👍1😁1
SiphonDNS: 通过 DNS 进行隐蔽数据泄露 https://ttp.report/evasion/2025/02/03/siphondns-covert-dns-exfiltration.html
TTP Report
SiphonDNS: covert data exfiltration via DNS
When you end up in a more strictly controlled environment, HTTP and DNS are likely the only protocols allowed to go outside. Furthermore, you can bet on both being proxied and highly monitored. This time, I’ll focus on some opportunities to hide traffic within…
❤2
KTransformers团队宣布支持DeepSeek-R1和V3模型,支持在单GPU(24GBVRAM)或多GPU和382GBDRAM的环境下运行。通过优化,实现了高达3~28倍的速度提升。提供了本地运行的示例,展示了如何在桌面环境中实现类似GPT-4的性能。
测试环境:
CPU:IntelXeonGold6454S(32核心/插槽,2插槽,2NUMA节点)
GPU:4090D(24GBVRAM)
内存:1TBDDR5-4800服务器DRAM
性能结果:
Prefill速度:最高可达97.32tokens/s
Decode速度:最高可达13.69tokens/s
相比llama.cpp,Prefill速度提升高达9.44倍,Decode速度提升3.03倍。
#AI
https://kvcache-ai.github.io/ktransformers/en/DeepseekR1_V3_tutorial.html
测试环境:
CPU:IntelXeonGold6454S(32核心/插槽,2插槽,2NUMA节点)
GPU:4090D(24GBVRAM)
内存:1TBDDR5-4800服务器DRAM
性能结果:
Prefill速度:最高可达97.32tokens/s
Decode速度:最高可达13.69tokens/s
相比llama.cpp,Prefill速度提升高达9.44倍,Decode速度提升3.03倍。
#AI
https://kvcache-ai.github.io/ktransformers/en/DeepseekR1_V3_tutorial.html
❤4👍1
反 Rootkit 技术 - 第 I 部分:取消 Map-Rootkit
https://eversinc33.com/posts/anti-anti-rootkit-part-i.html
反 Rootkit 技术 - 第 II 部分:破坏驱动程序和隐藏线程
https://eversinc33.com/posts/anti-anti-rootkit-part-ii.html
反 Rootkit 技术 - 第 III 部分:劫持指针
https://eversinc33.com/posts/anti-anti-rootkit-part-iii.html
https://eversinc33.com/posts/anti-anti-rootkit-part-i.html
反 Rootkit 技术 - 第 II 部分:破坏驱动程序和隐藏线程
https://eversinc33.com/posts/anti-anti-rootkit-part-ii.html
反 Rootkit 技术 - 第 III 部分:劫持指针
https://eversinc33.com/posts/anti-anti-rootkit-part-iii.html
Eversinc33
(Anti-)Anti-Rootkit Techniques - Part II: Stomped Drivers and Hidden Threads
Detectin driver 'stomping' and hiding system threads by manipulating the PspCidTable.
浙江大学出的这个开源的书籍「大模型基础」值得一看,行文风格挺不错的,易读、严谨、有深度的大模型教材。
https://github.com/ZJU-LLMs/Foundations-of-LLMs
https://github.com/ZJU-LLMs/Foundations-of-LLMs
GitHub
GitHub - ZJU-LLMs/Foundations-of-LLMs: A book for Learning the Foundations of LLMs
A book for Learning the Foundations of LLMs. Contribute to ZJU-LLMs/Foundations-of-LLMs development by creating an account on GitHub.
文章通过实验展示了在开源LLM中嵌入后门的可能性,并强调了嵌入风险的隐蔽性和检测的困难性。作者呼吁在使用LLM时保持警惕,无论其是否开源,并期待AI研究者开发出有效的检测和缓解方法。
#AI
https://blog.sshh.io/p/how-to-backdoor-large-language-models
https://github.com/sshh12/llm_backdoor?tab=readme-ov-file
#AI
https://blog.sshh.io/p/how-to-backdoor-large-language-models
https://github.com/sshh12/llm_backdoor?tab=readme-ov-file
blog.sshh.io
How to Backdoor Large Language Models
Making "BadSeek", a sneaky open-source coding model.
