Question Answering on TriviaQA (F1)
72.17F1 ScoreTIPS
Evaluation Results
| Method | Links | |
|---|---|---|
| TIPSBackbone=Qwen-2.5-7B Instruct2026.03 | 72.17 | |
| PPOBackbone=Qwen-2.5-7B Instruct2026.03 | 70.88 | |
| TIPSBackbone=Qwen-2.5-3B Instruct2026.03 | 66.43 | |
| MT-GRPO*Backbone=Qwen-2.5-7B Instruct2026.03 | 66.05 | |
| PPOBackbone=Qwen-2.5-3B Instruct2026.03 | 65.75 | |
| GRPOBackbone=Qwen-2.5-7B Instruct2026.03 | 63.58 | |
| GRPOBackbone=Qwen-2.5-3B Instruct2026.03 | 63.3 | |
| MT-PPOBackbone=Qwen-2.5-3B Instruct2026.03 | 62.85 | |
| MT-PPOBackbone=Qwen-2.5-7B Instruct2026.03 | 62.08 | |
| MT-GRPO*Backbone=Qwen-2.5-3B Instruct2026.03 | 59.51 |