notesum.ai

Published at November 6

QUILL: Quotation Generation Enhancement of Large Language Models

cs.CL

cs.AI

Released Date: November 6, 2024

Authors: Jin Xiao¹, Bowei Zhang¹, Qianyu He², Jiaqing Liang¹, Feng Wei³, Jinglei Chen³, Zujie Liang³, Deqing Yang¹, Yanghua Xiao²

Aff.: ¹School of Data Science, Fudan University; ²Shanghai Key Laboratory of Data Science, School of Computer Science, Fudan University; ³MYbank, Ant Group

Arxiv: http://arxiv.org/abs/2411.03675v1

Model	$S_{a}$	$S_{c}$	$S_{m}$	$S_{f}$	$S_{n}$	$Avg$
Chinese-oriented Models
ChatGLM3-6B	0.56	0.20	0.73	0.73	0.74	0.59
Qwen1.5-7B-Chat	0.63	0.15	0.71	0.72	0.70	0.58
Qwen1.5-14B-Chat	0.66	0.16	0.69	0.70	0.74	0.60
Qwen1.5-72B-Chat	0.72	0.16	0.70	0.71	0.71	0.60
English-oriented Models
Mixture-7B-v0.2	0.77	0.08	0.69	0.74	0.58	0.57
Llama2-7B-Chat-hf	0.46	0.15	0.73	0.72	0.77	0.57
Llama2-13B-Chat-hf	0.48	0.15	0.73	0.73	0.78	0.58
Llama2-70B-Chat-hf	0.60	0.11	0.69	0.71	0.66	0.55
Close-source Models
GPT-3.5-turbo	0.62	0.11	0.70	0.73	0.65	0.56
GPT-4o	0.79	0.23	0.71	0.74	0.62	0.62
Ours	1.00	1.00	0.77	0.77	0.81	0.87