又一个武林秘籍公开了，
接下来练成什么样子，就看各家自己的功夫了。
🚀 Introducing DeepInfer, our in-house inference framework, now fully open source!
Powering DeepSeek-V4.1-Flash at scale with high throughput and low latency. We’re sharing our production stack to help the community lower compute costs, reduce energy use, and make AI more affordable.
🔗 github.com/deepseek-ai/DeepI…
1/4
3
3
168
DeepSeek-V4.1-Flash-1017 在机器学习研究方面完胜 Astra。
子弹飞一飞，总是会有一些意外发生 ，🤡🤡🤡
蒸馏徒弟的比师傅还牛逼？？？😈😈😈
到底是几碗粉么？？？🤪🤪🤪
DeepSeek-V4.1-Flash-1017 mogs Astra when it comes to ML research.
Ex. take a benchmark, ask an agent to make a model of any variety that solves rather than solving it itself. Anthropic publicly admits sabotaging ML research. OpenAI too?
Big fat fish wins by default?
@teortaxesTex
85
完了，
不是我一个人明显感觉这周更快了，
奥特曼赔钱！！！
Something shady is happening with GPT-6.1 Sol usage limits inside OpenAI Codex since Friday:
Devs on Plus and $200 Pro are reporting a massive silent nerf:
- 5-hour Codex limits draining 4x–5x faster on the exact same workloads
- Sol Low/Medium burning to 0% in 30 mins (down from 2+ hours last week)
- Single merge conflicts eating 15%+ and idle resets waking up at 0%
Are you seeing Codex limits vanish faster this week?
4
187
这是有史以来最有价值的大模型工作产出，
它居然仅仅用了 9 个小时，就完全逆向反编译了 Linux 的源码，
还开放在一个网站上面🌚🌚🌚
Last night, a swarm of Opus 5.5 agents working for 9 hours and burning 254.2 million output tokens were able to fully reverse engineer and decompile the Linux source code
Full source archive available here kernel.org
Grab it before they take it down
1
4
320