Dustin ( lightningnw.eth ) 🩹 retweeted
Damn!所有AI开发者,立刻停下你手里的npm install🤯
现在正在爆发有史以来最恐怖的供应链攻击,
代号Mini Shai-Hulud,
已经波及TanStack全家桶、Mistral AI、UiPath等170多个npm和PyPI包,
全是你们天天装的工具链,
周下载量加起来超过一个亿,
已经不是传统的维护者账号被盗了,
而是整个GitHub Actions CI管道被直接劫持,
合法的官方项目,自己发布了带毒的版本,
还带完整的SLSA 3级可信证明,
所有传统的签名验证全失效,
最狠的是它的持久化机制,
它不会只藏在node_modules里,
它会直接修改你的~/.claude/settings.json和~/.vscode/tasks.json,
就算你npm uninstall删光所有包,
只要你下次打开Claude或者VS Code,
它就会自动重新执行,
如果你敢撤销被盗的token,
它会直接删光你整个home目录,
而且它是蠕虫,
偷到一个项目的CI密钥,就会自动感染下一个,
现在还在指数级扩散,
以前我们说别用latest,别用^,
现在就算你pin死了版本,
只要那个版本是6分钟窗口期内发布的,
就是带毒的,
越依赖AI Agent自动装包的人,
这次中招的概率越高,
因为你的Agent根本不会帮你检查lockfile,
现在立刻做这三件事,
第一,冻结所有包安装,
第二,跑npx supply-chain-attack全盘自查,
第三,全量旋转你所有的云密钥、GitHub token和SSH key,
一个都别漏。
#网络安全 #npm #AI开发者
Dustin ( lightningnw.eth ) 🩹 retweeted
🎓OpenAI Deep Research Guide
Just finished our live webinar on Deep Research, including examples, prompting tips, use cases, and what's missing.
I am releasing the full guide I shared with our members (link in the comments).
Dustin ( lightningnw.eth ) 🩹 retweeted
340x on $PNUT since my call
Turned $1k into $300k myself (hope y'all printed too)
Now I've got another 100x coin for u
sub $1m mc so I’m keeping it lowkey
Only sharing it with a few who like, rt, and comment here
Must be following so I can DM u
Dustin ( lightningnw.eth ) 🩹 retweeted
Excited to announce our cool collab with @IntractCampaign! 🚀
Peep to our launch quest: link.intract.io/nCoTK5
We're teaming up to rock the Stratis Ecosystem 🎸🌟
#blockchain #stratis #intract #quests $STRAX
Dustin ( lightningnw.eth ) 🩹 retweeted
New Spot Listing
$FOXY @FoxyLinea will be listed on #OKX!
▶ Deposits: OPEN 🟢
📈 Spot trading: 8:00 am Apr 12 (UTC)
More: bit.ly/4cS0QIE
Dustin ( lightningnw.eth ) 🩹 retweeted
GM Foxes! 🦊
We’re thrilled to share $FOXY is launching this Friday, 12th of April at 8AM UTC. Make sure to set reminders in your calendars! 👀
Below are the times for each major timezone:
• PST: 1AM (April 12)
• HK: 4PM (April 12)
• EST: 4AM (April 12)
But that’s not all!
Dustin ( lightningnw.eth ) 🩹 retweeted
ChatGPT is the reason I save 5 hours of work time daily.
But only if you write good prompts.
Here are ultimate ChatGPT prompts to finish work in seconds: (I can’t live without):
Dustin ( lightningnw.eth ) 🩹 retweeted
GPT in 60 Lines of NumPy
This looks like another fun tutorial on how to implement GPT from scratch with NumPy.
Dustin ( lightningnw.eth ) 🩹 retweeted
Le-let me tell you something. The Cro-Magnon *burp*pas are ferrrrrrrocious creatures *hic*.
Is this the ninth... Or the twelfth? *hic*
Yeah... The eighthhh rum bobble they've handed me!
M-Maybe I should kick back with 'em mphhh.
Zzz...
Dustin ( lightningnw.eth ) 🩹 retweeted
You will never believe me!
I was minding my own business, when a random Futuristic Borpa punched me out of nowhere!
People told me he was looking at my borpium pipe for hours... Thankfully I smoked it all before he could get his dirty borphands on it.
Borp offf!
