Abstract :
–
During Sébastien Durna’s internship, we asked you to share a reading list of HP papers of interest for you and your team. After that, we have ingested those papers - and more - into Large Language Models (LLMs) and Natural Language Processing (NLP) tools to generate pre-annotations. Those annotations will further serve to train BERT models to perform the Named Entity Extraction (NER) task, such as HelioBERT. In order to ensure the quality of those pre-annotations, we would like to invite you to participate in the first annotation review campaign. This seminar will be split into three parts : first, we will focus on the objectives of this work (Liza’s thesis, starting in October : can LLMs reproduce heliophysics pipelines ?). Second, we will introduce the pre-annotation pipeline developed by Sébastien. Third, there will be a demo session introducing the annotation guidelines and how to participate in this campaign on Label Studio.
Authors :
Liza Fretel, Sébastien Durna, Baptiste Cecconi
