In a new preprint paper, researchers from UC Santa Cruz show that it is possible to eliminate the most computationally expensive element of running large language models, called matrix multiplication ...

The new matrix has the same number of rows as the first matrix and the same number of columns as the second matrix. The matrix multiplication operator does not consistently propagate missing values.

This could eventually accelerate AI models like ChatGPT, which rely heavily on matrix multiplication to function ... bringing it closer to the ideal value of 2, which represents the theoretical ...