"bugs:fix lstm rnn.pack_padded_sequence RuntimeError, 详细信息: https://g…

[johnson7788]

…ithub.com/pytorch/pytorch/issues/43227"

Description：简要描述这次PR的内容 一个torch的bug: nn.utils.rnn.pack_padded_sequence: RuntimeError: 'lengths' argument should be a 1D CPU int64 tensor, but got 1D cuda:0 Long tensor 详情查看: https://github.com/pytorch/pytorch/issues/43227

Main reason: 做出这次修改的原因 兼容torch 1.70 GPU版本

Checklist 检查下面各项是否完成

Please feel free to remove inapplicable items for your PR.

• [bugfix] The PR title starts with [$CATEGORY] (例如[bugfix]修复bug，[new]添加新功能，[test]修改测试，[rm]删除旧代码)
• [ ] Changes are complete (i.e. I finished coding on this PR) 修改完成才提PR
• [ ] All changes have test coverage 修改的部分顺利通过测试。对于fastnlp/fastnlp/的修改，测试代码必须提供在fastnlp/test/。
• [ ] Code is well-documented 注释写好，API文档会从注释中抽取
• [ ] To the my best knowledge, examples are either not affected by this change, or have been fixed to be compatible with this change 修改导致例子或tutorial有变化，请找核心开发人员

Changes: 逐项描述修改的内容

• torch pack_padded_sequence函数兼容torch1.7.0 cuda 版本

Mention: 找人review你的PR

@修改过这个文件的人 @核心开发人员

[yhcc]

I didn't see you are trying to push to the master branch. Right now, we are developing in the dev branch, can you post a new pr to the dev branch instead of the master branch?

[johnson7788]

已提交到dev分支; Already PR to dev Branch