Multi-level Report Generation on USAFacts
3.47ReadabilityDeepAnalyze
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| DeepAnalyzeLevel=chart, Base Models=Qwen3-235B-A22B-Instruct-2507 & Qwen2.5-VL 72B-Instruct2026.01 | 3.47 | 3.67 | — | — | — | — | |
| DirectLevel=chart, Base Models=Qwen3-235B-A22B-Instruct-2507 & Qwen2.5-VL 72B-Instruct2026.01 | 3.2 | 2.83 | — | — | — | — | |
| DataNarrativeLevel=chart, Base Models=Qwen3-235B-A22B-Instruct-2507 & Qwen2.5-VL 72B-Instruct2026.01 | 2.17 | 2.23 | — | — | — | — | |
| EvidFuseLevel=chart, Base Models=Qwen3-235B-A22B-Instruct-2507 & Qwen2.5-VL 72B-Instruct2026.01 | 1.17 | 1.27 | — | — | — | — | |
| DataNarrativeLevel=chapter, Base Models=Qwen3-235B-A22B-Instruct-2507 & Qwen2.5-VL 72B-Instruct2026.01 | — | — | 2.7 | 2.33 | — | — | |
| DataNarrativeLevel=report, Base Models=Qwen3-235B-A22B-Instruct-2507 & Qwen2.5-VL 72B-Instruct2026.01 | — | — | — | — | 2.07 | 2.8 | |
| DeepAnalyzeLevel=chapter, Base Models=Qwen3-235B-A22B-Instruct-2507 & Qwen2.5-VL 72B-Instruct2026.01 | — | — | 3.87 | 3.7 | — | — | |
| DeepAnalyzeLevel=report, Base Models=Qwen3-235B-A22B-Instruct-2507 & Qwen2.5-VL 72B-Instruct2026.01 | — | — | — | — | 3.33 | 3.43 | |
| DirectLevel=chapter, Base Models=Qwen3-235B-A22B-Instruct-2507 & Qwen2.5-VL 72B-Instruct2026.01 | — | — | 2.37 | 2.9 | — | — | |
| DirectLevel=report, Base Models=Qwen3-235B-A22B-Instruct-2507 & Qwen2.5-VL 72B-Instruct2026.01 | — | — | — | — | 3.4 | 2.57 | |
| EvidFuseLevel=chapter, Base Models=Qwen3-235B-A22B-Instruct-2507 & Qwen2.5-VL 72B-Instruct2026.01 | — | — | 1.07 | 1.07 | — | — | |
| EvidFuseLevel=report, Base Models=Qwen3-235B-A22B-Instruct-2507 & Qwen2.5-VL 72B-Instruct2026.01 | — | — | — | — | 1.2 | 1.2 |