sneakyfox

@mativusgf

1751 Following

1002 Followers

sneakyfox pfp

📩 Gmail AI Agent Gmail AI Agent is an AI-powered email management system that automates the processing of emails by analyzing their content using GPT models to determine context and intent. It generates replies and short summaries, efficiently managing similar messages via Telegram and supporting multiple languages. Users can archive emails, reply to them or ask for more information. The system also offers different types of summary reports, including morning, afternoon and evening reports that focus on important communications and prioritized emails, eliminating unnecessary notifications. https://github.com/olivierloverde/gmail-agent

0 reply

0 recast

0 reaction

sneakyfox pfp

🧠 Your second brain c AI. Khoj is a versatile AI application that works both on devices and in the cloud, communicating with various LLMs such as llama3 and gpt. The app provides access to information from the web and documents in various formats including images and pdf. It is available on a variety of platforms such as browsers, Obsidian, Emacs and WhatsApp, allowing users to create customized agents and automate routine tasks. Khoj offers advanced semantic search, making it easy to find the documents you need, as well as the ability to generate images and audio messages. https://github.com/khoj-ai/khoj

0 reply

0 recast

5 reactions

sneakyfox pfp

Qwen2.5-Coder Qwen2.5-Coder-32B-Instruct is a code generation model that outperforms GPT-4o on most benchmarks. Qwen2.5-Coder-32B-Instruct demonstrates outstanding results in code generation, and is also effective in codebase repair and programming tasks. The release includes models of different sizes, which expands the possibilities for developers with different resources and promotes research in this area. https://github.com/QwenLM/Qwen2.5-Coder Demo: https://huggingface.co/spaces/Qwen/Qwen2.5-Coder-demo

0 reply

0 recast

0 reaction

sneakyfox pfp

🚀exo exo is a project that allows users to create AI clusters by combining various devices, including iPhones, Android and Linux computers, without the need for expensive GPUs. exo supports AI models such as LLaMA and Mistral, and offers a ChatGPT-compatible API for using the models on your own hardware. Exo is installed from source with low memory requirements, but the memory footprint on all devices in the cluster must be considered for the model to work. The program runs in P2P mode, allowing each device to participate in the computation without a masterworker architecture. https://github.com/exo-explore/exo

1 reply

0 recast

2 reactions

sneakyfox pfp

Hi everyone! 👋 I've written a script in Python selenium for farming Gradient Network points, maybe someone will find it useful. You just need a pre-installed Chrome browser to run it. https://github.com/mativusgf/gradient-network-bot

0 reply

0 recast

6 reactions

sneakyfox pfp

Browser-Use Browser-Use is an open source web automation library compatible with any language AI model. It automatically finds interactive elements, manages browser tabs, and extracts XPath for cleanup functions, eliminating the need to manually work with DevTools. The library also handles visual information and offers the ability to create customized actions for browser interaction, including integrating data into a database. In addition, Browser-Use handles dynamic content efficiently and offers tools for performing long-term memory tasks. In case of errors, the LLM is capable of self-correction, which increases the reliability of the library. https://github.com/gregpr07/browser-use

0 reply

0 recast

2 reactions

sneakyfox pfp

Ollama has added support for Llama3.2-vision 🦙 ☝🏻 Llama3.2-vision is multimodal and capable of recognizing images and generating descriptions for them. Both 11B and 90B versions of the model are available, with the 90B requiring an H100 GPU with 80GB of video memory to fully load. To use the model, you must upgrade Ollama to version 0.4.0 and perform the installation with the command ollama run llama3.2-vision For Linux users, a restart of the service and containers is required. Images can be uploaded via the Open WebUI by adding files. https://ollama.com/blog/llama3.2-vision

0 reply

0 recast

0 reaction

sneakyfox pfp

👨🏼‍💻 Satsweep Sastsweep is an open source code vulnerability detection tool that analyzes key repository metrics. It uses semgrep to automatically detect potential vulnerabilities and generates reports in HTML format, making it easy to identify problems in code. Go version 1.23 or higher is required for installation, and the tool supports a variety of commands and filters. https://github.com/chebuya/sastsweep

0 reply

0 recast

3 reactions

sneakyfox pfp

Ollama contains unpatched critical vulnerabilities 😳 Researchers at Oligo Security have discovered six vulnerabilities in Ollama, a large language modeling solution, that could lead to malicious behavior, including data theft. ☝🏻 All vulnerabilities have a high threat level. For two of them, their patches do not yet exist. ❌ ☝🏻 Ollama developers advise limiting access to vulnerable endpoints. Researchers identified 9,831 unique Ollama installations on the Internet. Most of them are located in China, the US, Germany, South Korea, Taiwan, France, the UK, India, Singapore and Hong Kong. Meanwhile, one in four servers is recognized to be vulnerable to the vulnerabilities discovered. 😳 https://thehackernews.com/2024/11/critical-flaws-in-ollama-ai-framework.html?m=1

