From d2f97fdc879604baaa4530b93b36029127d07ae3 Mon Sep 17 00:00:00 2001
From: zjun <zhangjun0@huawei.com>
Date: Tue, 6 Jul 2021 16:20:32 +0800
Subject: [PATCH] Ignore funcgraph grad

Signed-off-by: zjun <zhangjun0@huawei.com>
---
 .../pipeline/pynative/pynative_execute.cc      | 15 +++++++++++++--
 .../composite/multitype_ops/ones_like_impl.py  | 18 ++++++++++++++++++
 2 files changed, 31 insertions(+), 2 deletions(-)

diff --git a/mindspore/ccsrc/pipeline/pynative/pynative_execute.cc b/mindspore/ccsrc/pipeline/pynative/pynative_execute.cc
index 0b9902c6e58..eb42fc16546 100644
--- a/mindspore/ccsrc/pipeline/pynative/pynative_execute.cc
+++ b/mindspore/ccsrc/pipeline/pynative/pynative_execute.cc
@@ -2102,9 +2102,20 @@ void GradExecutor::CreateMakeTupleNodeForMultiOut(const std::string &cell_id, co
   // get input node and value
   std::vector<AnfNodePtr> inputs{NewValueNode(prim::kPrimMakeTuple)};
   ValuePtrList input_args;
+  std::vector<int> value_index;
   for (size_t i = 0; i < out_tuple.size(); i++) {
+    auto v = parse::data_converter::PyDataToValue(out_tuple[i]);
+    // Graph have no define for grad
+    if (v->isa<FuncGraph>()) {
+      continue;
+    }
+    value_index.emplace_back(i);
+    input_args.emplace_back(v);
     inputs.emplace_back(GetInput(out_tuple[i], false));
-    input_args.emplace_back(parse::data_converter::PyDataToValue(out_tuple[i]));
+  }
+  py::tuple value_outs(value_index.size());
+  for (size_t i = 0; i < value_index.size(); ++i) {
+    value_outs[i] = out_tuple[value_index[i]];
   }
   auto cnode = curr_g_->NewCNode(inputs);
   MS_LOG(DEBUG) << "Tuple output node info " << cnode->DebugString();
@@ -2116,7 +2127,7 @@ void GradExecutor::CreateMakeTupleNodeForMultiOut(const std::string &cell_id, co
     return;
   }
   // run ad for maketuple node
-  ValuePtr out_value = parse::data_converter::PyDataToValue(out);
+  ValuePtr out_value = parse::data_converter::PyDataToValue(value_outs);
   ad::GradPynativeOp(top_cell()->k_pynative_cell_ptr(), cnode, input_args, out_value);
 }
 
diff --git a/mindspore/ops/composite/multitype_ops/ones_like_impl.py b/mindspore/ops/composite/multitype_ops/ones_like_impl.py
index 1325cbce0a8..54b2a31b51b 100644
--- a/mindspore/ops/composite/multitype_ops/ones_like_impl.py
+++ b/mindspore/ops/composite/multitype_ops/ones_like_impl.py
@@ -58,6 +58,24 @@ def _ones_like_sparse_tensor(x):
     return F.make_sparse_tensor(F.sparse_tensor_get_indices(x), values, F.sparse_tensor_get_dense_shape(x))
 
 
+newenv = base.EnvInstance_()
+
+
+@ones_like_leaf.register("Function")
+def _ones_like_func(x):
+    """
+    Derivation of a function.
+
+    Args:
+        x (Function): x
+
+    Returns:
+        EnvInstance_, value is newenv.
+    """
+    # Unused parameters are placeholders.
+    return newenv
+
+
 ones_like = base.HyperMap(ones_like_leaf)
 """
 `ones_like` is a function which can generate a graph of `ones_like` operation according to input tensor dtype.