huang

EchoPod 听 podcast 学语言: SLOC panel that Github forget:

[huang @huang4fun]

@Joi2James https://t.co/9jFrxh1OJY

[huang @huang4fun]

@danellisona https://t.co/9jFrxh1OJY

[huang @huang4fun]

@real_kai42 kimi code usage 只显示重置剩余时长，不知道具体几点才重置，提了个 PR 支持显示剩余时长和 local reset time，这样可以一看就知道我可以在几点开始继续 vibe
不知道算不算小众需求🤣
https://t.co/ctm8hVemqB

[huang @huang4fun]

解锁新体验：攀岩会磨损指纹，导致 mac 指纹不识别🤣

[huang @huang4fun]

OctoCounts Chrome Store 已经有 5k+ weekly users 了，一个用来用于快速获取 github repo SLOC的小工具，很开心🥳

https://t.co/9jFrxh1OJY

[huang @huang4fun]

时间线上全是 KIMI3， 有人灰度到 DeepSeekV4 Pro 正式版，现在马上 qwen3.8 也要发了，都是比 Fable 5 差一些的 SOTA，开源和闭源差距会越来越小。只是不知道大模型时代开闭源之争会不会出现类似 Window vs Linux 的结局

[Qwen @Alibaba_Qwen]

Qwen3.8 is launching and going open-weight soon!🌐

With a massive 2.4T parameters, this model is continuously evolving. We believe it’s one of the most powerful model available today, compatible to leading frontier AI models , second only to Fable 5.

You don't have to wait to test it. Just now, the Qwen3.8-Max-Preview made its debut on Alibaba’s Token Plan, Qoder, and QoderWork. Be among the very first to try it out.

Can't wait to hear what you build. Stay tuned! 🚀 

Token Plan
international：https://t.co/YRvcGdB9Bv
China：https://t.co/PKMUNwUuRp

[huang @huang4fun]

我看不懂但我大受震撼

[很大很大的橙子 @0xVeryBigOrange]

Anthropic CEO Dario Amodei 向美国国会提交报告：
“经过调查，我们发现 Kimi K3 可能蒸馏了一个 Anthropic 尚未开发完成的模型。”

国会议员追问：“你们的模型还没做出来，它怎么蒸馏的？”

Amodei：“这正是我们召开听证会的原因。”

AI 安全领域迎来新课题：Future Model Leakage

[huang @huang4fun]

@xsser_w 127.0.0.1

[huang @huang4fun]

NVIDIA CMP 170HX 本来是老黄刀法将 A100 瑕疵品将 80G 显存固件限制到 10G，有人通过缓冲区溢出破解了限制，获取了比 A100 稍差但完全可以用的状态，矿卡要涨价了

https://t.co/dHU6vSH3M3

[huang @huang4fun]

DeepSeek 发布 DSpark，一种预测解码(Speculative Decoding)模型，可以实现60–85% 的 token 生成加速。
反观GPT5.6 也和fable 5 一样受限发布，不知道是否会出现开源模型因为便宜量大用户数据多反而会慢慢反超闭源模型

[Daniel Han @danielhanchen]

DeepSeek just released DSpark for V4 Flash & Pro, a new speculative decoding method boosting throughput by 51% to 400%!

DS also showed DSpark works well for other models like Gemma & Qwen

Github: https://t.co/EGVYpc1kcK
Paper: https://t.co/TaBMRVlaW9
HF: https://t.co/289jVU2pxh

[huang @huang4fun]

本来只是想简单估算一个开源 nes 实现代码量，github 自己不提供代码行数数据，又不想 clone 下来计算，vibe coding 了一个浏览器插件 OctoCounts 用于快速获取代码行数，发布一周有了 2497 个用户，vibe 是值得的
https://t.co/9jFrxh1OJY

[huang @huang4fun]

Update OctoCounts extension with skeleton loading, now it looks much better.

try it here: https://t.co/9jFrxh1OJY the SLOC panel that Github forget

[huang @huang4fun]

上面的数据是在 Apple M5 32G macOS 里跑出的数据，repo 里有 benchmark 的 script 可以一键生成 report，有兴趣的欢迎尝试， share 你的 benchmark 数据
https://t.co/Q2y0n1sNhL

[huang @huang4fun]

一直默认 mlx 支持的 qwen-asr 会比纯 cpu 实现的快，周末重新让 AI 优化了一波之前的 qwen-asr rust 实现，benchmark 里增加了mlx-audio python 版和另一个用 second-state 带 mlx backend 加速的实现
然后发现纯 CPU 的居然是最快的, 以为是个青铜结果居然是个王者，抽空继续深挖下原因

[huang @huang4fun]

My first browser extension: OctoCounts – the SLOC (Source Lines of Code) panel GitHub forgot, it can show the full SLOC info for public repo.
Check more detail and download the extension here: https://t.co/9jFrxh1OJY

[huang @huang4fun]

小米百万亿 Token 激励计划回复速度好快，黄金周休假得狠狠用起来

[huang @huang4fun]

DeepSeekV4 也发布了，A 家怕不是用 DeepSeek+GPT 来修复的吧🤣

[huang @huang4fun]

前脚 GPT5.5 发布，后脚 Anthropic 就发 blog 承认降智并且还修复了，演都不演了
https://t.co/0XP27N9fuQ

[huang @huang4fun]

DeepSeek-V4 top level 的性能，但token最高便宜将近 70x，DeepSeek 杀疯了

[DeepSeek @deepseek_ai]

🚀 DeepSeek-V4 Preview is officially live & open-sourced! Welcome to the era of cost-effective 1M context length.

🔹 DeepSeek-V4-Pro: 1.6T total / 49B active params. Performance rivaling the world's top closed-source models.
🔹 DeepSeek-V4-Flash: 284B total / 13B active params. Your fast, efficient, and economical choice.

Try it now at https://t.co/GCdiMzk1Dl via Expert Mode / Instant Mode. API is updated & available today!

📄 Tech Report: https://t.co/drlDrxkYtp
🤗 Open Weights: https://t.co/T13Y8i7SDM

1/n

[huang @huang4fun]

@Prince_Canuma You might also be interested in this Ninja input method:
https://t.co/p5h3XHH8vM

[huang @huang4fun]

Typing… Naruto style! 🥷✨
Real-time hand gesture T9 keyboard—runs entirely in your browser.
Try it now and see if you can keep up! 👀
🔗 try it online:https://t.co/MQmSRZl0iK
#WebDev #GestureTyping #Naruto #VibeCoding