diff --git a/mindspore/lite/include/delegate.h b/mindspore/lite/include/delegate.h
index c19f94a4d2f..b841e722d3b 100644
--- a/mindspore/lite/include/delegate.h
+++ b/mindspore/lite/include/delegate.h
@@ -29,9 +29,10 @@ using KernelIter = std::vector<kernel::Kernel *>::iterator;
 class DelegateModel {
  public:
   /// \brief Constructor of MindSpore Lite DelegateModel.
-  DelegateModel(std::vector<kernel::Kernel *> *kernels,
-                const std::map<kernel::Kernel *, const schema::Primitive *> primitives)
-      : kernels_(kernels), primitives_(primitives) {}
+  DelegateModel(std::vector<kernel::Kernel *> *kernels, const std::vector<tensor::MSTensor *> &inputs,
+                const std::vector<tensor::MSTensor *> &outputs,
+                const std::map<kernel::Kernel *, const schema::Primitive *> &primitives)
+      : kernels_(kernels), inputs_(inputs), outputs_(outputs), primitives_(primitives) {}
 
   /// \brief Destructor of MindSpore Lite DelegateModel.
   ~DelegateModel() = default;
@@ -61,9 +62,15 @@ class DelegateModel {
   /// \return The next iterator after graph_kernel, point to the next kernel that is not visited.
   KernelIter Replace(KernelIter from, KernelIter end, kernel::Kernel *graph_kernel);
 
+  const std::vector<mindspore::tensor::MSTensor *> &inputs() { return this->inputs_; }
+
+  const std::vector<mindspore::tensor::MSTensor *> &outputs() { return this->outputs_; }
+
  protected:
   std::vector<kernel::Kernel *> *kernels_;
-  const std::map<kernel::Kernel *, const schema::Primitive *> primitives_;
+  const std::vector<mindspore::tensor::MSTensor *> &inputs_;
+  const std::vector<mindspore::tensor::MSTensor *> &outputs_;
+  const std::map<kernel::Kernel *, const schema::Primitive *> &primitives_;
 };
 
 typedef void (*DelegateHook)(std::shared_ptr<Delegate> delegate);
diff --git a/mindspore/lite/src/lite_session.cc b/mindspore/lite/src/lite_session.cc
index 4911a819b12..5c713fb929b 100644
--- a/mindspore/lite/src/lite_session.cc
+++ b/mindspore/lite/src/lite_session.cc
@@ -499,7 +499,7 @@ int LiteSession::CompileGraph(Model *model) {
     return ret;
   }
   // scheduler kernels
-  Scheduler scheduler(context_, model, &tensors_, is_train_session_, delegate_);
+  Scheduler scheduler(context_, model, &tensors_, inputs_, outputs_, is_train_session_, delegate_);
   scheduler.SetupSchedulerCb(std::move(sched_cb_));
   ret = scheduler.Schedule(&kernels_);
   if (ret != RET_OK) {
diff --git a/mindspore/lite/src/scheduler.cc b/mindspore/lite/src/scheduler.cc
index 84e30a09eb2..556f7e43246 100644
--- a/mindspore/lite/src/scheduler.cc
+++ b/mindspore/lite/src/scheduler.cc
@@ -161,7 +161,17 @@ int Scheduler::ReplaceDelegateKernels(std::vector<kernel::LiteKernel *> *dst_ker
   for (size_t i = 0; i < dst_kernels->size(); i++) {
     kernels.push_back((*dst_kernels)[i]->kernel());
   }
-  DelegateModel *model = new (std::nothrow) DelegateModel(&kernels, primitives_);
+
+  std::vector<tensor::MSTensor *> input_ms_tensors;
+  input_ms_tensors.resize(inputs_.size());
+  (void)std::transform(inputs_.begin(), inputs_.end(), input_ms_tensors.begin(),
+                       [](lite::Tensor *tensor) { return reinterpret_cast<tensor::MSTensor *>(tensor); });
+  std::vector<tensor::MSTensor *> output_ms_tensors;
+  output_ms_tensors.resize(outputs_.size());
+  (void)std::transform(outputs_.begin(), outputs_.end(), output_ms_tensors.begin(),
+                       [](lite::Tensor *tensor) { return reinterpret_cast<tensor::MSTensor *>(tensor); });
+
+  DelegateModel *model = new (std::nothrow) DelegateModel(&kernels, input_ms_tensors, output_ms_tensors, primitives_);
   if (model == nullptr) {
     MS_LOG(ERROR) << "New delegate model failed.";
     return RET_NULL_PTR;
diff --git a/mindspore/lite/src/scheduler.h b/mindspore/lite/src/scheduler.h
index e4f0bc606b1..a3c91ffaaad 100644
--- a/mindspore/lite/src/scheduler.h
+++ b/mindspore/lite/src/scheduler.h
@@ -34,11 +34,14 @@
 namespace mindspore::lite {
 class Scheduler {
  public:
-  Scheduler(const InnerContext *ctx, Model *src_model, std::vector<Tensor *> *src_tensors, bool is_train_session,
-            std::shared_ptr<Delegate> delegate = nullptr)
+  Scheduler(const InnerContext *ctx, Model *src_model, std::vector<Tensor *> *src_tensors,
+            const std::vector<Tensor *> &input_tensors, const std::vector<Tensor *> &output_tensors,
+            bool is_train_session, std::shared_ptr<Delegate> delegate = nullptr)
       : context_(ctx),
         src_model_(src_model),
         src_tensors_(src_tensors),
+        inputs_(input_tensors),
+        outputs_(output_tensors),
         is_train_session_(is_train_session),
         delegate_(delegate) {}
   ~Scheduler() = default;
@@ -111,6 +114,8 @@ class Scheduler {
   const InnerContext *context_ = nullptr;
   Model *src_model_ = nullptr;
   std::vector<Tensor *> *src_tensors_;
+  const std::vector<Tensor *> &inputs_;
+  const std::vector<Tensor *> &outputs_;
   std::vector<size_t> graph_output_node_indexes_;
   std::map<int, OpParameter *> op_parameters_;
   bool is_train_session_ = false;