Producing publication-ready illustrations is a labor-intensive bottleneck within the analysis workflow. Whereas AI scientists can now deal with literature critiques and code, they wrestle to visually talk advanced discoveries. A analysis staff from Google and Peking College introduce new framework known as ‘PaperBanana‘ which is altering that through the use of a multi-agent system to automate high-quality educational diagrams and plots.
5 Specialised Brokers: The Structure
PaperBanana doesn’t depend on a single immediate. It orchestrates a collaborative staff of 5 brokers to rework uncooked textual content into skilled visuals.
Part 1: Linear Planning
- Retriever Agent: Identifies the 10 most related reference examples from a database to information the model and construction.
- Planner Agent: Interprets technical methodology textual content into an in depth textual description of the goal determine.
- Stylist Agent: Acts as a design advisor to make sure the output matches the “NeurIPS Look” utilizing particular shade palettes and layouts.
Part 2: Iterative Refinement
- Visualizer Agent: Transforms the outline into a visible output. For diagrams, it makes use of picture fashions like Nano-Banana-Professional. For statistical plots, it writes executable Python Matplotlib code.
- Critic Agent: Inspects the generated picture towards the supply textual content to search out factual errors or visible glitches. It gives suggestions for 3 rounds of refinement.
Beating the NeurIPS 2025 Benchmark
The analysis staff launched PaperBananaBench, a dataset of 292 take a look at instances curated from precise NeurIPS 2025 publications. Utilizing a VLM-as-a-Choose strategy, they in contrast PaperBanana towards main baselines.
| Metric | Enchancment over Baseline |
| Total Rating | +17.0% |
| Conciseness | +37.2% |
| Readability | +12.9% |
| Aesthetics | +6.6% |
| Faithfulness | +2.8% |
The system excels in ‘Agent & Reasoning’ diagrams, attaining a 69.9% total rating. It additionally gives an automatic ‘Aesthetic Guideline’ that favors ‘Soft Tech Pastels’ over harsh main colours.
Statistical Plots: Code vs. Picture
Statistical plots require numerical precision that customary picture fashions typically lack. PaperBanana solves this by having the Visualizer Agent write code as an alternative of drawing pixels.
- Picture Technology: Excels in aesthetics however typically suffers from ‘numerical hallucinations’ or repeated parts.
- Code-Primarily based Technology: Ensures 100% information constancy through the use of the Matplotlib library to render the ultimate plot.
Area-Particular Aesthetic Preferences in AI Analysis
In accordance with the PaperBanana model information, aesthetic decisions typically shift based mostly on the analysis area to match the expectations of various scholarly communities.
| Analysis Area | Visible ‘Vibe‘ | Key Design Components |
| Agent & Reasoning | Illustrative, Narrative, “Friendly” | 2D vector robots, human avatars, emojis, and “User Interface” aesthetics (chat bubbles, doc icons) |
| Laptop Imaginative and prescient & 3D | Spatial, Dense, Geometric | Digicam cones (frustums), ray traces, level clouds, and RGB shade coding for axis correspondence |
| Generative & Studying | Modular, Stream-oriented | 3D cuboids for tensors, matrix grids, and “Zone” methods utilizing mild pastel fills to group logic |
| Concept & Optimization | Minimalist, Summary, “Textbook” | Graph nodes (circles), manifolds (planes), and a restrained grayscale palette with single spotlight colours |
Comparability of Visualization Paradigms
For statistical plots, the framework highlights a transparent trade-off between utilizing a picture era mannequin (IMG) versus executable code (Coding).
| Characteristic | Plots by way of Picture Technology (IMG) | Plots by way of Coding (Matplotlib) |
| Aesthetics | Typically increased; plots look extra “visually appealing” | Skilled and customary educational look |
| Constancy | Decrease; vulnerable to “numerical hallucinations” or ingredient repetition | 100% correct; strictly represents the uncooked information supplied |
| Readability | Excessive for sparse information however struggles with advanced datasets | Constantly excessive; handles dense or multi-series information with out error |
Key Takeaways
- Multi-Agent Collaborative Framework: PaperBanana is a reference-driven system that orchestrates 5 specialised brokers—Retriever, Planner, Stylist, Visualizer, and Critic—to rework uncooked technical textual content and captions into publication-quality methodology diagrams and statistical plots.
- Twin-Part Technology Course of: The workflow consists of a Linear Planning Part to retrieve reference examples and set aesthetic pointers, adopted by a 3-round Iterative Refinement Loop the place the Critic agent identifies errors and the Visualizer agent regenerates the picture for increased accuracy.
- Superior Efficiency on PaperBananaBench: Evaluated towards 292 take a look at instances from NeurIPS 2025, the framework outperformed vanilla baselines in Total Rating (+17.0%), Conciseness (+37.2%), Readability (+12.9%), and Aesthetics (+6.6%).
- Precision-Centered Statistical Plots: For statistical information, the system switches from direct picture era to executable Python Matplotlib code; this hybrid strategy ensures numerical precision and eliminates “hallucinations” frequent in customary AI picture turbines.
Try the Paper and Repo. Additionally, be at liberty to observe us on Twitter and don’t neglect to affix our 100k+ ML SubReddit and Subscribe to our E-newsletter. Wait! are you on telegram? now you’ll be able to be a part of us on telegram as nicely.
