Table 14 Custom quality evaluation results comparison.
From: Localized large language model TCNNet 9B for Taiwanese networking and cybersecurity
Base model | Technical accuracy | Relevance | Detail and depth | Clarity and coherence |
|---|---|---|---|---|
gpt-4o | 3.871 | 3.912 | 4.121 | 4.324 |
TCNNet-9B | 3.631 | 3.839 | 4.087 | 4.110 |
gemini-1.5-pro | 3.214 | 3.452 | 4.021 | 4.016 |
Yi-1.5-9B | 2.845 | 2.820 | 3.423 | 3.786 |
