Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

【Hackathon 4th No.104】generate API对齐HF #5587

Open
wants to merge 13 commits into
base: develop
Choose a base branch
from

Conversation

zwhong714
Copy link

PR types

Application

PR changes

Description

提交RFC文档,sampling改写,新增contrastive search

@paddle-bot
Copy link

paddle-bot bot commented Apr 10, 2023

Thanks for your contribution!

@CLAassistant
Copy link

CLAassistant commented Apr 10, 2023

CLA assistant check
All committers have signed the CLA.

@gongel
Copy link
Member

gongel commented Apr 10, 2023

非常赞,感谢贡献,我们马上review哈。也辛苦签署下CLA

@zwhong714
Copy link
Author

OK

@sijunhe sijunhe changed the title 104 104号任务:generate API对齐HF Apr 11, 2023
paddlenlp/transformers/generation_utils.py Outdated Show resolved Hide resolved
paddlenlp/transformers/generation_utils.py Outdated Show resolved Hide resolved
paddlenlp/transformers/generation_utils.py Outdated Show resolved Hide resolved
Copy link
Member

@gongel gongel left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

改的很好!另外是否考虑做一些默认超参的对齐,比如我们这里的max_length和HF的max_new_tokens是一致的,HF的max_length会考虑输入的长度。

paddlenlp/generation/logits_process.py Show resolved Hide resolved
paddlenlp/transformers/generation_utils.py Outdated Show resolved Hide resolved
@gongel gongel changed the title 104号任务:generate API对齐HF 【Hackathon 4th No.104】generate API对齐HF Apr 12, 2023
104RFC.md Outdated Show resolved Hide resolved
paddlepaddle-2.4.1-cp310-cp310-macosx_10_14_universal2.whl Outdated Show resolved Hide resolved
104RFC.md Outdated Show resolved Hide resolved
paddlenlp/generation/logits_process.py Show resolved Hide resolved
paddlenlp/generation/logits_process.py Outdated Show resolved Hide resolved
paddlenlp/generation/stopping_criteria.py Outdated Show resolved Hide resolved
paddlenlp/transformers/generation_utils.py Outdated Show resolved Hide resolved
paddlenlp/transformers/generation_utils.py Outdated Show resolved Hide resolved
tests/generation/test_logits_process.py Outdated Show resolved Hide resolved
tests/generation/test_logits_process.py Show resolved Hide resolved
@zwhong714
Copy link
Author

hello,这边交了一版,不知道还有什么问题没有

@gongel
Copy link
Member

gongel commented Apr 23, 2023

conflicts解决下哈
另外大批量CI挂了

@zwhong714
Copy link
Author

目前无法处理的模型如下
无decoder hidden states
pytest PaddleNLP/tests/transformers/pegasus/test_modeling.py

multihead类型(OPT),与paddle.nn 中multiheadAttention 不一致
pytest PaddleNLP/tests/transformers/opt/test_modeling.py

multihead类型(GPT),与paddle.nn 中multiheadAttention 不一致
pytest PaddleNLP/tests/transformers/gpt/test_modeling.py

@gongel
Copy link
Member

gongel commented May 16, 2023

好的辛苦,目前改动文件太多,检查下是否代码风格的CI没过。

@zwhong714
Copy link
Author

好的辛苦,目前改动文件太多,检查下是否代码风格的CI没过。

好像只有test_generation_utils, 和test_generation_utils 改了,其它文件不知道是不是拉取develop版本的时候出现的问题

@zwhong714
Copy link
Author

这边我看日志,test和cli挂了好像就是因为那三个模型的问题

@sijunhe
Copy link
Collaborator

sijunhe commented May 24, 2023

@zwhong714 您好,感谢您参与飞桨黑客松以及为这个PR付出的努力。这个PR我们看下来问题不大,基本可以合入, 但是因为我们目前的一些内部原因暂时不能合入。我们后续会找合适的机会合入这个PR。黑客松活动这边我们算您完成这个任务了~

@github-actions
Copy link

This Pull Request is stale because it has been open for 60 days with no activity. 当前Pull Request 60天内无活动,被标记为stale。

@github-actions
Copy link

This Pull Request is stale because it has been open for 60 days with no activity. 当前Pull Request 60天内无活动,被标记为stale。

@github-actions github-actions bot added the stale label Oct 29, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

6 participants