Implementation for the paper "Compositional Attention Networks for Machine Reasoning" (Hudson and Manning, ICLR 2018)
A Constrained Text Generation Challenge Towards Generative Commonsense Reasoning
Implementation of ICML 23 Paper: Specializing Smaller Language Models towards Multi-Step Reasoning.
The MacGraph network. An attempt to get MACnets running on graph knowledge
Neural Logic Programming Language
Memory, Attention and Composition (MAC) Network for CLEVR implemented via KnetLayers
A smart meal planner for everyone, including diabetics
Knet implementation of "A simple neural network module for relational reasoning" (Relational Networks) https://arxiv.org/pdf/1706.01427.pdf
A lightweight expert-system framework for modeling and testing medical diagnostic rules, with multiple approaches.
Harmonic — independent third-party profile of a public API surface, by API Evangelist. Harmonic is a Palo Alto AI lab building Mathematical Superintelligence (MSI) — AI that reasons with rigorous, verifiable logic rather than statistical pattern matching. Its flagship product, Aristotle, is a formal reasoning agent that uses Lean 4 to prove and for
The website of Vincent Liu (WiseMotif.ai): The Grothendieckian Programme, Series 00–07 of the WiseMotif Monograph of Mathematical Intelligence — the overview and the seven works, with their DOIs.
Java and Python wrapper for the Julia language PDDL library (PDDL.jl)
Profile of Vincent Liu (WiseMotif.ai): mathematics and machine reasoning — The Grothendieckian Programme, Series 00–07.
The official site of Grothendieck Lab (WiseMotif.ai) and of the Grothendieckian Programme: the overview and the seven works, the method, the machine, the open problems.
Three papers on the Grothendieckian Programme — an overview, twelve theses and 57 open problems for mathematics and machine reasoning, with problem cards, problems.json and Lean statements. Take 57.
The Grothendieckian Programme: seven works, 34 volumes, 26,021 pages on machine reasoning about mathematics — a core, a topos-native kernel and five research subjects (RH, BSD, Hodge, abc, Langlands). Series 00, their overview, adds no mathematical result; none of the seven claims a proof of the problems it studies. DOI 10.5281/zenodo.23669107.
A research review about reasoning in Artificial Neural Networks.
Course Project, UniStuttgart, WiSe2425
A research program on whether adaptive systems can become appropriately different without losing their capacity for justified correction.
To associate your repository with the machine-reasoning topic, visit your repo's landing page and select "manage topics."