Add gpt synthetic dataset generator and update poetry

[qidanrui]

Firstly, install dbgpt_hub with pip install dbgpt-hub

Secondly, put Spider dataset under the dbgpt_hub/data folder

Then, run the following code:

from dbgpt_hub.data_generator import generate_dataset_with_gpt

args = {
            "model": "gpt-3.5-turbo-16k", 
            "prompt": COT_PROMPT,
            "num_text2sql_pair_each_db": 1,
            "table_file_path": os.path.join(ROOT_PATH, "dbgpt_hub/data/spider/tables.json"),
            "output_path": os.path.join(ROOT_PATH, "dbgpt_hub/data/spider/synthetic_data_with_gpt.json")
}

generate_dataset_with_gpt(args)

You can also run the generation code directly without any arguments:

from dbgpt_hub.data_generator import generate_dataset_with_gpt
generate_dataset_with_gpt()

Finally, we can generate synthetic dataset in the following figure:

[wangzaistone]

LGTM