- Download the zhwiki dataset from <https://dumps.wikimedia.org/zhwiki> for pre-training. Extract and clean text in the dataset with [WikiExtractor](https://github.com/attardi/wil
kiextractor). Convert the dataset to TFRecord format and move the files to a specified path.
- Download the CLUE dataset from <https://www.cluebenchmarks.com> for fine-tuning and evaluation.
- Download the zhwiki dataset for pre-training. Extract and clean text in the dataset with [WikiExtractor](https://github.com/attardi/wikiextractor). Convert the dataset to TFRecord format and move the files to a specified path.
- Download the CLUE dataset for fine-tuning and evaluation.
> Notes:
If you are running a fine-tuning or evaluation task, prepare the corresponding checkpoint file.