Wenjuan Han comments

Repositories
Issues
Comments

Results 8 comments of


                                            Wenjuan Han

Where is tool/eval_quac.py

Hi. Seems still missing.

Trained models

The trained model is also helpful to me. That will be great if it is released.

supported version of cuda ?

Same bug.

Model Weights

Same question.

Test on other datasets

Yes, the score is influenced if only one answer is fed to calculate the diversity score. The n-gram based score is calculated based on a group of similar sentences. If...

Which version of transformers do you use when calculating self-logic ?

I think the version should not make much difference. Would you describe this more specific?

输出解析

测试的时候用的是脚本，利用正则表达式进行匹配。

准确率相关

我们测了ChatGPT，参考 [Zero-Shot Information Extraction via Chatting with ChatGPT](https://arxiv.org/abs/2302.10205) 但是GPT3.5没有