From 059cb03d8854b484bb743bce10c64e5427620d55 Mon Sep 17 00:00:00 2001
From: chentianyu03 <chentianyu03@baidu.com>
Date: Mon, 1 Nov 2021 09:00:30 +0000
Subject: [PATCH] fix multioutputs bug

---
 paddle/pten/core/kernel_context.h | 2 +-
 paddle/pten/core/kernel_utils.h   | 4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/paddle/pten/core/kernel_context.h b/paddle/pten/core/kernel_context.h
index e7376cb510e86..5e9cccaf2b1dc 100644
--- a/paddle/pten/core/kernel_context.h
+++ b/paddle/pten/core/kernel_context.h
@@ -120,7 +120,7 @@ class KernelContext {
 
   template <typename TensorType>
   std::vector<TensorType*> MutableOutputAt(size_t start, size_t end) {
-    std::vector<TensorType> v;
+    std::vector<TensorType*> v;
     for (size_t i = start; i < end; ++i) {
       v.emplace_back(static_cast<TensorType*>(outputs_.at(i).get()));
     }
diff --git a/paddle/pten/core/kernel_utils.h b/paddle/pten/core/kernel_utils.h
index 6872d9632a3fd..d992821c2c0bf 100644
--- a/paddle/pten/core/kernel_utils.h
+++ b/paddle/pten/core/kernel_utils.h
@@ -146,7 +146,7 @@ using XPUContext = paddle::platform::XPUDeviceContext;
 
 #define PT_SPECIALIZE_KernelCallHelper_FOR_MULTI_OUTPUT(tensor_type)     \
   template <typename... Tail>                                            \
-  struct KernelCallHelper<const std::vector<tensor_type*>&, Tail...> {   \
+  struct KernelCallHelper<std::vector<tensor_type*>, Tail...> {          \
     template <int dev_ctx_idx,                                           \
               int in_idx,                                                \
               int attr_idx,                                              \
@@ -154,7 +154,7 @@ using XPUContext = paddle::platform::XPUDeviceContext;
               typename... PreviousArgs>                                  \
     static void Compute(KernelContext* ctx, PreviousArgs&... pargs) {    \
       const std::pair<int, int> range = ctx->OutputRangeAt(out_idx);     \
-      const std::vector<tensor_type*> arg = std::move(                   \
+      std::vector<tensor_type*> arg = std::move(                         \
           ctx->MutableOutputAt<tensor_type>(range.first, range.second)); \
       KernelCallHelper<Tail...>::                                        \
           template Compute<dev_ctx_idx, in_idx, attr_idx, out_idx + 1>(  \