Pith. sign in

REVIEW 5 cited by

Learning Long-Context Diffusion Policies via Past-Token Prediction

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 2505.09561 v2 pith:YWJKGJLS submitted 2025-05-14 cs.RO cs.AIcs.LG

classification cs.ROcs.AIcs.LG
keywords policyactionslong-contextpastpoliciesdiffusionlearningtraining
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Reasoning over long sequences of observations and actions is essential for many robotic tasks. Yet, learning effective long-context policies from demonstrations remains challenging. As context length increases, training becomes increasingly expensive due to rising memory demands, and policy performance often degrades as a result of spurious correlations. Recent methods typically sidestep these issues by truncating context length, discarding historical information that may be critical for subsequent decisions. In this paper, we propose an alternative approach that explicitly regularizes the retention of past information. We first revisit the copycat problem in imitation learning and identify an opposite challenge in recent diffusion policies: rather than over-relying on prior actions, they often fail to capture essential dependencies between past and future actions. To address this, we introduce Past-Token Prediction (PTP), an auxiliary task in which the policy learns to predict past action tokens alongside future ones. This regularization significantly improves temporal modeling in the policy head, with minimal reliance on visual representations. Building on this observation, we further introduce a multistage training strategy: pre-train the visual encoder with short contexts, and fine-tune the policy head using cached long-context embeddings. This strategy preserves the benefits of PTP while greatly reducing memory and computational overhead. Finally, we extend PTP into a self-verification mechanism at test time, enabling the policy to score and select candidates consistent with past actions during inference. Experiments across four real-world and six simulated tasks demonstrate that our proposed method improves the performance of long-context diffusion policies by 3x and accelerates policy training by more than 10x.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 5 Pith papers

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

  1. RoboTTT: Context Scaling for Robot Policies

    cs.RO 2026-07 conditional novelty 7.0 of 10

    A robot policy that updates its own weights during deployment can use 8,000 steps of history, steadily improving as context grows and enabling one-shot imitation from human videos.

  2. Present but Not Remembered: Auditing How Frozen VLAs Encode, Deploy, and Steer Visual History

    cs.CV 2026-07 conditional novelty 7.0 of 10

    VLAs linearly encode past frames as near-redundant copies of the present and deploy them only as architecture-conditional fallback or standing use, so steerability tracks deployment regime not encoding.

  3. Why Does Action Chunking Improve Behavioral Cloning Performance in Robotic Control?

    cs.RO 2026-08 conditional novelty 6.0 of 10

    Action chunking in robotic behavioral cloning works mainly because it acts as a delayed-prediction policy and an implicit ensemble, not because of temporal consistency or horizon reduction.

  4. HiMe: Hierarchical Embodied Memory for Long-Horizon Vision-Language-Action Control

    cs.RO 2026-07 conditional novelty 6.0 of 10

    A three-layer embodied memory (Executor, Sentry, Planner) with cross-modal Add/Update/Delete operations raises long-horizon VLA task progress to ~90% and reduces planner API calls versus flat memory.

  5. Training and Evaluating Diffusion Policies with Long Context Lengths

    cs.RO 2026-06 conditional novelty 6.0 of 10

    Naive long-context Diffusion Policies succeed with UNet+Cross-Attention and sufficient data; variable-history training cuts sample complexity in the low-data regime.

Pith tools