Implement binary io for faster/lighter serialization
Open
enhancement
good first issue
- Dominant language
- C++
- Stars
- 3.4k
- Forks
- 701
- PR merge metrics
- No merged PRs in 30d
Description
Saving a medium sized seq2seq model takes between 2x and 4x longer as before, is anyone else experiencing something similar?
EDIT: some more information:
- the model weighs 340MB once saved (which is less than before 2.0)
- There are 3 subcollections
Contributor guide
No contributing guide indexed for this repository
Assessment
This issue has not been assessed yet.