Narration Quality Evaluation on FigTalk-Gold (D1)
76Order MatchMINARD
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| MINARDRegime=MINARD (Figure + Document), Backbone=GPT-52026.06 | 76 | 83 | 84 | 81 | 82 | |
| MINARDRegime=MINARD (Figure + Document), Backbone=CLAUDE2026.06 | 75 | 82 | 79 | 76 | 78 | |
| MINARDRegime=MINARD (Figure + Document), Backbone=GEMINI2026.06 | 74 | 80 | 75 | 70 | 74 | |
| Figure-onlyRegime=Figure-only (F only), Backbone=GPT-52026.06 | 73 | 80 | 60 | 43 | 28 | |
| Figure-onlyRegime=Figure-only (F only), Backbone=CLAUDE2026.06 | 72 | 79 | 58 | 41 | 26 | |
| Figure-onlyRegime=Figure-only (F only), Backbone=GEMINI2026.06 | 71 | 78 | 56 | 39 | 24 | |
| Paper2VideoRegime=Paper2Video (Figure + Document), Backbone=GPT-52026.06 | 69 | 81 | 59 | 36 | 45 | |
| Paper2VideoRegime=Paper2Video (Figure + Document), Backbone=CLAUDE2026.06 | 68 | 79 | 57 | 33 | 42 | |
| Paper2VideoRegime=Paper2Video (Figure + Document), Backbone=GEMINI2026.06 | 66 | 77 | 54 | 31 | 39 |