MTN benchopt 1.5 API (#48)

example_config.yml

@@ -1,4 +1,4 @@
-objective-filter:
+objective:
   - L2 Logistic Regression[lmbd=0.1]
 dataset:
   - simulated[n_features=500,n_samples=200]

objective.py

@@ -11,8 +11,8 @@ def _compute_loss(X, y, lmbd, beta):
 
 
 class Objective(BaseObjective):
-    min_benchopt_version = "1.3"
     name = "L2 Logistic Regression"
+    min_benchopt_version = "1.5"
 
     parameters = {
         'lmbd': [1., 0.01]
@@ -27,10 +27,10 @@ def set_data(self, X, y, X_test=None, y_test=None):
         msg = "Logistic loss is implemented with y in [-1, 1]"
         assert set(self.y) == {-1, 1}, msg
 
-    def get_one_solution(self):
-        return np.zeros((self.X.shape[1]))
+    def get_one_result(self):
+        return {'beta': np.zeros((self.X.shape[1]))}
 
-    def compute(self, beta):
+    def evaluate_result(self, beta):
         train_loss = _compute_loss(self.X, self.y, self.lmbd, beta)
         test_loss = None
         if self.X_test is not None:

solvers/cd.py

@@ -131,4 +131,4 @@ def sparse_cd(X_data, X_indices, X_indptr, y, lmbd, L, n_iter,
         return w
 
     def get_result(self):
-        return self.w
+        return dict(beta=self.w)

solvers/chop.py

@@ -172,4 +172,4 @@ def run(self, n_iter):
             self.run_full_batch(n_iter)
 
     def get_result(self):
-        return self.beta.detach().cpu().numpy().flatten()
+        return dict(beta=self.beta.detach().cpu().numpy().flatten())

solvers/copt.py

@@ -107,4 +107,4 @@ def step(x):
         self.beta = result.x
 
     def get_result(self):
-        return self.beta.flatten()
+        return dict(beta=self.beta.flatten())

solvers/cuml.py

@@ -52,4 +52,4 @@ def run(self, n_iter):
         self.clf.fit(self.X, self.y)
 
     def get_result(self):
-        return self.clf.coef_.to_numpy().flatten()
+        return dict(beta=self.clf.coef_.to_numpy().flatten())

solvers/glmnet.py

@@ -82,4 +82,4 @@ def get_result(self):
         coefs = np.array(as_matrix(results["beta"], "matrix"))
         beta = coefs.flatten()
 
-        return beta
+        return dict(beta=beta)

solvers/lightning.py

@@ -26,4 +26,4 @@ def run(self, n_iter):
         self.clf.fit(self.X, self.y)
 
     def get_result(self):
-        return self.clf.coef_.flatten()
+        return dict(beta=self.clf.coef_.flatten())

solvers/python_gd.py

@@ -29,7 +29,7 @@ def run(self, n_iter):
         self.w = w
 
     def get_result(self):
-        return self.w
+        return dict(beta=self.w)
 
     def compute_lipschitz_constant(self):
         if not sparse.issparse(self.X):

solvers/sklearn.py

@@ -56,4 +56,4 @@ def run(self, n_iter):
         self.clf.fit(self.X, self.y)
 
     def get_result(self):
-        return self.clf.coef_.flatten()
+        return dict(beta=self.clf.coef_.flatten())

solvers/snapml.py

@@ -41,4 +41,4 @@ def run(self, n_iter):
         self.clf.fit(self.X, self.y)
 
     def get_result(self):
-        return self.clf.coef_.flatten()
+        return dict(beta=self.clf.coef_.flatten())