Distractor Generation on SciQ
25.5P@1GPT-3(few-shot-COT-k-NN)
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| GPT-3(few-shot-COT-k-NN)Approach=Prompting2026.04 | 25.5 | 8.5 | 17.03 | 31.08 | 35.89 | 0.1 | 0 | |
| GPT-3(few-shot-random)Approach=Prompting2026.04 | 25.1 | 8.37 | 16.13 | 30.17 | 34.57 | 0.2 | 0.1 | |
| T5(contrast)Approach=Text2Text2026.04 | 25 | 8.33 | 17.73 | 31.42 | 36.68 | 8.8 | 6.5 | |
| GPT-3(few-shot-k-NN)Approach=Prompting2026.04 | 24.9 | 8.3 | 16.53 | 30.45 | 35.24 | 0.1 | 0 | |
| GPT-3(few-shot-COT-random)Approach=Prompting2026.04 | 24.5 | 8.17 | 15.7 | 29.53 | 33.82 | 0.2 | 0 | |
| GPT-3(zero-shot)Approach=Prompting2026.04 | 24.4 | 8.13 | 15.47 | 29.3 | 33.45 | 0.1 | 0.3 | |
| T5(multi-task)Approach=Text2Text2026.04 | 21.2 | 7.07 | 14.27 | 25.47 | 28.52 | 18.9 | 11.1 | |
| TinyLlamaApproach=Instruction Tuning2026.04 | 20.5 | 6.83 | 15.4 | 27.07 | 32.73 | 2.4 | 1.9 | |
| T5-CSG-DS(beam)Approach=CSG-DS2026.04 | 20.3 | 6.77 | 14.33 | 26.2 | 31.3 | 29.1 | 0.2 | |
| T5(base)Approach=Text2Text2026.04 | 18.9 | 6.3 | 13.77 | 23.23 | 26.66 | 23.3 | 12.8 | |
| GPT-3-CSG-DS(clustering)Approach=CSG-DS2026.04 | 10.8 | 3.6 | 5.2 | 13.33 | 15.73 | 0.3 | 0 | |
| T5-CSG-DS(clustering)Approach=CSG-DS2026.04 | 10.5 | 3.5 | 6.3 | 14.35 | 17.95 | 22.3 | 0 |