Beyond Policy Rollouts: Embedding-Space Directional Exploration with Anchored Noise for LLM RLVR
EDEN is an exploration method for Large Language Model Reinforcement Learning from Human Feedback (RLVR) that intervenes at the input embedding stage to explore beyond a policy's standard rollouts. It works by conditioning the policy on perturbed inputs, combining Gaussian noise with an anchor embedding, and incorporating these informative trajectories into the policy updates.