Research map / Image generation / Text-to-image diffusion
Towards Language-Free Training for Text-to-Image Generation
Yufan Zhou, Ruiyi Zhang, Changyou Chen, Chunyuan Li, Chris Tensmeyer, Tong Yu, Jiuxiang Gu, Jinhui Xu, Tong Sun
Check 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 both ConG and ConD contrastive losses (ablation) | LAFITE with fixed perturbations · MS-COCO language-free | IS: 14.79 -> 27.20 | Improved |
| Add ConG contrastive loss (ablation) | LAFITE with fixed perturbations · MS-COCO language-free | IS: 22.28 -> 27.20 | Improved |
| Add ConD contrastive loss (ablation) | LAFITE with fixed perturbations · MS-COCO language-free | IS: 17.78 -> 27.20 | Improved |
| Add both ConG and ConD contrastive losses (ablation) | LAFITE with trainable perturbations · MS-COCO language-free | IS: 11.05 -> 22.23 | Improved |
| Add ConG contrastive loss (ablation) | LAFITE with trainable perturbations · MS-COCO language-free | IS: 19.14 -> 22.23 | Improved |
| Add ConD contrastive loss (ablation) | LAFITE with trainable perturbations · MS-COCO language-free | IS: 20.02 -> 22.23 | Improved |
| Use both the real-or-fake and semantic-alignment discriminator terms (ablation) | LAFITE with fixed perturbations · MS-COCO language-free | FID: 24.85 -> 18.04 | Improved |
| Use both the real-or-fake and semantic-alignment discriminator terms (ablation) | LAFITE with fixed perturbations · MS-COCO language-free | FID: 25.42 -> 18.04 | Improved |
| Use trainable instead of fixed perturbations (ablation) | LAFITE · MS-COCO language-free | IS: 27.20 -> 22.23 | Hurt |
| Use trainable instead of fixed perturbations (ablation) | LAFITE fine-tuned from a pre-trained model · CUB and MM CelebA-HQ | FID: CUB: 35.99 -> 37.91; MM CelebA-HQ: 15.74 -> 29.42 | Mixed |
| Use CLIP-ViT/B-16 instead of CLIP-ViT/B-32 (ablation) | LAFITE · MS-COCO supervised | FID: 8.62 -> 8.12 | Improved |
| Use a CLIP text encoder instead of RoBERTa-Base (ablation) | LAFITE fully supervised · MS-COCO | IS: 15.95 -> 24.54 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- DF-GAN — DF-GAN: A Simple and Effective Baseline for Text-to-Image Synthesis (CVPR 2022) · better on 2 of 2 shared results
- CogView — CogView: Mastering Text-to-Image Generation via Transformers (NeurIPS 2021) · better on 2 of 2 shared results
- DM-GAN — DM-GAN: Dynamic Memory Generative Adversarial Networks for Text-To-Image Synthesis (CVPR 2019) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Mining Fine-Grained Image-Text Alignment for Zero-Shot Captioning via Text-Only Training (AAAI 2024 · similarity 0.87)
- Customization Assistant for Text-to-Image Generation (CVPR 2024 · similarity 0.87)
- A Comprehensive Study of Decoder-Only LLMs for Text-to-Image Generation (CVPR 2025 · similarity 0.87)
- Text-Printed Image: Bridging the Image-Text Modality Gap for Text-centric Training of Large Vision-Language Models (CVPR 2026 · similarity 0.87)
- LLM Blueprint: Enabling Text-to-Image Generation with Complex and Detailed Prompts (ICLR 2024 · similarity 0.87)
- TIGeR: Unifying Text-to-Image Generation and Retrieval with Large Multimodal Models (ICLR 2025 · similarity 0.87)
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.
