DigiNews

Summary

alexzhang13/rlm presents Recursive Language Models (RLMs), a framework for building language-models that can perform recursive sub-call reasoning and interact with a code/execution environment. The repository includes an extensible inference engine, training harnesses, and support for multiple sandbox environments, with examples, documentation, and a minimal to full-stack setup. It emphasizes open-source, research-oriented tooling and integration with various model providers and sandboxes.