Question Answering on MuSiQue (EM)
17.05Exact Match (EM)TIPS
Evaluation Results
| Method | Links | |
|---|---|---|
| TIPSBackbone=Qwen-2.5 Instruct, Model Size=7B2026.03 | 17.05 | |
| PPOBackbone=Qwen-2.5 Instruct, Model Size=7B2026.03 | 8.94 | |
| GRPOBackbone=Qwen-2.5 Instruct, Model Size=3B2026.03 | 8.73 | |
| TIPSBackbone=Qwen-2.5 Instruct, Model Size=3B2026.03 | 8.73 | |
| MT-GRPO*Backbone=Qwen-2.5 Instruct, Model Size=7B2026.03 | 7.99 | |
| GRPOBackbone=Qwen-2.5 Instruct, Model Size=7B2026.03 | 7.2 | |
| MT-PPOBackbone=Qwen-2.5 Instruct, Model Size=7B2026.03 | 6.91 | |
| PPOBackbone=Qwen-2.5 Instruct, Model Size=3B2026.03 | 6.37 | |
| MT-PPOBackbone=Qwen-2.5 Instruct, Model Size=3B2026.03 | 6.16 | |
| MT-GRPO*Backbone=Qwen-2.5 Instruct, Model Size=3B2026.03 | 5.17 |