Research map / Image generation / Text-to-image diffusion
Do DALL-E and Flamingo Understand Each Other?
Hang Li, Jindong Gu, Rajat Koner, Sahand Sharifzadeh, Volker Tresp
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 |
|---|---|---|---|
| choose the caption whose SD reconstruction is closest to the input image | frozen BLIP + SD · NoCaps overall | CIDEr: 74.1 -> 80.3 | Improved |
| filter SD images by BLIP text reconstruction | SD · NoCaps | FID: 32.37 -> 29.59 | Improved |
| finetune BLIP with SD reconstruction loss | BLIP ViT-B · NoCaps overall | CIDEr: 109.7 -> 111.8 | Improved |
| finetune BLIP-2 with SD reconstruction loss | BLIP-2 · COCO Karpathy test | CIDEr: 145.8 -> 146.4 | Improved |
| finetune BLIP ViT-L with SD reconstruction loss | BLIP ViT-L · COCO Karpathy test | SPICE: 24.3 -> 24.1 | No clear effect |
| finetune SD with BLIP text reconstruction loss | SD · NoCaps | CLIP image distance: 0.4039 -> 0.3978 | Improved |
| finetune SD with BLIP text reconstruction loss | SD · NoCaps | FID: 21.19 -> 21.96 | Hurt |
| add caption supervision loss L_TG alongside reconstruction loss (ablation) | BLIP ViT-B · NoCaps overall | CIDEr: 102.3 -> 111.8 | Improved |
| add image reconstruction loss L_IR (ablation) | BLIP ViT-B · NoCaps overall | CIDEr: 109.7 -> 111.8 | Improved |
| add text reconstruction loss L_TR (ablation) | SD · NoCaps | FID: 24.32 -> 21.96 | Improved |
| finetune SD with MSE loss on COCO pairs only | SD · NoCaps | FID: 21.19 -> 24.32 | Hurt |
Closest papers by meaning
- Image Regeneration: Evaluating Text-to-Image Model via Generating Identical Image with Multimodal Large Language Models (AAAI 2025 · similarity 0.88)
- ScImage: How good are multimodal large language models at scientific text-to-image generation? (ICLR 2025 · similarity 0.87)
- Mastering Text-to-Image Diffusion: Recaptioning, Planning, and Generating with Multimodal LLMs (ICML 2024 · similarity 0.87)
- Exploring the Deep Fusion of Large Language Models and Diffusion Transformers for Text-to-Image Synthesis (CVPR 2025 · similarity 0.87)
- LLM Blueprint: Enabling Text-to-Image Generation with Complex and Detailed Prompts (ICLR 2024 · similarity 0.86)
- Localizing and Editing Knowledge In Text-to-Image Generative Models (ICLR 2024 · similarity 0.86)
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.
