Research map / Image generation / Text-to-image diffusion
Interleaving Reasoning for Better Text-to-Image Generation
Wenxuan Huang, Shuang Chen, Zheyong Xie, Shaosheng Cao, SHIXIANG TANG, Yufan Shen, Qingyu Yin, Wenbo Hu, Xiaoman Wang, Yuntian Tang, Junbo Qiao, Hangyu Guo and 6 more
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add high-quality image data training to BAGEL w/ self-CoT (ablation) | BAGEL w/ self-CoT · WISE | overall score: 0.70 -> 0.73 | Improved |
| add high-quality image training (ablation) | BAGEL w/ self-CoT · GenAI-Bench | overall score: 0.81 -> 0.80 | No clear effect |
| train on full interleaving reasoning thinking-image trajectories (ablation) | BAGEL w/ self-CoT + image training · WISE | overall score: 0.73 -> 0.76 | Improved |
| train on full interleaving reasoning thinking-image trajectories (ablation) | BAGEL w/ self-CoT + image training · GenAI-Bench | overall score: 0.80 -> 0.83 | Improved |
| add decomposed learning modes with two-stage training (ablation) | IRG · WISE | overall score: 0.76 -> 0.77 | Improved |
| add decomposed learning modes with two-stage training (ablation) | IRG · TIIF testmini (short) | overall score: 73.90 -> 76.00 | Improved |
| add reflection (second) turn to IRG (ablation) | IRG · WISE | overall score: 0.79 -> 0.77 | No clear effect |
| add reflection (second) turn to IRG (ablation) | IRG · WISE (MLLM judges) | avg rank score: 36.7% -> 63.3% | Improved |
| IRG vs BAGEL w/ self-CoT | BAGEL · GenEval | overall score: 0.79 -> 0.85 | Improved |
| IRG vs FLUX.1-dev on text rendering | IRG · OneIG-EN | text rendering: 0.523 -> 0.377 | Hurt |
| IRG vs BAGEL w/ self-CoT | BAGEL · OneIG-EN | overall score: 0.324 -> 0.415 | Improved |
Closest papers by meaning
- MMMG: A Massive, Multidisciplinary, Multi-Tier Generation Benchmark for Text-to-Image Reasoning (NeurIPS 2025 · similarity 0.92)
- Thinking-while-Generating: Interleaving Textual Reasoning throughout Visual Generation (CVPR 2026 · similarity 0.92)
- ROVER: Benchmarking Reciprocal Cross-Modal Reasoning for Omnimodal Generation (ICLR 2026 · similarity 0.91)
- DiffThinker: Towards Generative Multimodal Reasoning with Diffusion Models (ICML 2026 · similarity 0.91)
- GIR-Bench: Versatile Benchmark for Generating Images with Reasoning (ICLR 2026 · similarity 0.91)
- WEAVE: Unleashing and Benchmarking the In-context Interleaved Comprehension and Generation (CVPR 2026 · similarity 0.90)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
