huggingface / huggingface/datasets
[2.6.1][2.7.0] Upgrade `datasets` to fix `TypeError: can only concatenate str (not "int") to str`
- Dominant language
- Python
- Stars
- 22k
- Forks
- 3.4k
- Avg merge
- 5d 7h
- Merged PRs (30d)
- 17
Description
`datasets` 2.6.1 and 2.7.0 started to stop supporting datasets like IMDB, ConLL or MNIST datasets.
When loading a dataset using 2.6.1 or 2.7.0, you may this error when loading certain datasets:
```python
TypeError: can only concatenate str (not "int") to str
```
This is because we started to update the metadata of those datasets to a format that is not supported in 2.6.1 and 2.7.0
This change is required or those datasets won't be supported by the Hugging Face Hub.
Therefore if you encounter this error or if you're using `datasets` 2.6.1 or 2.7.0, we encourage you to update to a newer version.
For example, versions 2.6.2 and 2.7.1 patch this issue.
```python
pip install -U datasets
```
All the datasets affected are the ones with a ClassLabel feature type and YAML "dataset_info" metadata. More info [here](https://github.com/huggingface/datasets/issues/5275).
We apologize for the inconvenience.
Contributor guide
Assessment
This issue has not been assessed yet.