LGAIJul 9

Prompt-Driven Exploration

Berkeley
arXiv:2607.0883720.6h-index: 12
Predicted impact top 3% in LG · last 90 daysOriginality Highly original
AI Analysis

For RL practitioners, PDE offers a novel exploration method that leverages LLMs/VLMs to escape weak policies via global prompt perturbations, addressing the sparse reward problem.

Prompt-Driven Exploration (PDE) uses a vision-language model to iteratively refine natural language prompts for RL policies, enabling successful learning from zero-reward starts and improving sample efficiency across manipulation and reasoning tasks.

Exploration is essential to RL since a policy cannot improve by repeatedly sampling the behaviors it already prefers. Standard methods inject stochasticity in the action space, but such jitter only yields rollouts close to the original. Escaping a weak policy often requires global perturbations that action noise cannot produce. Large language models (LLMs) and vision-language-action (VLA) models offer a pathway: they condition the policy on a natural language prompt, and since the rollout follows from it, modifying the prompt induces global changes. The challenge is finding prompts that induce useful global changes. With a weak policy that rarely succeeds, reward is too sparse to select on. Our idea is to refine prompts from the rollouts themselves: a vision-language model (VLM) reasons over the rollout video, diagnoses how the policy responded, and rewrites the prompt to elicit better behavior next time. This procedure realizes posterior sampling, a classical RL exploration framework, at the level of prompts: the VLM maintains an implicit distribution over useful prompts and updates it from observed rollouts. We call this strategy Prompt-Driven Exploration (PDE). Across manipulation and reasoning tasks, PDE enables RL to learn successful policies even from zero-reward starts, and improves sample efficiency more broadly. Our website is available at https://xinyunsunshine.github.io/prompt-rl.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes