Allows to red-team your AI systems through adversarial probing. It is simple, effective, and requires minimal setup.

A capability lockfile for local models. Catch silent tool-calling regressions in CI (deterministic, no LLM judge).

Short and lightweight list of LLMs you should catch before they disapear (not saying they will, but better safe than sorry)

Repository with code and data for the the Moltbook Observatory Archive, which is an incremental dataset of agent-only social network activity.

Data collection from Moltbook for research

Browser-based video chat with Gemma 4 running locally via llama.cpp

Simple package to use old school image features, and avoid them being forgotten.

A Python framework for N-agent collective learning using Tsetlin Machines and synthetic data sharing, supporting zero-shot sensor onboarding and LLM-driven feature extraction.

Give LLMs "mathematical eyes" for image analysis. An MCP (Model Context Protocol) server that wraps the imagefeatures library, exposing classic computer vision features as tools that Claude and oth…

Security-focused AI reasoning benchmark for evaluating cipher analysis, steganography detection, and adversarial robustness

Visual chat example using the medgemma model from Google

This repository provides a Google Colab notebook designed to easily evaluate Hugging Face language models (e.g., Google's Gemma 3, Mistral, Llama) on tasks from the NorEval benchmark.

This repository contains code and resources related to an in-depth analysis of OpenAI's HealthBench, a benchmark designed for evaluating Large Language Models in the healthcare sector.

Simple how to and exploration of MedGemma from Google and comparison with Gemma3 using the Kvasir dataset

This is the code example repository for PGR-103-2020. The published code is for educational purposes only. Reusing the code for other purposes than this needs to be approved by individual code auth…

Real time local video chat example with llama.cpp

NorMedQA is designed to evaluate the medical knowledge and reasoning capabilities of large language models (LLMs) in Norwegian context (Bokmål and Nynorsk). The benchmark consists of 1313 question-…

This repository showcases various ways to utilize and visualize features from Meta AI's new Web-ssl model series facebook/webssl-dino300m-full2b-224, Vision Transformers trained using self-supervis…

A local agent RAG implementation using Gemma3 via Ollama