Leveraging Pretrained Language Models for Enhanced Entity Matching: A Comprehensive Study of Fine-Tuning and Prompt Learning Paradigms

Table 8

The loss values of different methods at each training epoch on the structured DBLP-Scholar dataset.

Epochs	FT	HP	SP
1	0.1561	0.1158	0.1314
2	0.0711	0.0663	0.0720
3	0.0477	0.0483	0.0508
4	0.0325	0.0299	0.0307
5	0.0204	0.0203	0.0203