Real-time reward debugging and hacking detection for reinforcement learning

Ultramodern ML cookicutter

🧠 NeuroForge is an intuitive drag-and-drop tool for building and training neural networks, featuring data preprocessing, interactive visualizations, and automated model architecture design. Built with PyTorch and Streamlit, it simplifies the deep learning workflow from data preparation to model deployment with GPU acceleration support.

🥧 [Under Construction] Development toolkit & template: Intelligent lazy loading (96% faster startup), Docker deployment guide available. Advanced AI context engineering, production project frameworks.

MLGuard – Lightweight ML experiment manager and leakage detection toolkit

ScrollScribe is a Python CLI toolkit that grab docs or index website pages and converts them into clean local Markdown using browser automation and LLM filtering, perfect for building RAG datasets.

RetroML is a nostalgic, configuration-driven AutoML pipeline that brings a ‘90s flair to modern machine learning workflows. It allows users to run ML classification tasks via JSON configs, with optional support for interactive UIs and demos (e.g., customer churn prediction).

A modular deep learning framework for training and evaluating image classification models on datasets like CIFAR-10 and MNIST. Supports configurable CNN architectures, automated training, and performance visualization using Python and TensorFlow.

Application providing an efficient and user-friendly GUI for captioning PNG datasets in bulk with comma-separated TXT files.

Automated data leakage detection and ML validation strategy advisor for reliable machine learning model evaluation.

Готовая среда Claude Code для ML/AI: hooks, skills, subagents, persistent memory, auto-fix loop, secret protection. Установка одной командой.

Script for dumping frame images from video

🔍 The fairness monitoring library for AI - audit, measure, and mitigate demographic bias in your ML models with one line of code

A binary-level cognition engine for MindsEye. Decode, label, map, and traverse binary as time-patterns. Builds signatures, provenance, and time-aware meters for advanced ML and agentic systems.

🔍 Detect reward hacking in RL training with RewardScope. Track reward components and visualize agent behavior to enhance learning efficiency.