From 866db7e1bb22abcc8be7ea23abaeaf60a251dbc5 Mon Sep 17 00:00:00 2001
From: Randy Shuai <rashuai@microsoft.com>
Date: Thu, 11 May 2023 11:49:37 -0700
Subject: [PATCH] clear fetch before emplace

---
 onnxruntime/core/framework/cloud_invoker.cc | 1 +
 1 file changed, 1 insertion(+)

diff --git a/onnxruntime/core/framework/cloud_invoker.cc b/onnxruntime/core/framework/cloud_invoker.cc
index 0f07ac29464e5..a2e9ec97cf931 100644
--- a/onnxruntime/core/framework/cloud_invoker.cc
+++ b/onnxruntime/core/framework/cloud_invoker.cc
@@ -187,6 +187,7 @@ onnxruntime::Status OpenAIInvoker::Send(const CloudEndPointConfig& run_options,
   auto* output_string = output_tensor->MutableData<std::string>();
   *output_string = string_buffer.ss_.str();
   auto tensor_type = DataTypeImpl::GetType<Tensor>();
+  ort_outputs.clear();
   ort_outputs.emplace_back(output_tensor.release(), tensor_type, tensor_type->GetDeleteFunc());
   return Status::OK();
 }