Open githubzmt opened 11 months ago
Hi, there! I found that the data consist of "paragraph, sentence, answer, question and entity". Would you tell me if you use the sentence as the context? Also, in the val and test set, the number of question-answer pairs belonging to one paragraph is more than one, how do you process the data?
Thank you for your open-source code, but when running the code, I found that the experimental results did not achieve the corresponding effect. Is it related to the running process, and there is no dictionary parameter for saving the model in the code. Therefore, can you provide the corresponding code to run the process.