REVIEW 20 cited by
TOPReward: Token Probabilities as Hidden Zero-Shot Rewards for Robotics
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
TOPReward: Token Probabilities as Hidden Zero-Shot Rewards for Robotics
read the original abstract
General-purpose robot learning requires dense, instruction-conditioned feedback that can distinguish meaningful task progress from stalled, failed, or partially completed behavior. Yet obtaining such feedback at scale remains difficult, since existing approaches often rely on manual progress annotations, task-specific demonstrations, or reward models trained on curated robot datasets. We introduce TOPReward, a training-free progress reward method that probes pretrained Video-Language Models (VLMs) through their internal token probabilities rather than asking them to generate numerical progress values. Given a video prefix and a language instruction, TOPReward measures the model's likelihood that the instructed task has been completed, converting latent video-language understanding into a dense reward signal without task-specific reward-model training or manually annotated progress labels. We evaluate TOPReward on ManiRewardBench, our real-world manipulation benchmark spanning 130 unique tasks and four robot platforms, as well as on Open X-Embodiment datasets. Across these settings, TOPReward substantially outperforms prior training-free VLM reward methods on open-source models and is competitive with a trained reward-model baseline on progress-estimation metrics, while requiring no reward-model training. Additional analyses show that the reward is sensitive to the specified instruction and is not explained by time index alone. Finally, TOPReward supports downstream applications including success detection and offline reward-weighted behavior cloning.
Forward citations
Cited by 20 Pith papers
-
WARP-RM: A Warp-Augmented Relative Progress Reward Model for Data Curation
WARP trains a reward model on time-warped successful demonstrations to produce frame-level progress estimates that upweight high-advantage chunks during behavior cloning, maintaining high success rates on suboptimal d...
-
Improving Robotic Generalist Policies via Flow Reversal Steering
Flow Reversal Steering steers flow matching generalist policies by reversing suboptimal actions to nearby better modes, enabling improved zero-shot control, quick distillation, and RL bootstrapping in robotic manipulation.
-
Freeform Preference Learning for Robotic Manipulation
Language-conditioned multi-axis human preferences yield denser rewards and steerable robot policies that outperform sparse and binary-preference baselines by 38 points on long-horizon manipulation.
-
Freeform Preference Learning for Robotic Manipulation
Freeform Preference Learning trains language-conditioned multi-axis reward models from human pairwise preferences to produce steerable and compositional robot policies that outperform sparse and binary-preference base...
-
WARP-RM: A Warp-Augmented Relative Progress Reward Model for Data Curation
Self-supervised relative progress velocities from time-warped demos let WARP-BC reweight action chunks so imitation policies stay robust as suboptimal teleop data is admitted.
-
WARP-RM: A Warp-Augmented Relative Progress Reward Model for Data Curation
Self-supervised relative progress from time-warped demos reweights BC action chunks, sustaining ~19/20 success and up to ~18× throughput on mixed-quality T-shirt folding where vanilla BC fails.
-
Beyond Monotonic Progress: Retry-Supervised Value Learning for Robot Imitation
ReTVL uses retry events as sparse supervision to train mistake-sensitive value functions that reweight demonstration chunks for improved behavior cloning on real-robot manipulation tasks.
-
Beyond Monotonic Progress: Retry-Supervised Value Learning for Robot Imitation
Sparse retry keypoints plus pairwise preference learning yield mistake-sensitive values that reweight mixed-quality demos and raise real-robot imitation success over progress-based baselines.
-
RARM: Confidence-Gated Progress Reward Modeling for RL in Manipulation
RARM is a lightweight visual comparator trained once on general videos that supplies dense progress rewards to RL by matching rollout clips to a reference demonstration and gating rewards on match confidence.
-
SARM2: Multi-Task Stage Aware Reward Modeling for Self Improving Robotic Manipulation
SARM2 presents RM, a multi-task stage-aware reward model achieving 80% lower value-estimation MSE, which when used in SPIRAL boosts manipulation task success from ~50% to near-perfect on several benchmarks.
-
StressDream: Steering Video World Models for Robust Policy Evaluation and Improvement
StressDream optimizes initial noise in diffusion video world models using VLM semantic and plausibility objectives to steer generations toward specified high-impact outcomes for improved policy evaluation.
-
Stable-Layers: Fine-Tuning Image Layer Decomposition Models with VLM-Scored Reinforcement Learning
Stable-Layers applies Flow-GRPO with LoRA and a two-stage VLM scoring pipeline to improve layer decomposition without paired supervision, yielding stronger separation and lower reconstruction error on Crello.
-
Beyond Pixels: Learning Invariant Rewards for Real-World Robotics From a Few Demonstrations
A framework learns invariant symbolic reward functions from few demonstrations that generalize zero-shot to variations in robotic manipulation tasks.
-
GuidedVLA: Specifying Task-Relevant Factors via Plug-and-Play Action Attention Specialization
GuidedVLA improves VLA success rates by manually supervising separate attention heads in the action decoder with auxiliary signals for task-relevant factors.
-
GuidedVLA: Specifying Task-Relevant Factors via Plug-and-Play Action Attention Specialization
GuidedVLA improves VLA generalization by supervising individual attention heads with manually defined auxiliary signals for three task-relevant factors.
-
Grounded World Model for Semantically Generalizable Planning
A vision-language-aligned world model turns visuomotor MPC into a language-following planner that reaches 87% success on 288 unseen semantic tasks where standard VLAs drop to 22%.
-
Robometer: Scaling General-Purpose Robotic Reward Models via Trajectory Comparisons
Robometer combines intra-trajectory progress supervision with inter-trajectory preference supervision on a 1M-trajectory dataset to learn more generalizable robotic reward functions than prior methods.
-
LLM-as-a-Verifier: A General-Purpose Verification Framework
Computing the expectation over scoring-token logits instead of taking argmax enables verification to scale along granularity, repetition, and criteria decomposition, achieving state-of-the-art on four agentic benchmar...
-
LLM-as-a-Verifier: A General-Purpose Verification Framework
Expecting over scoring-token logits yields continuous, scalable verification that improves agent trajectory selection and dense RL rewards across coding, robotics, and medical benchmarks.
-
World Value Models for Robotic Manipulation
World Value Model (WVM) integrates world models with value estimation to achieve SOTA Value-Order Correlation on expert and suboptimal robotic data and improves downstream policy performance.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.