CH Chiang, H Lee - arXiv preprint arXiv:2402.05629, 2024 - arxiv.org
… Long-form generations from large language models (LLMs) contains a … The factuality of the whole generation is the percentage of … in the generation. Figure 1 shows such a case, where …
LFR Ribeiro, Y Zhang, C Gardent… - Transactions of the …, 2020 - direct.mit.edu
… In contrast, models based on local aggregation learn the … graph and how their naturallanguage descriptions are related … build a graph encoder to aggregate the local context by utilizing …
Y Jeon, GG Lee - arXiv preprint arXiv:2403.04111, 2024 - arxiv.org
… This paper explores the task of languageagnostic speaker replication, a novel endeavor … language they are speaking. Towards this end, we introduce a multi-level attention aggregation …
S Li, T Xiao, H Li, B Zhou, D Yue… - Proceedings of the …, 2017 - openaccess.thecvf.com
… Since there is no person dataset or benchmark with textual description available, we collect a large-scale person description dataset with detailed naturallanguage annotations and …
… For languagegeneration tasks, we find that RAG models … To test RAG’s naturallanguage generation (NLG) in a … a retrieved item, but on aggregating content from several pieces of …
… that can be applied to naturallanguagegeneration tasks of similar nature, including sum… Instead, we investigate aggregation functions to consolidate layer-wise information, forming …
… (EQA) is a fundamental task in naturallanguage processing (Parsing… aggregation problem as a free text sequence aggregation problem and adopt the free text sequence aggregation …
P Mehta, P Majumder - Information Processing & Management, 2018 - Elsevier
… conditions under which performance of an existing aggregation system can be improved. In … other existing aggregation techniques. As an example, we show that all these aggregation …
… Naturallanguagegeneration tasks can be generally divided into two phases: content … Aggregation via set partitioning for naturallanguagegeneration. In HLT-NAACL, 359–366. …