The Problem With Large Language Models

You have a data problem here. There are attempts to take genetic information from microbes and ignore the semantics. So this approach would actually not work. In addition, it would just predict the next token, which is not necessarily what you want. What we have in the microbiome is the tiny tiny tiny fractures, orders of smugnet you smaller than what would be required. And so many, many problems in applying this approach to genetic information today from microbiome. Take all this data, it's tiny compared to what you need in order to take this approach.