Transformers are Minimax Optimal Nonparametric In-Context Learners | Read Paper on Bytez