diff --git a/paddle/fluid/pybind/op_function_generator.cc b/paddle/fluid/pybind/op_function_generator.cc
index f9d11e8154f43fe7b32dedd515ec128c748f6f5a..32e14dafb644bfbd0b905f6a02b50f5c7f3b001f 100644
--- a/paddle/fluid/pybind/op_function_generator.cc
+++ b/paddle/fluid/pybind/op_function_generator.cc
@@ -71,6 +71,9 @@ std::map<std::string, std::set<std::string>> op_ins_map = {
     {"adam",
      {"Param", "Grad", "LearningRate", "Moment1", "Moment2", "Beta1Pow",
       "Beta2Pow", "MasterParam"}},
+    {"adamw",
+     {"Param", "Grad", "LearningRate", "Moment1", "Moment2", "Beta1Pow",
+      "Beta2Pow", "MasterParam"}},
 };
 
 // NOTE(zhiqiu): Like op_ins_map.
@@ -110,6 +113,9 @@ std::map<std::string, std::set<std::string>> op_outs_map = {
     {"adam",
      {"ParamOut", "Moment1Out", "Moment2Out", "Beta1PowOut", "Beta2PowOut",
       "MasterParamOut"}},
+    {"adamw",
+     {"ParamOut", "Moment1Out", "Moment2Out", "Beta1PowOut", "Beta2PowOut",
+      "MasterParamOut"}},
 };
 
 // NOTE(zhiqiu): Commonly, the outputs in auto-generated OP function are
@@ -129,7 +135,8 @@ std::map<std::string, std::set<std::string>> op_passing_outs_map = {
      {"ParamOut", "Moment1Out", "Moment2Out", "Beta1PowOut", "Beta2PowOut",
       "MasterParamOut"}},
     {"adamw",
-     {"ParamOut", "Moment1Out", "Moment2Out", "Beta1PowOut", "Beta2PowOut"}},
+     {"ParamOut", "Moment1Out", "Moment2Out", "Beta1PowOut", "Beta2PowOut",
+      "MasterParamOut"}},
     {"average_accumulates",
      {"out_sum_1", "out_sum_2", "out_sum_3", "out_num_accumulates",
       "out_old_num_accumulates", "out_num_updates"}},
diff --git a/python/paddle/optimizer/adamw.py b/python/paddle/optimizer/adamw.py
index 10d6af651777e2ec91af0bd03fddd4fd4f0232c7..34fb201d8ccaf7bb801de53f30b3f2718838d83c 100644
--- a/python/paddle/optimizer/adamw.py
+++ b/python/paddle/optimizer/adamw.py
@@ -298,14 +298,14 @@ class AdamW(Adam):
             _beta2 = self._beta2 if not isinstance(
                 self._beta2, Variable) else self._beta2.numpy().item(0)
 
-            _, _, _, _, _, _ = _C_ops.adam(
+            _, _, _, _, _, _ = _C_ops.adamw(
                 param_and_grad[0], param_and_grad[1], lr, moment1, moment2,
                 beta1_pow_acc, beta2_pow_acc, master_weight, param_and_grad[0],
                 moment1, moment2, beta1_pow_acc, beta2_pow_acc, master_weight,
                 'epsilon', self._epsilon, 'lazy_mode', self._lazy_mode,
                 'min_row_size_to_use_multithread', 1000, 'beta1', _beta1,
                 'beta2', _beta2, 'coeff', self._coeff, 'multi_precision',
-                find_master)
+                find_master, "lr_ratio", lr_ratio_)
 
             return None