Pith. sign in

REVIEW 9 cited by

Vision-Language Models are Zero-Shot Reward Models for Reinforcement Learning

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2310.12921 v2 pith:2MNSEPXW submitted 2023-10-19 cs.LG cs.AI

classification cs.LGcs.AI
keywords rewardmodelsvlm-rmsvlmslearningprompttasksbaseline
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Reinforcement learning (RL) requires either manually specifying a reward function, which is often infeasible, or learning a reward model from a large amount of human feedback, which is often very expensive. We study a more sample-efficient alternative: using pretrained vision-language models (VLMs) as zero-shot reward models (RMs) to specify tasks via natural language. We propose a natural and general approach to using VLMs as reward models, which we call VLM-RMs. We use VLM-RMs based on CLIP to train a MuJoCo humanoid to learn complex tasks without a manually specified reward function, such as kneeling, doing the splits, and sitting in a lotus position. For each of these tasks, we only provide a single sentence text prompt describing the desired task with minimal prompt engineering. We provide videos of the trained agents at: https://sites.google.com/view/vlm-rm. We can improve performance by providing a second "baseline" prompt and projecting out parts of the CLIP embedding space irrelevant to distinguish between goal and baseline. Further, we find a strong scaling effect for VLM-RMs: larger VLMs trained with more compute and data are better reward models. The failure modes of VLM-RMs we encountered are all related to known capability limitations of current VLMs, such as limited spatial reasoning ability or visually unrealistic environments that are far off-distribution for the VLM. We find that VLM-RMs are remarkably robust as long as the VLM is large enough. This suggests that future VLMs will become more and more useful reward models for a wide range of RL applications.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 9 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. SOLE-R1: Video-Language Reasoning as the Sole Reward for On-Robot Reinforcement Learning

    cs.RO 2026-03 conditional novelty 6.5 of 10

    A video-language model with per-timestep spatiotemporal CoT and dense progress prediction can serve as the sole reward for zero-shot online robot RL on 24 unseen manipulation tasks.

  2. Weights or Skills? A Survey of Robot-Learning Techniques: from Action-Predicting Weights to Robots that Write their Own Skills

    cs.RO 2026-08 conditional novelty 6.0 of 10

    A taxonomy of robot learning on a weights-versus-skills axis, with a five-rung self-improvement ladder whose top cell (feedback plus memory plus search) holds only a few recent systems.

  3. MARS-RA: Rank Aggregation for Credit Assignment via Multimodal Comparisons in Embodied Multi-Agent Cooperation

    cs.AI 2026-07 conditional novelty 6.0 of 10

    Credit assignment via LMM pairwise comparisons plus Bradley–Terry rank aggregation and potential-based shaping improves cooperative MARL under sparse rewards and dynamic agent counts.

  4. PhyAgentOS: A Self-Evolving Operating System for Embodied Agents with Decoupled Cognitive Planning and Physical Execution

    cs.RO 2026-07 conditional novelty 6.0 of 10

    A file-based operating-system layer with a session verifier and persistent memory improves embodied-agent task completion on game, simulated, and real-robot platforms without retraining policies.

  5. TOPReward: Token Probabilities as Hidden Zero-Shot Rewards for Robotics

    cs.RO 2026-02 conditional novelty 6.0 of 10

    The log-probability a VLM assigns to 'True' for 'does this video prefix complete the task?' is used as a zero-shot dense progress reward that outperforms GVL on open-source models.

  6. FOUNDER: Grounding Foundation Models in World Models for Open-Ended Embodied Decision Making

    cs.RO 2025-07 conditional novelty 6.0 of 10

    FOUNDER maps foundation-model embeddings of text or video prompts into world-model goal states and rewards policies by predicted temporal distance to those goals, improving reward-free multi-task offline control.

  7. RobotSmith: Generative Robotic Tool Design for Acquisition of Complex Manipulation Skills

    cs.RO 2025-06 conditional novelty 6.0 of 10

    RobotSmith autonomously designs, 3D-prints, and uses task-specific tools for robotic manipulation, raising task success from 2.8% (no tool) to 50% in simulation.

  8. LLM-as-a-Verifier: A General-Purpose Verification Framework

    cs.AI 2026-07 conditional novelty 5.0 of 10

    Expecting over scoring-token logits yields continuous, scalable verification that improves agent trajectory selection and dense RL rewards across coding, robotics, and medical benchmarks.

  9. Advancing Autonomous VLM Agents via Variational Subgoal-Conditioned Reinforcement Learning

    cs.LG 2025-02 reject novelty 4.0 of 10

    VSC-RL combines VLM-generated subgoals with a subgoal-conditioned AWR-style RL objective and claims improved sample efficiency over DigiRL and WebRL on AitW and WebArena-Lite.

Pith tools