Keep on to blur preview images; turn off to show them clearly
RT @nypost: Doc who fled China hides in US over fears communists want revenge on COVID lab theories https://t.co/JFxXI0Rsfs https://t.co/G4…
“The woman scooping Silicon Valley” — BBC・hacker turned builder, blogger & consultant・ex: Threads, Instagram, etc
SplatPainter: Interactive Authoring of 3D Gaussians from 2D Edits via Test-Time Training Contributions: 1. A Novel Feedforward Editing Framework: We introduce SplatPainter, a state-aware feedforward model for the interactive and continuous editing of 3DGS. By leveraging TTT, our method iteratively incorporates user edits from 2D views, directly updating the attributes of a compact and feature-rich Gaussian representation without destructive regeneration. 2. Versatile and Intuitive Applications: We evaluate our framework on two novel and practical applications: high-fidelity local detail refinement and consistent global relighting. The unified approach provides users with precise control over both fine-grained texture details and broad appearance changes from 2D edits. 3. State-of-the-Art Performance at Interactive Speeds: Through extensive experiments, we show that SplatPainter significantly outperforms existing methods in both quality and speed. Our approach achieves superior visual results while operating at interactive rates, making it a truly practical solution for unlocking fluid, real-time creative workflows in 3D content authoring.
Paper: https://t.co/fD0PVFVhpM Project: https://t.co/9lt4aJl1va
训练好一个机器学习模型往往只是第一步，最头疼的其实是向老板或业务方解释：为什么模型给出了这个预测结果？面对“黑盒”模型，常常哑口无言。 最近在 GitHub 挖掘到一个名为 Shapash 的 Python 库，专门致力于让机器学习模型变得“说人话”，把复杂的算法逻辑翻译成可视化的语言。 能直接生成一个交互式的 Web 仪表盘，不仅清晰展示特征对模型的影响权重，还支持从全局到局部的深度探索。最关键的是，所有图表都配有清晰的业务标签，非技术人员也能一眼看懂。 GitHub：https://t.co/WiC2uPKqEK 兼容性颇佳，支持 Catboost、Xgboost、LightGBM 以及 Sklearn 等主流模型框架。 除了可视化，还能一键生成独立的 HTML 审计报告，甚至提供了用于生产环境部署的轻量级预测器，通过 pip 即可快速安装使用。 如果你需要经常向非技术团队汇报模型成果，或者需要对模型进行合规审计，这个项目能帮我们省下大量解释的时间。
💡 挖掘开源的价值 🧑🏻💻 坚持分享 GitHub 上高质量、有趣、实用的教程、AI工具、前沿 AI 技术 🧐 A list cool, interesting projects of GitHub. ✏️ 公众号：GitHubDaily
Success usually comes from 1000 minuscule victories, not 1 big break. Find a way to get a single tiny victory every day: - 1 hour of training - 1 hour of coding - 1 hour of reading - 1 hour of therapy Overnight success takes years to happen. Take your time and keep grinding.
Product & Devs Growth @Cyfrin | Ex @Alchemy | Created @cyfrinupdraft and @AlchemyLearn | Robotics | Making web3 mainstream
Shocking but true: LLMs learn to reason better when you give them the wrong traces. https://t.co/Y8o0Y9jEHb
Professor of computer science at UW and author of '2040' and 'The Master Algorithm'. Into machine learning, AI, and anything that makes me curious.
Shocking but true: LLMs learn to reason better when you give them the wrong traces. https://t.co/Y8o0Y9jEHb
Professor of computer science at UW and author of '2040' and 'The Master Algorithm'. Into machine learning, AI, and anything that makes me curious.