--- title: The Void - Buleyean RL Demo emoji: 🕳️ colorFrom: gray colorTo: indigo sdk: gradio sdk_version: 5.0.0 app_file: app.py pinned: true --- # The Void -- Buleyean RL Demo An interactive demo comparing Buleyean RL-trained language models against their base counterparts. Buleyean RL transforms high-entropy model outputs into low-entropy, structured responses by applying a probability-theoretic reinforcement learning framework rooted in Buleyean probability. ## Models - **SmolLM2-360M**: Ultra-lightweight model, ideal for fast comparisons - **Qwen2.5-0.5B**: Small but capable instruction-following model ## How It Works 1. Enter a prompt or select an example 2. Choose a model from the dropdown 3. See side-by-side outputs: base model vs Buleyean-trained 4. Compare entropy metrics, optimality gap, and token efficiency