Data-driven Report Generation on USAFacts (test)
3.5ReadabilityDirect
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| DirectBase Model=Qwen3-VL-32B-Instruct, Evaluation Protocol=Human Ranking2026.01 | 3.5 | 3.7 | 3.6 | 3.9 | 3.9 | 3.6 | |
| DataNarrativeBase Model=Qwen3-VL-32B-Instruct, Evaluation Protocol=Human Ranking2026.01 | 2.7 | 2.2 | 2.4 | 2.5 | 1.9 | 1.9 | |
| DeepAnalyzeBase Model=Qwen3-VL-32B-Instruct, Evaluation Protocol=Human Ranking2026.01 | 2.3 | 2 | 1.8 | 2.2 | 2.1 | 3 | |
| EvidFuseBase Model=Qwen3-VL-32B-Instruct, Evaluation Protocol=Human Ranking2026.01 | 1.8 | 2.3 | 1.8 | 1.8 | 1.5 | 1.5 |