Question Answering on TAQA 2022
27.9F2022 ScoreLLaMA2-70B + target-year finetuning
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| LLaMA2-70B + target-year finetuningAlignment Target Year=20222024.02 | 27.9 | 40.7 | 55.4 | |
| LLaMA2-70B + time-aware promptingAlignment Target Year=20222024.02 | 27.4 | 40.3 | 54.1 | |
| LLaMA2-70B + temporal-adaptive finetuningAlignment Strategy=adaptive2024.02 | 25.7 | 40 | 57.9 | |
| LLaMA2-70B + target-year finetuningAlignment Target Year=20212024.02 | 20.8 | 39.3 | 56.7 | |
| LLaMA2-70B + time-aware promptingAlignment Target Year=20212024.02 | 19.7 | 38.9 | 56.7 | |
| LLaMA2-70BTraining Cutoff=Sep. 20222024.02 | 17.2 | 34.1 | 55.9 | |
| ChatGPTTraining Cutoff=Sep. 20212024.02 | 17.1 | 32.5 | 46 | |
| Tülu2-70BTraining Cutoff=-2024.02 | 16.2 | 32.2 | 50.1 | |
| LLaMA1-65BTraining Cutoff=Aug. 20222024.02 | 12.1 | 26.8 | 54.5 | |
| GPT-3Training Cutoff=Sep. 20212024.02 | 10 | 22.9 | 46.5 |