Both you and your sibling are approaching LLMs like it is some sort of science. If you do that there is no wonder you have a lot of unanswered questions. LLMs a

Both you and your sibling are approaching LLMs like it is some sort of science. If you do that there is no wonder you have a lot of unanswered questions. LLMs are not a science, they are applied statistics. Making predictions to evaluate hypothesis and constructing theories around the hyperparameters of LLMs is no different then making predictions to evaluate hypothesis and constructing theories around the configurations of Nuclear Power Plants, the latter of course being applied physics.