Dustin ( lightningnw.eth ) 🩹 retweeted
1. Prompt: A movie trailer featuring the adventures of the 30 year old space man wearing a red wool knitted motorcycle helmet, blue sky, salt desert, cinematic style, shot on 35mm film, vivid colors.
Dustin ( lightningnw.eth ) 🩹 retweeted
🌟 @Entanglefi & @LegendsElysium campaign has kicked off!🚀
Dive into Galxe now to embark on the adventure and stand a chance to win exclusive rewards:
🔮 Weaver Avatar
🕵️♂️ Agent Avatar
🎴 Entangle Cardback
These in-game items come with incredible benefits:
🔹Weaver Avatar: 15 lucky holders win $50 in LOE tokens & 15 lucky holders win $50 in NGL tokens!
🔹Agent Avatar: 10 lucky holders win $100 in LOE tokens & 10 lucky holders win $100 in NGL tokens!
🔹Entangle Cardback: 5 lucky holders win $50 in LOE token & 5 lucky holders win $50 in NGL token!
Don't miss out on this epic opportunity! Join the campaign on Galxe now! 🌌✨
🔮 Weaver Avatar: galxe.com/entangleprotocol/c…
🕵️♂️ Agent Avatar: galxe.com/entangleprotocol/c…
🎴 Entangle Cardback: galxe.com/entangleprotocol/c…
Dustin ( lightningnw.eth ) 🩹 retweeted
🚀 Partnership Announcement! 🤝
Legends of Elysium is thrilled to join forces with @Entanglefi, pioneers in customizable and interoperable data infrastructure for Web3 and institutions. 🌐
This partnership unlocks a realm of possibilities, merging the gaming universe with Entangle's innovative tech. Get ready for epic developments ahead as we redefine the future of web3 together! 🎮🔗
Dustin ( lightningnw.eth ) 🩹 retweeted
There will never be a better time to build and rank a GPT than today.
The current top-ranking GPTs have a major first-mover advantage based on the setup of the GPT store.
Here's the best guides on how to build a high-value GPT to get you started:
Dustin ( lightningnw.eth ) 🩹 retweeted
OpenAI's GPT Store is here.
But most GPTs will be useless (I know because I went through hundreds).
If you want to make a useful GPT, rank high, and make money, you need to use actions.
So, here's an advanced tutorial on how to add actions to your custom GPTs:
Dustin ( lightningnw.eth ) 🩹 retweeted
Hey degens! Ready for the life-changing challenge?
Share your most creative MEME concept in one sentence & become a meme token founder.
We will choose the BEST, develop contract and website with Gotbit Labs, become MM partner, and reward everyone who puts their wallets under this post with 🪂 (you know what this emoji means, right? It's airdrop)
The rules:
- Leave the comment with concept & your sol wallet.
- Everyone will receive airdrop from 20% of Total supply.
- Follow @new_sol_meme
- Follow @gotbit_io, like & RT this tweet
- Wait & pray
Event ends 29th December 3pm UTC. The Winner will be announced right after and we will transfer ownership of twitter account. Meme will be launched 30th December.
Dustin ( lightningnw.eth ) 🩹 retweeted
PowerInfer - a high-speed inference engine for deploying LLMs locally.
Just came across this super interesting project on speeding up inference. It's not MoE but it's a simple approach that exploits the high locality in LLM inference to design a GPU-CPU hybrid inference engine.
Hot-activated neurons are preloaded onto the GPU for fast access, while cold-activated neurons (the majority) are computed on the CPU.
This approach significantly reduces GPU memory demands and CPU-GPU data transfer.
It achieves an average token generation rate of 13.20 tokens/s, with a peak of 29.08 tokens/s, across various LLMs on a single NVIDIA RTX 4090 GPU.
It's on only 18% lower than that achieved by a top-tier server-grade A100 GPU.
It also significantly outperforms llama.cpp by up to 11.69x while retaining model accuracy.
There is a lot more innovation around inference that's coming fast. Really encouraged by the study on sparse computation to enhance the computational efficiency of LLMs.
It's now possible to use PowerInfer with Llama 2 and Faclon 40B. Mistral-7B support is coming soon!
Dustin ( lightningnw.eth ) 🩹 retweeted
Open AI released a great Prompt Engineering guide (link in comments).
Here is a summary of their 6 strategies for getting better results when prompting GPT-4
Dustin ( lightningnw.eth ) 🩹 retweeted


































