[Adapt] [seminar]An overview of faithfulness in natural language generation(I)

[贾琪]

Hi Adapters,


With the birth of Chatgpt, the magic of large language models is blowing everybody's mind. Such models are gradually becoming an indispensable part of our lives with diverse applications, such as the New Bing. Meanwhile, more and more factual errors are found by users, raising the importance and urgency of research on the faithfulness/hallucinations of these models.


In this talk, I'll provide an overview of faithfulness in natural language generation, including the definition, the categorization, and the contributors. It also includes some representative and state-of-the-art evaluation metrics for this problem. 




Time: Wed 4:00 pm
Venue: SEIEE 3-404

Best Regards,
Angel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://cs.sjtu.edu.cn/pipermail/adapt/attachments/20230214/a2fddfc6/attachment.htm>