# Recursive Language Models

**Recursive Language Models (RLMs)** enable large language models (LLMs) to process prompts significantly longer than their typical context windows by allowing them to **decompose** and **recursively call** themselves on prompt snippets, enhancing their performance.

- **Open sourced Loop Attention for Qwen3-0.6B:** Two-pass global + local attention with a learnable gate (code + weights + training script)
  
  **Loop Attention** enhances Qwen3-0.6B by implementing a **two-pass attention mechanism** that combines global and local contexts through a **learnable gate**, allowing for adaptive focus during processing.

- **Sirius DB**  
  **Sirius is a GPU-native SQL engine** that integrates seamlessly with existing databases like DuckDB, achieving over **10× speedup** in query execution without requiring major system changes.

- **LEMMA:** A Rust-based Neural-Guided Theorem Prover with 220+ Mathematical Rules  
  **LEMMA** is a **Rust-based** theorem prover that integrates **220+ mathematical rules** with a **neural-guided** approach, enhancing the efficiency of symbolic computation through **Monte Carlo Tree Search**.

- **Announcing OpenAI Grove Cohort 2**