Dataset issues

[zhangyizhuoduanweiwei]

Hello, the dataset in the URL you provided, https://github.com/liyucheng09/Selective_Context/releases/download/v0.1.0rc1/datasets_dumps.zip, seems incomplete. There is no data in the news folder, it is empty. How can I prepare this part of the dataset?

[zhangyizhuoduanweiwei]

I would also like to ask you a question. In the main.py file, task_name: {task_name} seems to be undefined. Should it be modified like this:
for task_type in task_types:
task_name = task_type
logging.info(f'arxiv_path: {arxiv_path}, news_path: {news_path}, conversation_path: {conversation_path}, save_to_path: {save_to_path}, num_articles: {num_articles}, model_name: {model_name}, task_name: {task_name}')
Or should it be deleted?

[zhangyizhuoduanweiwei]

I would like to ask, what is the execution logic of the code? Is the main file executed first? But it reports an error saying that there is no dataset in the news folder.

[zhangyizhuoduanweiwei]

The get_self_information in the context_manager.py file uses the gpt3 model, but I don’t have the openai_api_key. Can this part be replaced with another open source model?