Harry00 -
: Unlike autoregressive LLMs, it uses energy minimization to "reason" through problems.
: This foundational paper introduces a mathematical model for human long-term memory using high-dimensional binary vectors and Hamming distance for addressing. harry00
: This work details how to perform "binding" of information (connecting concepts) using circular convolution, a technique Harry00 utilizes for bitwise reasoning without standard backpropagation. : Unlike autoregressive LLMs, it uses energy minimization
: It avoids traditional training data and GPU-heavy gradients. : Unlike autoregressive LLMs
The MLE-Morpho-Logic-Engine is built on several landmark papers in neural computing and vector logic:
: This modern paper connects traditional associative memories to the attention mechanisms used in current LLMs, providing the energy minimization framework that the MLE project aims to optimize. Key Technical Aspects
