Official repository for the paper "A Modern Self-Referential Weight Matrix That Learns to Modify Itself" (ICML 2022 & NeurIPS 2021 Deep RL Workshop) and "Accelerating Neural Self-Improvement via Bootstrapping" (ICLR 2023 Workshop)
reinforcement-learning transformers pytorch fast-weights torchbeast fast-weight-programmers self-referential-weight-matrix rnn-based-few-shot-learning self-referential-learning
-
Updated
Jun 11, 2025 - Python