what? I was working on this stuff 15 years ago and it was already an old field at that point. the tech is unambiguously not old. they just managed to train an LLM with significantly more parameters than we could manage back then because of computing power enhancements. undoubtedly, there have been improvements in the algorithms but it’s ahistorical to call this new tech.
what? I was working on this stuff 15 years ago and it was already an old field at that point. the tech is unambiguously not old. they just managed to train an LLM with significantly more parameters than we could manage back then because of computing power enhancements. undoubtedly, there have been improvements in the algorithms but it’s ahistorical to call this new tech.