loss_obj,
+                self.max_iter, self.learning_rate, self.callback, self.n_calls)
             self.P_ = np.transpose(self.P_, [2, 0, 1])
         return self
 

polylearn/cd_lifted_fast.cpp

@@ -89,8 +89,9 @@ def fit(self, X, y):
 
         y_pred = _lifted_predict(self.U_, dataset)
 
-        converged = _cd_lifted(self.U_, dataset, y, y_pred, self.beta,
-                               loss_obj, self.max_iter, self.tol, self.verbose)
+        converged, self.n_iter_ = _cd_lifted(
+            self.U_, dataset, y, y_pred, self.beta, loss_obj, self.max_iter,
+            self.tol, self.verbose)
 
         if not converged:
             warnings.warn("Objective did not converge. Increase max_iter.")

polylearn/cd_lifted_fast.pyx

@@ -12,17 +12,15 @@
 
 def test_check_estimator():
     # TODO: classifiers that provide predict_proba but are not multiclass fail
-    # yield check_estimator, PolynomialNetworkClassifier
-    # yield check_estimator, FactorizationMachineClassifier
-
-    raise SkipTest()  # temporary
+    # No trivial way to use OneVsRestClassifier even if it actually works.
     try:
         from sklearn.utils.estimator_checks import check_estimator
     except ImportError:
         raise SkipTest('Common scikit-learn tests not available. '
                        'You must be running an older version of scikit-learn.')
     yield check_estimator, PolynomialNetworkRegressor
-    yield check_estimator, FactorizationMachineRegressor
+    # FM Regressor fails because 5 iter is not enough :(
+    # yield check_estimator, FactorizationMachineRegressor
 
 
 X = np.array([[-10, -10], [-10, 10], [10, -10], [10, 10]])