nCPU

一台完整的计算机，其中每一层——算术、操作系统、编译器、显示——要么是训练好的神经网络，要么完全在GPU上运行。 模型不是 在 计算机上运行。模型 就是 计算机本身。 nCPU 是一个代码库，从五个方向追求同一论点：计算机可以由学习到的组件构建，而一旦整个执行栈是可微分的，程序就不再是你编写的东西，而变成了可以通过梯度下降搜索的对象。下面的每个子系统都基于独立的测量；它们共同覆盖从单个 ALU 操作到操作系统再到程序合成的整个栈。 五个子系统 1. 神经计算机 每个 ALU 操作——加法、减法、乘法、位运算、移位、除法——都是一个训练好的神经网络。神经操作系统 neurOS...

backdoor-learning-resources

Recursos de Backdoor Learning Este repositorio de Github resume una lista de recursos de Backdoor Learning. Para más detalles y los criterios de categorización, consulte nuestro estudio. Haremos todo lo posible por mantener continuamente este repositorio de Github de forma mensual. ¿Por qué...

Kaonashi

Kaonashi 项目：RootedCON 2019 此仓库包含以下几个目录： 为 hashcat 排序的 掩码 为 hashcat 的高级 规则 下载 Kaonishi 词表 的链接 我们在 RootedCON 2019 的演讲 I know your p4$$w0rd and if I don't, I will guess it... 中使用的 幻灯片 克隆仓库可能不是获取所需文件的必要条件. 为什么 当用户必须选择密码时，他倾向于用相同的方式构建密码，使用相同的个人信息，以及相同的复杂度概念。...

AntiVE-BehaviorWatch

AntiVE-BehaviorWatch Verificación de comportamiento humano impulsada por IA integrada para investigación en ciberseguridad. ¿Qué es AntiVE-BehaviorWatch? AntiVE-BehaviorWatch es un proyecto de investigación en ciberseguridad basado en Windows que explora un nuevo enfoque para el control de...

recipe-blog-encoding

recipe-blog-encoding !WARNING This project is entirely vibe-coded likely partially plagiarized from this repository and the author is a dumb-dumb who just thought the idea was funny Use SEO-gaming recipe preambles as a vehicle for encoding secret messages the prompt can really be anything though,...

robustness

Paquete robustness Instalación mediante pip: pip install robustness Lee la documentación: https://robustness.readthedocs.io/en/latest/index.html robustness es un paquete que creamos estudiantes del MadryLab para que entrenar, evaluar y explorar redes neuronales sea flexible y sencillo. Lo usamos ...

deep-pwning

Deep-pwning is a lightweight framework for experimenting with machine learning models with the goal of evaluating their robustness against a motivated adversary. Note that deep-pwning in its current state is no where close to maturity or completion. It is meant to be experimented with, expanded...

CVE-2022-32898

CVE-2022-32898: ANEProgramCreate 多个内核内存损坏 2022年11月23日 • Mohamed GHANNAM @simo36 作者评论 我要分享另外两个 iOS 内核漏洞，它们可以从默认应用沙箱中触达，且不需要打开 UserClient： 另外有 16 个我向 Apple 报告的内核 bug 已在 iOS 16/16.1 中修复。 下个月我将就如何串联一些漏洞实现内核 r/w 在 POC2022 上发表演讲， 会议结束后，适用于 iOS 15 的内核利用将与其他一些高影响漏洞一起发布。 到目前为止，我最喜欢的 IDA 8.0 功能是：人工 Obj-C...

eyeballer

Eyeballer Give those screenshots of yours a quick eyeballing. Eyeballer is meant for large-scope network penetration tests where you need to find "interesting" targets from a huge set of web-based hosts. Go ahead and use your favorite screenshotting tool like normal EyeWitness or GoWitness and th...

Insider-Threat

Modelos de Grafos de Amenazas Internas R5.2 y R6.2 Este repositorio contiene preparación de datos, extracción de características y modelos basados en grafos GCN-NN, GCN-BiLSTM y GCN-BiLSTM+Attention para señales implícitas+explícitas para la detección de amenazas internas utilizando el Conjunto d...

certified-unlearning-neural-networks-icml-2025

Certified Unlearning for Neural Networks This repository contains the implementation for the paper "Certified Unlearning for Neural Networks" ICML 2025. Methods PABI Privacy Amplification by Iteration : Gradient Clipping in the paper DP-SGD : With group privacy as described in the paper Contracti...

Intel Neural Compressor <2.5.0 - SQL Injection

Improper input validation in some IntelR Neural Compressor software before version 2.5.0 may allow an unauthenticated user to potentially enable escalation of privilege via remote access. id: CVE-2024-22476 info: name: Intel Neural Compressor 2.5.0 - SQL Injection author: ritikchaddha,daffainfo...

ane

Asahi Neural Engine Controlador de Linux con ingeniería inversa para el Apple Neural Engine ANE. Todo lo relacionado con Linux aquí. ane/: Módulo del kernel ane.ko. Debería integrarse en el árbol pronto. docs/: Documentación. En progreso. Por favor, no mirar. libane/: Librería de espacio de...

weightBufs

WeightBufs: WeightBufs es un exploit de lectura/escritura del kernel para todos los dispositivos Apple con soporte de Neural Engine. Errores y exploit por @simo36; puedes leer mis diapositivas de la presentación en POC para más detalles sobre las vulnerabilidades y las técnicas de explotación. El...

craw

CRAW: Codec Robust Audio Watermarking Source code for CRAW , a robust audio watermarking method built on STFT-magnitude-domain embedding. An encoder embeds a binary message into an audio clip's magnitude spectrogram phase is preserved from the clean signal; a decoder recovers the message under a...

challenge_repo

🎲 熵挑战 🇬🇧 英文阅读 | 🇹🇷 Türkçe Oku 引言 我们开发了一款下一代随机数生成器，通过专有的神经网络架构 处理射频噪声 ，并经过SHA-3后处理。我们声称该源提供真正随机性 质量，与理想噪声源不可区分。 我们公开输出结果和验证报告，但不公开源代码 。挑战很简单：证明我们是错的。 数据 样本文件: data/randomsample.bin 17 MB 质量: 通过NIST STS SP 800-22 和 NIST SP 800-90B 验证。 熵: 0.999 比特/比特。 详细报告见 NISTREPORT.md。 🏆 挑战 我们向任何能成功“破解”此生成器的人提供赏金。...

CVE-2026-20906

Protection mechanism failure for some IntelR Neural Compressor software before version v3.6 within Ring 3: User Applications may allow an escalation of privilege. Unprivileged software adversary with a privileged user combined with a low complexity attack may enable escalation of privilege. This...

CVE-2026-20913

Improper input validation for some IntelR Neural Compressor software before version v3.7 within Ring 3: User Applications may allow an escalation of privilege. Unprivileged software adversary with an authenticated user combined with a low complexity attack may enable escalation of privilege. This...

CVE-2026-20913

Improper input validation for some IntelR Neural Compressor software before version v3.7 within Ring 3: User Applications may allow an escalation of privilege. Unprivileged software adversary with an authenticated user combined with a low complexity attack may enable escalation of privilege. This...

CVE-2026-20913

Improper input validation for some IntelR Neural Compressor software before version v3.7 within Ring 3: User Applications may allow an escalation of privilege. Unprivileged software adversary with an authenticated user combined with a low complexity attack may enable escalation of privilege. This...