notesum.ai

Published at December 4

TOOL-ED: Enhancing Empathetic Response Generation with the Tool Calling Capability of LLM

cs.CL

Released Date: December 4, 2024

Authors: Huiying Cao¹, Yiqun Zhang, Shi Feng, Xiaocui Yang, Daling Wang, Yifei Zhang

Aff.: ¹Northeastern University, Shenyang, China

Arxiv: http://arxiv.org/pdf/2412.03096v1

Refer to caption

Model	BLEU-1/2/3/4	B-S	ROU-1/2/L.	Dist-1/2
CEM*	0.1332/0.0630/0.0351/0.0209	0.8603	0.1625/0.0418/0.1508	0.0066/0.0299
EmpDG	0.1857/0.0873/0.051/0.0316	0.8611	0.1697/0.0453/0.155	0.0181/0.0694
HEF	0.1164/0.0365/0.0171/0.0084	0.8539	0.1471/0.0186/0.1196	0.0336/0.2096
KEMP*	0.1902/0.0767/0.0362/0.0195	0.8531	0.1618/000296/0.1418	0.0041/0.0204
IAMM	0.1505/0.0651/0.0360/0.0215	0.8633	0.1580/0.0347/0.1446	0.0098/0.0302
MOEL	0.1726/0.0773/0.0433/0.0264	0.8582	0.1653/0.0367/0.1502	0.0038/0.0160
MIME	0.2182/0.096/0.0492/0.02796	0.8613	0.1882/0.0387/0.1642	0.0032/0.0124
vicuna_base	0.1106/0.0412/0.0209/0.0111	0.8522	0.1525/0.0271/0.1254	0.0274/0.1833
vicuna_oneshot	0.1204/0.0449/0.0227/0.012	0.8544	0.1550/0.0259/0.1273	0.0283/0.1793
vicuna_lora	0.1935/0.0879/0.0514/0.0316	0.8723	0.1740/0.0403/0.1561	0.0296/0.1424
\hdashlinevicuna_tool_comet*	0.1859/0.0903/0.0544/0.0339	0.8755	0.1888/0.0543/0.1735	0.0299/0.1453
vicuna_tool_cicero*	0.1877/0.0904/0.0545/0.0341	0.8751	0.1889/0.0534/0.1732	0.0300/0.1447
qwen_base	0.1046/0.0349/0.016/0.0077	0.8456	0.1356/0.0186/0.1072	0.2322/0.1720
qwen_oneshot	0.1158/0.0253/0.1780/0.0078	0.8497	0.1431/0.0211/0.1158	0.0281/0.1904
qwen_lora	0.1861/0.0883/0.0517/0.0317	0.8760	0.1846/0.0483/0.1696	0.0249/0.1172
\hdashlineqwen_tool_comet*	0.1936/ 0.0961/0.0579/0.036	0.8765	0.1941/0.0579/0.1793	0.0283/0.1358
qwen_tool_cicero*	0.1907/0.0943/0.0569/0.0352	0.8766	0.1913/0.0563/0.1768	0.0289/0.1395