Temporal Question Answering on TimeQA Easy v1
58R-1 ScoreCoT+RL pipeline
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| CoT+RL pipelinePrompt Type=W/ ORIGINAL CONTEXT (C), Model=Qwen2.5-1.5B, Training Method=RL2026.02 | 58 | 45.6 | 57.8 | 72.2 | 43.41 | |
| SFTPrompt Type=W/ TIME SUB-C, Model=Qwen2.5-7B2026.02 | 48.4 | 31.4 | 48.3 | 60.9 | 1.01 | |
| CoT+RL pipelinePrompt Type=W/ ORIGINAL CONTEXT (C), Model=Qwen2.5-0.5B, Training Method=RL2026.02 | 46.9 | 35.5 | 46.8 | 66 | 32.48 | |
| SFTPrompt Type=W/ TIME SUB-C, Model=Qwen2.5-1.5B2026.02 | 44.2 | 28.3 | 44 | 58.7 | 1.08 | |
| SFTPrompt Type=W/ ORIGINAL CONTEXT (C), Model=Qwen2.5-7B2026.02 | 39.6 | 19.6 | 39.6 | 56.4 | 3.7 | |
| SFTPrompt Type=W/ ORIGINAL CONTEXT (C), Model=Qwen2.5-1.5B2026.02 | 35.9 | 18.3 | 35.7 | 54.3 | 1.81 | |
| SFTPrompt Type=W/ KGS (WHOLE C) (FAISS), Model=Qwen2.5-7B2026.02 | 32.2 | 19.8 | 32 | 51.1 | 1.4 | |
| LoRA (r=32)Prompt Type=W/ ORIGINAL CONTEXT (C), Model=Qwen2.5-1.5B2026.02 | 32.1 | 18.4 | 31.6 | 51.9 | 1.29 | |
| SFTPrompt Type=W/ KGS (WHOLE C) (FAISS), Model=Qwen2.5-1.5B2026.02 | 28.5 | 15.9 | 28.4 | 49 | 0.07 | |
| SFTPrompt Type=W/ KGS (CHUNKED C) (FAISS), Model=Qwen2.5-7B2026.02 | 27.1 | 15.4 | 27 | 49.6 | 0.43 | |
| SFTPrompt Type=W/ KGS (CHUNKED C) (FAISS), Model=Qwen2.5-1.5B2026.02 | 23.8 | 15.1 | 23.6 | 46.3 | 2.5 | |
| ClassifierPrompt Type=W/ ORIGINAL CONTEXT (C), Model=Qwen2.5-7B2026.02 | 15.9 | 12.6 | 15.8 | 38.7 | 11.41 |