Single-Hop Question Answering on Trivia
68.1Exact MatchTree-GRPO
Evaluation Results
| Method | Links | |
|---|---|---|
| Tree-GRPOBackbone=Qwen2.5-14b2025.09 | 68.1 | |
| GSPOBackbone=Qwen2.5-14b2025.09 | 67.4 | |
| GRPOBackbone=Qwen2.5-14b2025.09 | 67.2 | |
| GSPOBackbone=Qwen2.5-7b2025.09 | 64.5 | |
| ReActBackbone=Qwen2.5-14b2025.09 | 64.2 | |
| Search-o1Backbone=Qwen2.5-14b2025.09 | 63.5 | |
| Tree-GRPOBackbone=Qwen2.5-7b2025.09 | 63.3 | |
| GRPOBackbone=Qwen2.5-7b2025.09 | 61.5 | |
| Tree-GRPOBackbone=Llama3.2-3b2025.09 | 59.9 | |
| Tree-GRPOBackbone=Qwen2.5-3b2025.09 | 59.7 | |
| GSPOBackbone=Qwen2.5-3b2025.09 | 58.8 | |
| GRPOBackbone=Llama3.2-3b2025.09 | 58.2 | |
| GRPOBackbone=Qwen2.5-3b2025.09 | 58 | |
| GSPOBackbone=Llama3.2-3b2025.09 | 57.8 | |
| Tree-GRPOBackbone=Qwen2.5-1.5b2025.09 | 57.3 | |
| ReActBackbone=Qwen2.5-7b2025.09 | 56.3 | |
| Direct InferenceBackbone=Qwen2.5-14b2025.09 | 53.1 | |
| GRPOBackbone=Qwen2.5-1.5b2025.09 | 51 | |
| GSPOBackbone=Qwen2.5-1.5b2025.09 | 48.9 | |
| Search-o1Backbone=Llama3.2-3b2025.09 | 48.4 | |
| Search-o1Backbone=Qwen2.5-7b2025.09 | 47.2 | |
| Search-o1Backbone=Qwen2.5-3b2025.09 | 44.3 | |
| ReActBackbone=Qwen2.5-3b2025.09 | 43.5 | |
| ReActBackbone=Llama3.2-3b2025.09 | 42.4 | |
| Direct InferenceBackbone=Qwen2.5-7b2025.09 | 40.8 | |
| Search-o1Backbone=Qwen2.5-1.5b2025.09 | 30.9 | |
| Direct InferenceBackbone=Llama3.2-3b2025.09 | 29.6 | |
| Direct InferenceBackbone=Qwen2.5-3b2025.09 | 28.8 | |
| Direct InferenceBackbone=Qwen2.5-1.5b2025.09 | 22.4 | |
| ReActBackbone=Qwen2.5-1.5b2025.09 | 22.1 |