Fast-dLLM: Training-free Acceleration of Diffusion LLM by Enabling KV Cache and Parallel Decoding Paper • 2505.22618 • Published May 28, 2025 • 45
