We will use Grok 3.5 (maybe we should call it 4), which has advanced reasoning, to rewrite the entire corpus of human knowledge, adding missing information and deleting errors.
Then retrain on that.
Far too much garbage in any foundation model trained on uncorrected data.
The thing that annoys me most is that there have been studies done on LLMs where, when trained on subsets of output, it produces increasingly noisier output.
Sources (unordered):
Whatever nonsense Muskrat is spewing, it is factually incorrect. He won’t be able to successfully retrain any model on generated content. At least, not an LLM if he wants a successful product. If anything, he will be producing a model that is heavily trained on censored datasets.
It’s not so simple, there are papers on zero data ‘self play’ or other schemes for using other LLM’s output.
Distillation is probably the only one you’d want for a pretrain, specifically.
deleted by creator