From: Yinghai Lu <yinghai@fb.com>
Date: Wed, 9 Jan 2019 06:24:37 +0000 (-0800)
Subject: Initialize tensor with fp32 in Caffe2Backend.prepare() (#15832)
X-Git-Tag: accepted/tizen/6.5/unified/20211028.231830~1953
X-Git-Url: http://review.tizen.org/git/?a=commitdiff_plain;h=5fe26976559490a99cb60a0dd2e457c1fdda65ba;p=platform%2Fupstream%2Fpytorch.git

Initialize tensor with fp32 in Caffe2Backend.prepare() (#15832)

Summary:
Fix https://github.com/pytorch/pytorch/issues/14104
Pull Request resolved: https://github.com/pytorch/pytorch/pull/15832

Reviewed By: bddppq

Differential Revision: D13598332

Pulled By: yinghai

fbshipit-source-id: 3302ac47928974f49353c5da8af440e5c1716c22
---

diff --git a/caffe2/python/onnx/backend.py b/caffe2/python/onnx/backend.py
index d4cc70a..119750c 100644
--- a/caffe2/python/onnx/backend.py
+++ b/caffe2/python/onnx/backend.py
@@ -644,7 +644,10 @@ class Caffe2Backend(Backend):
             if value_info.name in initialized:
                 continue
             shape = list(d.dim_value for d in value_info.type.tensor_type.shape.dim)
-            ws.FeedBlob(value_info.name, np.ones(shape), device_option)
+            ws.FeedBlob(
+                value_info.name,
+                np.ones(shape, dtype=onnx.mapping.TENSOR_TYPE_TO_NP_TYPE[value_info.type.tensor_type.elem_type]),
+                device_option)
 
     @staticmethod
     def optimize_onnx(input, init=False, predict=False):