Prompts for publication-quality research figures
Browse 60+ curated prompts for transformer architectures, training pipelines, statistical charts, mechanism diagrams and more — copy, tweak, generate.
59 prompts
How to write a great prompt
The highest-rated prompts share four ingredients. Lead with the goal, list the components, lock the layout, then specify the style.
1. Goal
One sentence telling the model exactly what figure to produce. Keep it under 25 words.
2. Components
Bullet the modules, stages, data series or molecules to include — generators reproduce labels word for word.
3. Layout
Pick one orientation (left-to-right, top-down, 2x2 grid, radial) and stick to it; mixed layouts confuse the model.
4. Style
Specify palette, typography, target venue (NeurIPS, Nature, IEEE) and any preservation rules for image-to-image edits.
Frequently asked questions
Browse all prompt examples
Direct links to every category and featured prompt.
- Transformer Encoder-Decoder Architecture
- Multi-Agent Collaboration Framework
- Decoder-Only LLM Architecture (GPT-Style)
- Vision Transformer (ViT) Architecture
- Diffusion Model Forward & Reverse Process
- GAN Training Loop
- Retrieval-Augmented Generation (RAG) Pipeline
- Mixture-of-Experts (MoE) Layer
- Graph Neural Network Message Passing
- U-Net Encoder-Decoder for Segmentation
- End-to-End Segmentation Training Pipeline
- Multi-Agent Code-Generation Orchestrator
- Therapeutic siRNA Design Pipeline
- AI Protein Design Workflow
- Multimodal Fusion Pipeline (Image + Text)
- Federated Learning Round
- LLM Evaluation Framework
- Risk Analysis Framework: FMEA / FTA / DSM
- Multi-Phase Qualification Framework
- Wildfire Risk & Evacuation Framework