huggingface / huggingface/datasets
Is the 10k files / folder limit a hard limit for a dataset repo?
- Dominant language
- Python
- Stars
- 22k
- Forks
- 3.4k
- Avg merge
- 5d 7h
- Merged PRs (30d)
- 17
Description
### Feature request
Can the hard limit of 10k files per folder be extended with acceptable loss in performance?
### Motivation
I'm uploading a lance dataset to huggingface hub and have a folder inside lance internals (`data/*.lance/_transactions`) that has > 20k atomic transaction records and my commits are being rejected.
```
Bad request for commit endpoint:
Your push was rejected because it contains too many files per directory. Each directory in your git repo can only contain up to 10000 files. Offending reference: refs/heads/main Offending directories: /data/expression.lance/_transactions/
```
### Your contribution
Open to suggestions for how to make a PR
Contributor guide
Assessment
This issue has not been assessed yet.