Table 3 Performance comparison of different large language models.

From: Leveraging large language models and embedding representations for enhanced word similarity computation

Method

RG65

MC30

YP130

MED38

r

ρ

r

ρ

r

ρ

r

ρ

SBERT + Bloom-7B1

0.7947

0.7724

0.7931

0.7922

0.7251

0.6973

0.7189

0.7203

SBERT + Qwen-7B-Chat-Int4

0.8188

0.8057

0.8056

0.8184

0.7373

0.7000

0.7381

0.7354

SBERT + Qwen-7B-Chat

0.8374

0.8089

0.8297

0.8429

0.7599

0.7259

0.7421

0.7428

SBERT + Deepseek-7B

0.8457

0.8225

0.8300

0.8405

0.7648

0.7325

0.7478

0.7582

SBERT + ChatGPT-3.5-turb

0.8594

0.8294

0.8403

0.8273

0.7792

0.7307

0.7511

0.7736

SBERT + ChatGPT-4

0.8723

0.8469

0.8444

0.8340

0.7820

0.7321

0.7642

0.7839

Back to article page