Question Answering on EXPERTQA (test)
19.27Claim RecallChatGPT
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| ChatGPTLearning Mode=ICL2024.02 | 19.27 | 47.79 | 47.3 | 48.68 | |
| M_dist + f.g.RLRefinement=Fine-grained Reinforcement Learning2024.02 | 15.53 | 49.73 | 51.11 | 45.92 | |
| M_dist + f.g.RSRefinement=Fine-grained Reward Shaping2024.02 | 15.48 | 59.46 | 57.58 | 44.67 | |
| M_distType=Distilled baseline2024.02 | 15.28 | 49.03 | 46.22 | 40.63 | |
| M_dist + f.g.(RS+RL)Refinement=Combined RS and RL2024.02 | 15.23 | 58.94 | 59.8 | 42.13 | |
| LLAMA-2-7BLearning Mode=ICL2024.02 | 10.32 | 10.09 | 7.79 | 34.27 |