diff --git a/paddle/fluid/ir_adaptor/translator/op_compat_gen.py b/paddle/fluid/ir_adaptor/translator/op_compat_gen.py
index ea844c659554d..1035674cdb3d9 100644
--- a/paddle/fluid/ir_adaptor/translator/op_compat_gen.py
+++ b/paddle/fluid/ir_adaptor/translator/op_compat_gen.py
@@ -164,6 +164,9 @@ def insert_new_mutable_attributes(
         "atol_tensor": "TolTensor",
         "out": "Out",
     }
+    op_arg_name_mappings['push_sparse_v2'].update(
+        {"out_grad_in": "Out@GRAD", "out_grad_out": "Out@GRAD"}
+    )
 
     op_name_normailzer_template = env.get_template("op_compat_info.cc.j2")
     with open(output_source_file, 'wt') as f:
diff --git a/paddle/fluid/pir/dialect/op_generator/ops_api_gen.py b/paddle/fluid/pir/dialect/op_generator/ops_api_gen.py
index 04da5c4be230c..f38261bd775cc 100644
--- a/paddle/fluid/pir/dialect/op_generator/ops_api_gen.py
+++ b/paddle/fluid/pir/dialect/op_generator/ops_api_gen.py
@@ -145,6 +145,8 @@
     'uniform_random_batch_size_like',
     'c_reduce_min',
     'c_reduce_min_',
+    'push_sparse_v2',
+    'push_sparse_v2_',
 ]
 
 
diff --git a/paddle/fluid/pir/dialect/operator/ir/ops.yaml b/paddle/fluid/pir/dialect/operator/ir/ops.yaml
index 947e6306c3032..67f65298e0407 100644
--- a/paddle/fluid/pir/dialect/operator/ir/ops.yaml
+++ b/paddle/fluid/pir/dialect/operator/ir/ops.yaml
@@ -979,6 +979,17 @@
     func : prod
   backward : prod_grad
 
+- op : push_sparse_v2
+  args : (Tensor[] ids, Tensor[] w, Tensor[] out_grad_in, int embeddingdim = 11, int tableid = 0, str accessorclass = "", str ctrlabelname = "", int paddingid = 0, bool scalesparsegrad = true, str[] inputnames = {}, bool is_distributed = true)
+  output : Tensor[](out_grad_out){out_grad_in.size()}
+  infer_meta :
+    func : UnchangedMultiInferMeta
+    param : [out_grad_in]
+  kernel :
+    func : push_sparse_v2
+    data_type : out_grad_in
+  inplace: (out_grad_in -> out_grad_out)
+
 - op : randint
   args : (int low, int high, IntArray shape, DataType dtype=DataType::INT64, Place place={})
   output : Tensor(out)
diff --git a/paddle/fluid/pir/dialect/operator/utils/utils.cc b/paddle/fluid/pir/dialect/operator/utils/utils.cc
index eff7fecb844fb..18daeb8565580 100644
--- a/paddle/fluid/pir/dialect/operator/utils/utils.cc
+++ b/paddle/fluid/pir/dialect/operator/utils/utils.cc
@@ -72,7 +72,8 @@ const std::unordered_set<std::string> LegacyOpList = {
     paddle::onednn::dialect::LrnOp::name(),
     paddle::onednn::dialect::LrnGradOp::name(),
 #endif
-    CReduceMinOp::name()};
+    CReduceMinOp::name(),
+    PushSparseV2Op::name()};
 
 enum class AttrType {
   UNDEFINED = 0,
diff --git a/paddle/phi/api/yaml/op_compat.yaml b/paddle/phi/api/yaml/op_compat.yaml
index 08172e9190b92..3c0cb80949235 100755
--- a/paddle/phi/api/yaml/op_compat.yaml
+++ b/paddle/phi/api/yaml/op_compat.yaml
@@ -2461,6 +2461,14 @@
   outputs :
     out : Out
 
+- op : push_sparse_v2
+  inputs :
+    { x : Ids, W : w}
+  outputs :
+    out : Out
+  extra :
+    attrs : [int embeddingdim = 11, int tableid = 0, str accessorclass = "", str ctrlabelname = "", int paddingid = 0, bool scalesparsegrad = true, 'str[] inputnames = {}', bool is_distributed = true]
+
 - op : put_along_axis
   backward : put_along_axis_grad
   inputs :