visualization

The paper proposes Smooth Rank t-SNE (SR-tSNE), a t-SNE variant with a differentiable sigmoid rank term that preserves the ordering of neighbors, plus an Ordinal DLP metric; it outperforms 12 baselines across eight benchmarks and improves downstream tasks like label transfer, single-cell clustering, and algorithmic trading.

A user praises the "Chronological Map of Jiangshan" website — an interactive timeline application of Chinese historical maps spanning from the Warring States period to modern times, supporting the evolution of territories and administrative divisions, comparison of ancient and modern place names, scroll-to-zoom, and event jumping. The project is open-sourced on GitHub.

The author compiled and shared 10 open-source AI education projects they have been following recently, covering interactive labs for primary and secondary schools in China, secondary school science curricula abroad, and AI visualization learning tools, arguing that AI will reinvent education.

维护者分享了基于 2xRTX 3090 的本地 LLM Pareto 前沿基准图表，指出自 DeepSeek-V4-Flash-Vision-Exp 和 Qwen3.8-Flash-Next 发布两个月来排名未变，并邀请社区协作完善细粒度本地模型能力测绘。

The author developed an educational visualization page that reveals the six core stages of AI search, from understanding user needs to displaying citations, helping users intuitively understand the principles of GEO (Generative Engine Optimization) and how AI selects content.

展示了一个16神经元小网络逐步学会分离双螺旋的可视化视频，并介绍其由开源的SenseNova 6.8 Code2Video模型通过代码自动生成，包括叙事节奏。

A content creator recommends the open-source Agent Skill "diagram-design," which automatically renders long-form text, processes, and relationships into professional single-file HTML+SVG diagrams. It supports 45 chart types, brand customization, and imports from draw.io/Mermaid/Excalidraw, and has already garnered nearly 60,000 Stars.

A free, open-source WebGL explorer that visualizes your X (Twitter) activity over time using your X data export, with Birdclaw powering the backend, helping surface inflection points, top posts, topic clusters, and trends.

Recommended an open-source project, jiangshan-map (Jiangshan Chronology), which brings over 2,400 years and 18 historical periods of Chinese history into a single interactive map. It supports dragging a timeline to view territorial evolution, and clicking on ancient cities and passes to see their names through the ages and related major events. Great for learning history.

I built an open-source instrument for exploring LLM internals combining research on jspace, j++, LARQL/VINDEX, and Cammarata's neuron poles, released as logos-theodolite on GitHub with j-lenses originally fitted for gemma4-31b.

Whiteboard 是一款开源的 IDE，定位为「为深思熟虑的软件设计而生的画布」，可在统一界面中审查交互式可视化、代码 diff 与 AI 代理轨迹。

Runloop 是一个本地优先、开源的 3D 小镇可视化工具，把 Claude Code 代理会话以建筑形式呈现，支持实时观察、回放和对比运行记录。

一篇关于复数单位根的教学笔记，从实数开方出发，介绍用极坐标计算 n 次单位根、其在单位圆上的旋转几何意义，以及与正多边形和群论的联系，并配有可交互的可视化图示。

介绍了在浏览器中运行的交互式线性规划工具，让用户可以直接在网页环境里构建、求解并可视化线性规划问题。

An interactive tool that visualizes regular expressions as a 3D marble run, making regex patterns easier to understand through playful, animated mechanics.

开源项目 Anatomy 是一个 Claude Code 技能，能将抽象技术概念转化为可交互的等距机械图（基于 SVG/WebGL），开源一周即获 12,000 stars，并兼容 OpenCode、Codex 等多种 AI 编码环境。

A free, MIT-licensed open-source tool called Anima Agent skill that turns a single biology prompt into detailed, interactive scientific visualizations—such as switchable mechanisms, inspectable molecular structures, and animated processes—rendered as standalone HTML or SVG. The skill instructs Claude to research the biology before drawing, and includes a reusable rendering toolkit with working examples.

开发者把家里的 AI 助手和编码代理做成像素办公室里的小同事，用等距仪表盘实时查看各代理的状态，现已开源并支持自托管。

Phloem is a developer tool that visualizes Git history using slime-mold-style growth algorithms, turning commit graphs into organic, intuitive representations.

MathVis introduces a 512-task benchmark evaluating LLMs' ability to generate mathematical diagrams and perform diagram-assisted reasoning, with GPT-6-astra, Claude Fable 5, and Gemini 3.1 Pro achieving only 31.25%, 26.56%, and 22.85% Pass@1, revealing that models still rely primarily on symbolic reasoning rather than visual diagrams.