This PR resolves issue #47 by clarifying the meaning of the sampling weight in the task name argument within the training README. It also provides detailed instructions on how users can add their own data to the DATASETS by creating a jsonl file and including it in the task name with an appropriate sampling weight. The changes aim to make the training process more accessible and understandable for users looking to customize their training datasets.
This PR resolves issue #47 by clarifying the meaning of the sampling weight in the task name argument within the training README. It also provides detailed instructions on how users can add their own data to the DATASETS by creating a
jsonlfile and including it in the task name with an appropriate sampling weight. The changes aim to make the training process more accessible and understandable for users looking to customize their training datasets.