Skip to content

FG for Text-To-Text Transfer Transformer #3763

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

Merged
merged 19 commits into from
Dec 21, 2022
Merged

Conversation

FrostML
Copy link
Contributor

@FrostML FrostML commented Nov 15, 2022

PR types

New features

PR changes

Models

Description

FG for Text-To-Text Transfer Transformer.

@FrostML
Copy link
Contributor Author

FrostML commented Dec 8, 2022

因格式导致冲突,将待 review 完成后处理。

Copy link
Contributor

@guoshengCS guoshengCS left a comment

Choose a reason for hiding this comment

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

是否补充下faster_generation下的例子和文档,给出性能数据,其他没有什么问题

@codecov
Copy link

codecov bot commented Dec 16, 2022

Codecov Report

Merging #3763 (413bc3a) into develop (fcfe692) will decrease coverage by 1.17%.
The diff coverage is 9.61%.

❗ Current head 413bc3a differs from pull request most recent head 341e4c0. Consider uploading reports for the commit 341e4c0 to get more accurate results

@@             Coverage Diff             @@
##           develop    #3763      +/-   ##
===========================================
- Coverage    34.06%   32.89%   -1.18%     
===========================================
  Files          405      400       -5     
  Lines        56613    56642      +29     
===========================================
- Hits         19288    18634     -654     
- Misses       37325    38008     +683     
Impacted Files Coverage Δ
paddlenlp/transformers/generation_utils.py 71.98% <ø> (-1.88%) ⬇️
...nlp/ops/faster_transformer/transformer/decoding.py 7.50% <7.65%> (-0.72%) ⬇️
paddlenlp/transformers/t5/modeling.py 16.74% <13.51%> (-70.22%) ⬇️
...ster_transformer/transformer/faster_transformer.py 12.42% <15.00%> (+0.17%) ⬆️
paddlenlp/transformers/t5/configuration.py 42.85% <0.00%> (-57.15%) ⬇️
paddlenlp/transformers/mbart/modeling.py 80.73% <0.00%> (-1.21%) ⬇️
paddlenlp/transformers/unimo/modeling.py 82.08% <0.00%> (-0.80%) ⬇️
paddlenlp/transformers/configuration_utils.py 67.09% <0.00%> (-0.78%) ⬇️
... and 13 more

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

@FrostML FrostML merged commit 68c4b09 into PaddlePaddle:develop Dec 21, 2022

layer_outputs = recompute(
create_custom_forward(layer_module),
layer_outputs = self.recompute_training(
Copy link
Collaborator

Choose a reason for hiding this comment

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

动转静为什么会走到 recompute 的分支?

@gongel gongel mentioned this pull request Mar 28, 2023
1 task
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants