Add needed param to warning for L2 in SDCA

src/Microsoft.ML.StandardLearners/Standard/LinearClassificationTrainer.cs

@@ -207,10 +207,9 @@ internal virtual void Check(IHostEnvironment env)
                 {
                     using (var ch = env.Start("SDCA arguments checking"))
                     {
-                        ch.Warning("The specified l2Const = {0} is too small. SDCA optimizes the dual objective function. " +
-                            "The dual formulation is only valid with a positive L2 regularization. Also, an l2Const less than {1} " +
-                            "could drastically slow down the convergence. So using l2Const = {1} instead.", L2Const);
-
+                        ch.Warning($"The L2 regularization constant must be at least {L2LowerBound}. In SDCA, the dual formulation " +
+                            $"is only valid with a positive constant, and values below {L2LowerBound} cause very slow convergence. " +
+                            $"The original {nameof(L2Const)} = {L2Const}, was replaced with {nameof(L2Const)} = {L2LowerBound}.");
                         L2Const = L2LowerBound;
                         ch.Done();
                     }
@@ -1752,4 +1751,4 @@ public static CommonOutputs.BinaryClassificationOutput TrainBinary(IHostEnvironm
 
         }
     }
-}
+}