修改tensor.data的返回，返回的张量不包含梯度信息

[Qin-sx]

PR Category

User Experience

PR Types

Improvements

Description

tensor.data功能和pytorch对齐，返回不包含梯度信息的tensor
测试代码：

import numpy as np
import paddle
import torch

np_data = np.array([[1.0, 2.0, 3.0], [4.0, 5.0, 6.0]], dtype=np.float32)
paddle_tensor = paddle.to_tensor(np_data, stop_gradient=False)
paddle_tensor_data = paddle_tensor.data
print(paddle_tensor_data)

np_data = np.array([[1.0, 2.0, 3.0], [4.0, 5.0, 6.0]], dtype=np.float32)
torch_tensor = torch.tensor(np_data, requires_grad=True)
torch_tensor_data = torch_tensor.data
print(torch_tensor_data)

[paddle-bot]

你的PR提交成功，感谢你对开源项目的贡献!
请关注后续CI自动化测试结果，详情请参考Paddle-CI手册。
Your PR has been submitted. Thanks for your contribution!
Please wait for the result of CI firstly. See Paddle CI Manual for details.

[zhwesky2010]

Tensor.data剥离了反向，应该不需要AutogradMeta了吧

[Qin-sx]

嗯，收到，已修改。测试了一下，不需要重新设置stopgradient，新的tensor没有梯度信息。

[Qin-sx]

CI / Mac-CPU / Build and test (pull_request)这个检查没有找到在哪里可以重新启动

[swgu98]

CI / Mac-CPU / Build and test (pull_request)这个检查没有找到在哪里可以重新启动

非required的失败不影响合入

[zhwesky2010]

LGTM