Skip to main content
. Author manuscript; available in PMC: 2024 Apr 30.
Published in final edited form as: Proc Conf Assoc Comput Linguist Meet. 2023 Jul;2023:10520–10542. doi: 10.18653/v1/2023.acl-long.587

Figure 10:

Figure 10:

Three abstracts generated from model checkpoints after Relevance Calibration (Summary 1), Fine-Tuning (PRIMERA FT checkpoint, Summary 2), and after Faithfulness Calibration (Summary 3). Red Text has been annotated as being part of an intrinsic error while Purple Text is extrinsic. The annotator rated Summary 1 as the most relevant and Summary 3 the least relevant.