Short-text Multi-doc Question Answering on RGB noise robustness testbed (test)
96EM (Noise 0)Our model (PAM QA)
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Our model (PAM QA)2023.11 | 96 | 90.67 | 90 | 85.5 | 67.33 | |
| GPT3.5-Turbo2023.11 | 95.67 | 94.67 | 91 | 87.67 | 70.67 | |
| Qwen-14B-Chat2023.11 | 94.67 | 92 | 88 | 85.3 | 69.67 | |
| Baichuan2-13B-Chat2023.11 | 93 | 90.33 | 89 | 82.33 | 63.33 | |
| ChatGLM3-6Bstatus=Released after current work2023.11 | 91.67 | 90 | 89 | 84.67 | 66.33 | |
| ChatGLM2-6B2023.11 | 86.67 | 82.33 | 76.67 | 72.33 | 54 |