Table 11 Experiment results of classification models
From: HsscBERT: pre-training domain model for the full text of Chinese humanity and social science
Title | Abstract | |||||
|---|---|---|---|---|---|---|
Accuracy (%) | Macro Avg (%) | Weighted Avg (%) | Accuracy (%) | Macro Avg (%) | Weighted Avg (%) | |
BERT-base-Chinese | 65.28 | 61.78 | 63.42 | 70.57 | 67.65 | 69.11 |
Chinese-Roberta-wwm-ext | 58.66 | 52.89 | 56.12 | 64.82 | 58.98 | 62.37 |
HsscBERT_e3 | 65.74 | 61.95 | 63.75 | 71.63 | 68.69 | 70.09 |
HsscBERT_e5 | 65.47 | 61.83 | 63.52 | 71.51 | 68.94 | 69.98 |
LLAMA3.1-8B | 61.42 | 56.82 | 61.22 | 64.01 | 60.90 | 62.39 |
GPT3.5-turbo | 42.56 | 4.08 | 42.14 | 44.90 | 8.30 | 43.93 |
GPT4-turbo | 62.58 | 41.36 | 60.87 | 63.54 | 43.27 | 63.13 |