🤯1
Microsoft 激活脚本 (MAS)——自行分析是否有问题,慎重使用。
https://github.com/massgravel/Microsoft-Activation-Scripts
https://github.com/massgravel/Microsoft-Activation-Scripts
GitHub
GitHub - massgravel/Microsoft-Activation-Scripts: Open-source Windows and Office activator featuring HWID, Ohook, TSforge, and…
Open-source Windows and Office activator featuring HWID, Ohook, TSforge, and Online KMS activation methods, along with advanced troubleshooting. - massgravel/Microsoft-Activation-Scripts
Linux系统中有软链接和硬链接的概念,这篇文章详细解释他们之间的差别。
https://bhoot.dev/2024/on-linux-file-and-links/
https://bhoot.dev/2024/on-linux-file-and-links/
Linux硬链接:在进行文件备份时,硬链接能在不额外占用大量存储空间的情况下创建文件副本。因为多个硬链接指向同一个数据块,对原文件进行修改,备份的硬链接文件也能实时体现这些变化,保证数据一致性,同时删除其中一个硬链接不会影响其他链接和数据块,提高数据安全性。例如,在重要配置文件备份场景中,系统管理员可以为关键的配置文件创建多个硬链接,分布在不同目录下。当系统出现故障需要恢复配置时,只要数据块未被删除,任何一个硬链接都可用于恢复配置。
思考:如遇勒索,只要其中还有未被加密的数据块,是否可以恢复?
思考:如遇勒索,只要其中还有未被加密的数据块,是否可以恢复?
👍2
超强职场AI工具
文案创作
ChatGPT https://chatgpt.com/
Claude https://claude.ai/
Deepseek https://www.deepseek.com/
日常办公
Gamma https://gamma.app/zh-tw
RecCloud https://reccloud.com/tw/
Formulabot https://www.formulabot.com/
资料搜寻
Perplexity https://www.perplexity.ai/
Gemini https://gemini.google.com/
Bohrium https://www.dp.tech/en/product/bohrium
图片生成
Midjourney https://www.midjourney.com/
Recraft https://www.recraft.ai/
Stability.ai https://stability.ai/stable-image
文案创作
ChatGPT https://chatgpt.com/
Claude https://claude.ai/
Deepseek https://www.deepseek.com/
日常办公
Gamma https://gamma.app/zh-tw
RecCloud https://reccloud.com/tw/
Formulabot https://www.formulabot.com/
资料搜寻
Perplexity https://www.perplexity.ai/
Gemini https://gemini.google.com/
Bohrium https://www.dp.tech/en/product/bohrium
图片生成
Midjourney https://www.midjourney.com/
Recraft https://www.recraft.ai/
Stability.ai https://stability.ai/stable-image
ChatGPT
ChatGPT helps you get answers, find inspiration, and be more productive.
👍5❤1
一种高性能分布式文件系统,旨在应对 AI 训练和推理工作负载的挑战。
https://github.com/deepseek-ai/3FS
https://github.com/deepseek-ai/3FS
GitHub
GitHub - deepseek-ai/3FS: A high-performance distributed file system designed to address the challenges of AI training and inference…
A high-performance distributed file system designed to address the challenges of AI training and inference workloads. - deepseek-ai/3FS
每一轮大的技术风口并不完全是一哄而上的炒作,里面包含了一些真正的机会,值得关注和跟上。这也是为什么周刊每一次都对新技术倍加关注、积极评价的原因。
当然,赶上技术风口的前提,还是要有真才实学,能做出实打实的产品。否则,真遇到风口,你也无法脱颖而出,拿到技术炒作的红利。
当然,赶上技术风口的前提,还是要有真才实学,能做出实打实的产品。否则,真遇到风口,你也无法脱颖而出,拿到技术炒作的红利。
👍7
mshta.exe 从命令行执行原始脚本
https://github.com/SecurityAura/DE-TH-Aura/blob/main/100DaysOfKQL/Day%2080%20-%20mshta.exe%20Executing%20Raw%20Script%20From%20Command%20Line.md
https://github.com/SecurityAura/DE-TH-Aura/blob/main/100DaysOfKQL/Day%2080%20-%20mshta.exe%20Executing%20Raw%20Script%20From%20Command%20Line.md
GitHub
DE-TH-Aura/100DaysOfKQL/Day 80 - mshta.exe Executing Raw Script From Command Line.md at main · SecurityAura/DE-TH-Aura
Repository where I hold random detection and threat hunting queries that I come up with based on different sources of information (or even inspiration). - SecurityAura/DE-TH-Aura
👍1