【Hackathon 6th No.9】Update stack op to support zero sized tensor -part

[NKNaN]

PR Category

User Experience

PR Types

Improvements

Description

Support zero sized tensor stack operation:

x1 = paddle.ones([0, 1, 2])
x2 = paddle.ones([0, 1, 2])
out = paddle.stack([x1, x2])
# out -> shape = [2, 0, 1, 2]

x1 = paddle.ones([0, 1, 2])
x2 = paddle.ones([0, 1, 2])
out = paddle.stack([x1, x2], axis=1)
# out -> shape = [0, 2, 1, 2]

The gradient for each zero sized tensor in the input list of stack is None

[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.

[luotao1]

LGTM

1. 这里覆盖率没过的原因是什么？本地能覆盖到么？
2. 可以在下一个PR中，stack的注释里填加下 support zero sized tensor 的说明，放上 Description里的示例代码，同中文。

[NKNaN]

1. 这里覆盖率没过的原因是什么？本地能覆盖到么？

在kernel里cout来看的话确实应该 stack_kernel.cc 和 stack_grad_kernel.cc 都跑到了

2. 可以在下一个PR中，stack的注释里填加下 support zero sized tensor 的说明，放上 Description里的示例代码，同中文。

好的

[paddle-ci-bot]

Sorry to inform you that f4cc809's CIs have passed for more than 7 days. To prevent PR conflicts, you need to re-run all CIs manually.

[luotao1]

请重新触发下 CI