Show HN: Linear RNN/Reservoir hybrid generative model, one C file (no deps.)

Category: ai-ml

Tags: machine-learning, rnn, generative-models, c-language, neural-networks

Score: 6.3/10 (Innovation: 7, Technical: 8, Documentation: 3, Utility: 4)

A single-file C implementation of a hybrid linear RNN/reservoir computing model for generative sequence tasks, combining elements from transformers (SwiGLU, ALiBi, slot memory) with recurrent architectures for efficient inference. It's interesting as a research artifact exploring memory-efficient alternatives to transformers with claimed linear generation complexity.

Target audience: ai-researchers, ml-engineers, low-level-optimization-devs

Repository: https://raw.githubusercontent.com/bggb7781-collab/lrnnsmdds/refs/heads/main/lrnnsmdds

View on Hacker News