Description

Some hf dataset, such as openai/gsm8k has multi config (e.g., "main" and "socratic"), directly load dataset will lead to error. Instead, we should use the user specified config_name if provided, and if not, we simply try to load_dataset with dataset name.

This PR also update examples/grpo_gsm8k/gsm8k.yaml's buffer.train_dataset.path from sqlite:////gsm8k.db to sqlite:///gsm8k.db, since sqlite:////gsm8k.db will create the db file under the root directory (some may have no permission under a server's root directory), instead sqlite:///gsm8k.db will create the db file under the current direcotry.

Checklist

Please check the following items before code is ready to be reviewed.

• Code has passed all tests
• Docstrings have been added/updated in Google Style
• Documentation has been updated
• Code is ready for review