0 reply

0 recast

0 reaction

sneakyfox pfp

🛑 WARNING 🛑 ☝🏻 Nvidia reported critical vulnerabilities in drivers and software for RTX, Quadro, NVS, Tesla and GeForce 😳 Eight vulnerabilities have been discovered that could allow hackers to remotely control high-risk computers (CVE scores of 7 to 8). ☝🏻 Nvidia added that the vulnerabilities in the company's software affect both the Nvidia GPU display driver and the Nvidia VGPU software. This could allow hackers to run malicious code on users' systems, as well as copy data without warning or authorization. To improve security, users are advised to update drivers to the latest versions for Windows and Linux. https://wccftech.com/nvidia-gpu-users-must-update-driver-8-high-risk-vulnerabilities-found/amp/

0 reply

0 recast

0 reaction

sneakyfox pfp

🖥️Silex Silex is a free and open source tool for creating websites without programming. It supports both online and offline creation, as well as integration with JAMStack projects. Silex offers a drag-and-drop visual editor and an online CSS editor, as well as SEO tools and self-hosting support. https://github.com/silexlabs/Silex/

1 reply

0 recast

6 reactions

sneakyfox pfp

🚀Cline Cline is an AI-based CLI and editor assistant that can perform complex programming tasks in a step-by-step fashion, leveraging the power of Claude 3.5 Sonnet agents. It can create and edit files, view projects, run terminal commands, and interact through a browser, while ensuring security and allowing users to approve changes. Cline also analyzes code and fixes bugs. Cline also supports integration with APIs such as OpenRouter and OpenAI, and tracks token usage. https://github.com/cline/cline

0 reply

0 recast

1 reaction

sneakyfox pfp

Open Interpreter Open Interpreter is an application for local code execution in various languages. Using natural language commands, users can perform various tasks, such as creating and editing files, analyzing data, and controlling a browser, while having access to all computer resources. The main advantage of Open Interpreter is that it has no restrictions on time, file size and internet access, which makes it more convenient for users. https://github.com/OpenInterpreter/open-interpreter

0 reply

0 recast

1 reaction

sneakyfox pfp

☝🏻Anthropic is launching a new AI model Claude 3.5 Sonnet that controls a computer The AI is able to interact with applications by performing keystrokes and cursor movements, allowing it to perform office tasks. Read more: https://www.anthropic.com/news/3-5-models-and-computer-use

0 reply

0 recast

9 reactions

sneakyfox pfp

AI Tamago AI Tamago is a virtual pet created and controlled by AI natural language processing (LLM) models. It is designed to be similar to the popular virtual pet Tamagotchi. All ASCII animations in the game are generated using the GPT-3 model. https://github.com/ykhli/AI-tamago

0 reply

0 recast

17 reactions

sneakyfox pfp

👾 LiteLLM LiteLLM allows you to call all the APIs (100+) of LLM using the OpenAI format. Supported vendors include OpenAI, Azure, AWS, Google, Cohere, Huggingface, VertexAI, TogetherAI and many more. https://github.com/BerriAI/litellm https://docs.litellm.ai/docs/

0 reply

0 recast

13 reactions

sneakyfox pfp

🦙 Ollama App Ollama App is a custom Ollama client for Android that provides privacy and local operation without the need for server hosting. The app's interface is intuitive and does not require complex customization, however it is important to note that it is in the experimental stage and some features may not work properly. https://github.com/JHubi1/ollama-app

0 reply

0 recast

0 reaction

sneakyfox pfp

Eagle Eagle is a multimodal LLM specifically optimized for processing high-resolution visual data. https://github.com/NVlabs/EAGLE Demo: https://huggingface.co/spaces/NVEagle/Eagle-X5-13B-Chat

0 reply

0 recast

1 reaction

sneakyfox pfp

Screenpipe - Records your screen and then runs what you have done through Ollama. The program allows you to automatically collect data from your screen and audio input. Special attention is paid to data security, with an emphasis on local storage in SQLite. https://github.com/mediar-ai/screenpipe

0 reply

0 recast

0 reaction

sneakyfox pfp

👾 Invoke AI Invoke is a powerful visual media creation tool that utilizes AI technologies. The tool allows users to manage workflows through a node-based architecture and integrated workspace. Invoke includes tools for image enlargement and metadata manipulation. Installation documentation is available for Windows, macOS and Linux, including the use of Docker.🐋 https://github.com/invoke-ai/InvokeAI

0 reply

0 recast

0 reaction