Blog Posts
GoLLIE: Guideline-following Large Language Model for Information Extraction
Oscar Sainz, Iker García-Ferrero, Rodrigo Agerri, Oier Lopez de Lacalle, German Rigau and Eneko Agirre
TL;DR: We present GoLLIE, a Large Language Model trained to follow annotation guidelines. GoLLIE outperforms previous approaches on zero-shot Information Extraction and allows the user to perform inferences with annotation schemas defined on the fly. Different from previous approaches, GoLLIE is able to follow detailed definitions and does not only rely on the knowledge already encoded in the LLM.
Did ChatGPT cheat on your test? (2023)
Oscar Sainz, Jon Ander Campos, Iker García-Ferrero, Julen Etxaniz, Eneko Agirre
TL;DR: Large Language Models have seen trillions of tokens. However, who knows what is inside? Recent works have evaluated those models in many different tasks, but, did they make sure the model had not already seen the training or even the evaluation datasets? In this blog post, we show that some popular benchmark datasets are already memorized by ChatGPT and that one can prompt ChatGPT to regenerate them.