diff --git a/paddle/phi/api/yaml/legacy_api.yaml b/paddle/phi/api/yaml/legacy_api.yaml
index 3e2e0f70a9260ff6430d65ea1d43daa1dc2a11e2..69f37c374cffd5bdf4672eea8e6220318eb6492e 100644
--- a/paddle/phi/api/yaml/legacy_api.yaml
+++ b/paddle/phi/api/yaml/legacy_api.yaml
@@ -280,6 +280,16 @@
     func : bce_loss
   backward : bce_loss_grad
 
+- api : bilinear_tensor_product
+  args : (Tensor x, Tensor y, Tensor weight, Tensor bias)
+  output : Tensor
+  infer_meta :
+    func : BilinearTensorProductInferMeta
+  kernel :
+    func : bilinear_tensor_product
+  optional : bias
+  backward : bilinear_tensor_product_grad
+
 # bitwise_and
 - api : bitwise_and
   args : (Tensor x, Tensor y)
diff --git a/paddle/phi/api/yaml/legacy_backward.yaml b/paddle/phi/api/yaml/legacy_backward.yaml
index 73749b94870a634ef76ae8aa20ccdbd3f1cfaf21..aa83bb54a03152d39385a20e7f0e7cd5556936fc 100644
--- a/paddle/phi/api/yaml/legacy_backward.yaml
+++ b/paddle/phi/api/yaml/legacy_backward.yaml
@@ -251,6 +251,15 @@
     func : bce_loss_grad
   inplace : (out_grad -> input_grad)
 
+- backward_api : bilinear_tensor_product_grad
+  forward : bilinear_tensor_product (Tensor x, Tensor y, Tensor weight, Tensor bias) -> Tensor(out)
+  args : (Tensor x, Tensor y, Tensor weight, Tensor out_grad)
+  output : Tensor(x_grad), Tensor(y_grad), Tensor(weight_grad), Tensor(bias_grad)
+  infer_meta :
+    func : BilinearTensorProductGradInferMeta
+  kernel :
+    func : bilinear_tensor_product_grad
+
 - backward_api : brelu_grad
   forward : brelu (Tensor x, float t_min, float t_max) -> Tensor(out)
   args : (Tensor x, Tensor out_grad, float t_min, float t_max)
diff --git a/python/paddle/fluid/tests/unittests/test_bilinear_tensor_product_op.py b/python/paddle/fluid/tests/unittests/test_bilinear_tensor_product_op.py
index 5301924927087050ea36b6e510550abc2f16a05e..6eb7e8547d27c3b0038984dccdddfc967c6ad4cc 100644
--- a/python/paddle/fluid/tests/unittests/test_bilinear_tensor_product_op.py
+++ b/python/paddle/fluid/tests/unittests/test_bilinear_tensor_product_op.py
@@ -18,6 +18,7 @@ import unittest
 import numpy as np
 import paddle.fluid as fluid
 from op_test import OpTest
+import paddle
 
 
 class TestDygraphBilinearTensorProductAPIError(unittest.TestCase):
@@ -41,6 +42,7 @@ class TestBilinearTensorProductOp(OpTest):
 
     def setUp(self):
         self.op_type = "bilinear_tensor_product"
+        self.python_api = paddle.nn.functional.bilinear
         batch_size = 6
         size0 = 5
         size1 = 4
@@ -63,10 +65,10 @@ class TestBilinearTensorProductOp(OpTest):
         self.outputs = {'Out': output + bias}
 
     def test_check_output(self):
-        self.check_output()
+        self.check_output(check_eager=True)
 
     def test_check_grad_normal(self):
-        self.check_grad(['X', 'Y', 'Weight', 'Bias'], 'Out')
+        self.check_grad(['X', 'Y', 'Weight', 'Bias'], 'Out', check_eager=True)
 
 
 if __name__ == "__main__":
diff --git a/python/paddle/nn/functional/common.py b/python/paddle/nn/functional/common.py
index e10a1c106914144988dc8ee28a5557a89415f254..b5e34199aafbf604fdccf7a4528315e97821f344 100644
--- a/python/paddle/nn/functional/common.py
+++ b/python/paddle/nn/functional/common.py
@@ -850,7 +850,9 @@ def bilinear(x1, x2, weight, bias=None, name=None):
 
     """
 
-    if in_dynamic_mode():
+    if in_dygraph_mode():
+        return _C_ops.final_state_bilinear_tensor_product(x1, x2, weight, bias)
+    elif _non_static_mode():
         return _C_ops.bilinear_tensor_product(x1, x2, weight, bias)
 
     check_variable_and_dtype(x1, 'x1', ['float32', 'float64'], 'bilinear')