Text-to-interleaved generation on RecipeGen (test)
4.35Temporal Coherence (GPT-4o)Doubao
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| DoubaoModel Type=Proprietary Large Multimodal Model2025.12 | 4.35 | 4.1 | 4.25 | 4.05 | 4.95 | 4.65 | 0.25 | |
| LoomFramework=Multi-turn framework2025.12 | 4.25 | 4.15 | 3.75 | 3.35 | 4.7 | 4.3 | 0.269 | |
| Anole2025.12 | 1.55 | 1.05 | 1.35 | 1.05 | 1.95 | 1.35 | 0.219 | |
| BagelFramework=Multi-turn dialogue framework, Prompts=Step-by-step textual prompts (GPT-4o generated)2025.12 | 1.4 | 1.25 | 1.55 | 1.05 | — | — | 0.217 | |
| Janus-ProFramework=Multi-turn dialogue framework, Prompts=Step-by-step textual prompts (GPT-4o generated)2025.12 | 1.05 | 1.1 | 1.1 | 1 | — | — | 0.105 |