开启时会模糊预览图，关闭后正常显示
What's interesting is that there was a divergence. In 2022, closed labs were doing PPO, and open models… didn't exist. In 2023, we mostly did SFT (LoRA or not) and then got DPO species copium. GRPO in 2024 was… unexpected. And we don't know what closed labs use now.
btw they also were surprised by how GRPO has fared
RT @indie_maker_fox: 在Reddit上发现的一篇好文章 如何使用Payload CMS和Vercel AI SDK来构建AI应用，看完很受启发，AI应用有很多通用的事情需要处理，而Payload CMS提供的一些功能刚好能满足这些需求，前途无量 Buil…
🔥 The best AI SaaS boilerplate - https://t.co/VyNtTs0jSX 🚀 The best directory boilerplate with AI - https://t.co/wEvJ1Dd8aR 🎉 https://t.co/bh1RxeERuY & https://t.co/zubXJCoY92 & https://t.co/tfQf8T7gGF
#Steam 客户端基本完成架构过渡，从明年 1 月 1 日起全面停止更新 32 位版。如果用户运行的是 64 位 Windows 10/11 那没有任何问题，Steam 也会自己更新到 x64 架构，但如果用户运行的是 32 位 Windows 10 那后续无法接收 Steam 更新，随着时间推移出现 BUG 也没法修复：https://t.co/F3GlmwRaYm
独立科技网站 - 蓝点网 / 感谢关注 订阅频道：https://t.co/xzeoUEoPcU 联系方式：https://t.co/LJK1g3biPp
![MatSpray: Fusing 2D Material World Knowledge on 3D Geometry
Contributions:
• World Material Fusion: A plug-and-play pipeline that, to our knowledge, is the first to fuse swappable diffusion-based 2D PBR priors ("world material knowledge") with 3D Gaussian material optimization. It uses Gaussian ray tracing and PBR consistent supervision to obtain relightable assets.
• Neural Merger: A softmax neural merger that aggregates per-Gaussian, multi-view material estimates, suppresses baked-in lighting, and enforces cross-view consistency while stabilizing joint environment map optimization.
• Faster Reconstruction: A simple projection and optimization scheme that reconstructs high-quality relightable 3D materials with 3.5× less per-scene optimization time than IRGS [9].](/_next/image?url=https%3A%2F%2Fpbs.twimg.com%2Famplify_video_thumb%2F2003360301333512198%2Fimg%2FQkqLL_37pcs3nDBx.jpg&w=3840&q=75 "MatSpray: Fusing 2D Material World Knowledge on 3D Geometry
Contributions:
• World Material Fusion: A plug-and-play pipeline that, to our knowledge, is the first to fuse swappable diffusion-based 2D PBR priors (\"world material knowledge\") with 3D Gaussian material optimization. It uses Gaussian ray tracing and PBR consistent supervision to obtain relightable assets.
• Neural Merger: A softmax neural merger that aggregates per-Gaussian, multi-view material estimates, suppresses baked-in lighting, and enforces cross-view consistency while stabilizing joint environment map optimization.
• Faster Reconstruction: A simple projection and optimization scheme that reconstructs high-quality relightable 3D materials with 3.5× less per-scene optimization time than IRGS [9].")
MatSpray: Fusing 2D Material World Knowledge on 3D Geometry Contributions: • World Material Fusion: A plug-and-play pipeline that, to our knowledge, is the first to fuse swappable diffusion-based 2D PBR priors ("world material knowledge") with 3D Gaussian material optimization. It uses Gaussian ray tracing and PBR consistent supervision to obtain relightable assets. • Neural Merger: A softmax neural merger that aggregates per-Gaussian, multi-view material estimates, suppresses baked-in lighting, and enforces cross-view consistency while stabilizing joint environment map optimization. • Faster Reconstruction: A simple projection and optimization scheme that reconstructs high-quality relightable 3D materials with 3.5× less per-scene optimization time than IRGS [9].
Paper: https://t.co/ZIQjEjjtUM Project: https://t.co/SMjrucR9r1
Reminder that 220 cm tall Chinese gfs are not supposed to exist under a naive bell curve model. That's a 2*10^-18 event. Except clearly at least 2 exist. You can apply the same reasoning generally, eg to "implausible" IQ values of Terry Tao or JVN.
We're in a race. It's not USA vs China but humans and AGIs vs ape power centralization. @deepseek_ai stan #1, 2023–Deep Time «C’est la guerre.» ®1
Was out canoeing today and thought of an AI-based sharing component to replace our existing sharing feature for flows. Kinda like it.
Marketer, self-taught developer, and founder of @Bento and https://t.co/lcsIohchEv. Designing a quiet family life in 福岡, Japan. DMs open if you need email help 🌿
