honestly i dont actually expect it to improve anything - kinda did it as a rebellious shower thought when the jev homescreen said "we are not a chatbot"
the underlying mechanism has to sample tokens (1000/90000 for openai's tokenizer) by some shallow heuristic to build the decision pool (more details are in https://github.com/giga-james/jevgpt/blob/main/docs/experime... !), while standard autoregressive models have access to the entire token vocabulary to predict the next word
the underlying mechanism has to sample tokens (1000/90000 for openai's tokenizer) by some shallow heuristic to build the decision pool (more details are in https://github.com/giga-james/jevgpt/blob/main/docs/experime... !), while standard autoregressive models have access to the entire token vocabulary to predict the next word