Popular GitHub Repositories for Capacity Bounds

The code of our paper "InfLLM: Unveiling the Intrinsic Capacity of LLMs for Understanding Extremely Long Sequences with Training-Free Memory"

auto_LiRPA: An Automatic Linear Relaxation based Perturbation Analysis Library for Neural Networks and General Computational Graphs

PyTorch reimplementation of the paper "Swin Transformer V2: Scaling Up Capacity and Resolution" [CVPR 2022].

Certified defense to adversarial examples using CROWN and IBP. Also includes GPU implementation of CROWN verification algorithm (in PyTorch).

core code for High-Capacity Convolutional Video Steganography with Temporal Residual Modeling

Efficient, lightweight variational inference and approximation bounds

implement basic and contextual MAB algorithms for recommendation system

Pytorch implementation of our paper accepted by ECCV2022 -- Dynamic Dual Trainable Bounds for Ultra-low Precision Super-Resolution Networks

Implementation of the paper "Improving Optimization Bounds using Machine Learning: Decision Diagrams meet Deep Reinforcement Learning".

Bastien One is being developed as autonomous A.I. bot with the capacity to complete complex tasks - either by itself or by creating additional specialised bots should it be required.

[NeurIPS '24 Spotlight] PertEval: Unveiling Real Knowledge Capacity of LLMs via Knowledge-invariant Perturbations

Custom YOLO11m model for detecting and classifying car body damage (99% shattered glass, 96% flat tire detection accuracy)—optimized for high-capacity inference and assistive use in inspection and service workflows like BMW pre-loaner inspections.

A summary of my Individual Project for University of Southampton. This project aimed to inform a user if they were performing a movement correctly based on a pretrained model, with the capacity to train new models given sufficient data. View the results and methods at Final_Report.pdf

AI for the game "Connect Four". Available on PyPI.

Example machine learning implementation to predict the residual bending moment capacity of corroded reinforced concrete beams tested under monotonic three or four-point bending. Data is collected from 54 experimental programs available in the literature.

Welcome to the ongoing adventures of Stan-The Growing Stump in his never-ending quest for understanding and connection. Stan's journey explores the bounds of artificial intelligence within the framework of a Discord bot, navigating the complexities of memory, interaction, and existential musing.

A Comparative Evaluation of Active Learning Methods in Deep Recommendation

?? ? ? A project based in Autonomous and Intelligent Agents. This project was built using Java, JADE (JAVA Agent DEvelopment Framework) and Rapid Miner Studio. The scenario chosen for this project was to combat fires with the use of flying and terrestrial autonomous and intelligent vehicles, like trucks and airplanes. In this scenario we have, as agents, the trucks, airplanes and the headquarters (responsible for coordinating the airplanes and trucks). And also, as an environment, the fires and the water supply stations (rivers, seas, lagoons, among others). Airplanes and trucks would have limited fuel and water supply capacity.

Code for the paper "Finite-Time Frequentist Regret Bounds of Multi-Agent Thompson Sampling on Sparse Hypergraphs", AAAI Conference on Artificial Intelligence 2024

This repo analyzes capacity bounds of communication systems with quantization and spectral constraints. It includes theoretical derivations and numerical evaluations of mutual information under coarse quantization and bandwidth limitations, with applications in modern wireless systems.