Greg Schoeninger

How RWKV-7 Goose Works 🪿 + Notes from the Author

How Phi-4 Cracked Small Multimodality

Training a Rust 1.5B Coder LM with Reinforcement Learning (GRPO)

Why GRPO is Important and How it Works

🧠 GRPO VRAM Requirements For the GPU Poor

How DeepSeek R1, GRPO, and Previous DeepSeek Models Work