diff --git a/skglm/datafits/single_task.py b/skglm/datafits/single_task.py
index 1dfd79e85..18be14b87 100644
--- a/skglm/datafits/single_task.py
+++ b/skglm/datafits/single_task.py
@@ -498,8 +498,8 @@ def gradient_scalar_sparse(self, X_data, X_indptr, X_indices, y, Xw, j):
             grad += X_data[i] * (np.exp(Xw[idx_i]) - y[idx_i])
         return grad / len(y)
 
-    def intercept_update_self(self, y, Xw):
-        pass
+    def intercept_update_step(self, y, Xw):
+        return np.sum(self.raw_grad(y, Xw))
 
 
 class Gamma(BaseDatafit):
@@ -555,8 +555,8 @@ def gradient_scalar(self, X, y, w, Xw, j):
     def gradient_scalar_sparse(self, X_data, X_indptr, X_indices, y, Xw, j):
         pass
 
-    def intercept_update_self(self, y, Xw):
-        pass
+    def intercept_update_step(self, y, Xw):
+        return np.sum(self.raw_grad(y, Xw))
 
 
 class Cox(BaseDatafit):