Recurrent Looped Transformer

AI 摘要

The Recurrent Looped Transformer is a subject of a tech report authored by Yifan Zhang in September 2026. The report, available on GitHub, details a model where the probability of the next token, p_\Theta(x_{t+1}\mid x_{1:t}), is determined by a softmax function applied to a weighted and RMS-normalized state, specifically softmax(W_o RMSNorm_o(s_t))_{x_{t+1}}.