Question Answering on HotpotQA 1.0 (dev)
77.2EMSGIC
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| SGICBackbone=Llama2-7B-Chat2025.06 | 77.2 | 80.5 | — | — | — | — | — | — | |
| SGICBackbone=GPT-4o2025.06 | 76.5 | 70.8 | — | — | — | — | — | — | |
| SGICBackbone=GPT-4o-mini2025.06 | 74.1 | 66.7 | — | — | — | — | — | — | |
| GPT-4oFine-tuning=None2025.06 | 73.7 | 68.1 | — | — | — | — | — | — | |
| GPT-4o-miniFine-tuning=None2025.06 | 69.2 | 63 | — | — | — | — | — | — | |
| Llama2-7B-ChatTuning=LoRA Tuning2025.06 | 69.1 | 73.5 | — | — | — | — | — | — | |
| SGICBackbone=Phi-3.5-mini2025.06 | 55.3 | 60.2 | — | — | — | — | — | — | |
| Phi-3.5-miniTuning=Full Tuning2025.06 | 42.8 | 50.3 | — | — | — | — | — | — | |
| Ding (2019)2019.09 | — | — | 37.6 | 49.4 | 23.1 | 58.5 | 12.2 | 35.3 | |
| hierarchical pipeline system for MRSpipeline_mode=whole pipeline2019.09 | — | — | 46.5 | 58.8 | 39.9 | 71.5 | 26.6 | 49.2 | |
| Yang (2018)2019.09 | — | — | 24.7 | 34.4 | 5.3 | 41 | 2.5 | 17.7 |