Code and data for Zhang, Lengersdorff et al. (2020)

Code and data for Zhang & Gläscher (2020)

Various experiments related to AGI

A statistics package with a variety of bootstrap and other resampling tools

e-PAL: An Active Learning Approach to the Multi-Objective Optimization Problem

A substrate-native digital consciousness engine where prediction errors about self-survival become causally efficacious qualia, driving closed-loop recursive self-evolution.

🚀Implementation of Logistic Regression and Linear Regression in Python for Classification Problems🏗

MATLAB implementation | Li et al., IEEE TCE 2026 | Sorting-and-Prediction Integration for high-capacity reversible data hiding

A statistics package with a variety of bootstrap and other resampling tools. This repository is synced to the same-named repository owned by GNU-Octave. It exists to facilitate publication of the developmental version of the statistics-resampling toolbox at MathWorks FileExchange.

An R Package for quickly performing regressions over many groups and evaluating leave-one-out prediction error.

Prediction-error optimization. The agent guesses before each experiment. Wrong predictions are the learning signal. One file, zero dependencies. pip install theorist

we fit various splines to model the COVID-19 daily positive case numbers in Florida from 3/3/20 – 3/7/21.

Error-aware Bayesian Meta-Predictions and ensembles

What if the AI agent guessed first? 251 experiments. The predicting agent adapted to a 5x compute shift on try one. The reflecting agent took 7. The searching agent never caught up.

A minimal protocol for testing functional consciousness-like loops in persistent-state LLM agents.

A 2025 'ChatGPT makes students dumber' study predicted why Karpathy's 2026 AI agent would plateau. The fix worked on AI too.

This is a project from Wesleyan Computational Memory Lab (now at Swarthmore) that utilizes GPT-2 model to study human memory structure during event segmentation and recall.

AHA as prediction error

Recursive research intelligence that gets smarter every time you use it. MCP server + CLI + REST API. Prediction-error learning, surprise detection, persistent brain.