REVIEW 16 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
Learning Long-Context Diffusion Policies via Past-Token Prediction
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.
Forward citations
Cited by 16 Pith papers
-
RoboTTT: Context Scaling for Robot Policies
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.
-
Present but Not Remembered: Auditing How Frozen VLAs Encode, Deploy, and Steer Visual History
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.
-
DSSP: Diffusion State Space Policy with Full-History Encoding
DSSP is a history-conditioned diffusion state space policy that uses SSMs to encode full observation streams with an auxiliary dynamics objective and hierarchical fusion, achieving SOTA results with reduced model size...
-
RoboMME: Benchmarking and Understanding Memory for Robotic Generalist Policies
A standardized 16-task memory benchmark and 14 controlled VLA variants show perceptual memory (especially frame sampling + AdaLN modulation) is strongest overall, while symbolic subgoals win on counting and short-hori...
-
HiMe: Hierarchical Embodied Memory for Long-Horizon Vision-Language-Action Control
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.
-
Remember what you did?: Learning Behavioral Memories for Partially Observable Object Manipulation
CAMP learns a compressed behavioral memory from action history to enable success in long-horizon partially observable object manipulation without extra supervision, showing gains over baselines in real-robot and simul...
-
EventVLA: Event-Driven Visual Evidence Memory for Long-Horizon Vision-Language-Action Policies
EventVLA introduces foundational visual anchors and a Keyframe Evidence Memory module that predicts future keyframe probabilities from VLA embeddings to improve long-horizon task success by an average of 40% on 17 sim...
-
Training and Evaluating Diffusion Policies with Long Context Lengths
Naive long-context Diffusion Policies succeed with UNet+Cross-Attention and sufficient data; variable-history training cuts sample complexity in the low-data regime.
-
Expanding Spatial and Temporal Context for Robotic Imitation Learning With Scene Graphs
Dynamic scene graphs serve as explicit memory to improve imitation learning policies for spatial-temporal reasoning under partial observability in mobile and tabletop manipulation.
-
Persistent Visual Memory: Sustaining Perception for Deep Generation in LVLMs
PVM adds a parallel branch to LVLMs that directly supplies visual embeddings to prevent attention decay over long generated sequences, yielding accuracy gains on reasoning tasks with minimal overhead.
-
RoboMME: Benchmarking and Understanding Memory for Robotic Generalist Policies
RoboMME is a new benchmark with 16 tasks and 14 memory-augmented VLA variants that shows memory effectiveness is highly task-dependent.
-
EventVLA: Event-Driven Visual Evidence Memory for Long-Horizon Vision-Language-Action Policies
EventVLA uses a foresight-driven Keyframe Evidence Memory module to selectively store sparse task-critical visual events predicted from VLA latent embeddings, reporting +40% average success rate gains on 17 simulation...
-
Persistent Visual Memory: Sustaining Perception for Deep Generation in LVLMs
PVM adds a parallel learnable branch to LVLMs that supplies visual embeddings on demand to structurally prevent attention decay and visual signal dilution during deep autoregressive generation.
-
Gated Memory Policy
GMP selectively activates and represents memory via a gate and lightweight cross-attention, yielding 30.1% higher success on non-Markovian robotic tasks while staying competitive on Markovian ones.
-
RLDX-1 Technical Report
RLDX-1 achieves 86.8% success on complex ALLEX humanoid manipulation tasks where prior VLAs reach only around 40%.
-
RLDX-1 Technical Report
RLDX-1 outperforms frontier VLAs such as π0.5 and GR00T N1.6 on dexterous manipulation benchmarks, reaching 86.8% success on ALLEX humanoid tasks versus around 40% for the baselines.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.