mlsanityData to 14 Oct 2026

ATHENA: Accelerated Multi-Task Heterogeneous Influence Functions for Robot Data Curation

arXiv 2606.16208 ↗First posted 15 Jun 2026cs.ROcited by 1

On the map: Vision-language-action models → Reinforcement learning and post-training

also in 1 more
Abstract (arXiv)

In robot imitation learning, influence functions provide a principled approach to quantify each demonstration's effect on robot task outcomes, yet scaling them to billion-parameter Vision-Language-Action (VLA) models is limited by computational and multitask bottlenecks. To this end, we propose ATHENA, an influence function framework tailored for multitask VLA data curation at a billion-parameter scale. Concretely, it leverages the Kronecker structure of linear-layer gradients to reduce projection cost, and approximates dense Hessian inversion with a rank-r Random Truncated Approximation, achieving about a 313.4x speedup in influence computation. Furthermore, ATHENA formulates global and local interactive influence to balance data curation across 50 jointly trained tasks. Extensive evaluations on RoboTwin 2.0 and real-robot deployment, covering 9.34 and 6.90 hours of demonstrations, respectively, show that ATHENA matches or exceeds full-data joint fine-tuning using only 50% of demonstrations in simulation and 66.7% of data across six real-robot tasks. Overall, ATHENA demonstrates its effectiveness for data curation in billion-parameter multitask VLA fine-tuning.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in Reinforcement learning and post-training, one of 13 directions in Vision-language-action models. Builds on 49 of its 55 references that are in the map. First posted on arXiv 15 Jun 2026.

Builds on

49 of 55 references.

Closest ancestors · 1

Same direction, newest first

  1. 2024STRAP: Robot Sub-Trajectory Retrieval for Augmented Policy Learning19 Dec

Other references · 48

  1. 2023RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Controlcited by 3,080
  2. 2024OpenVLA: An Open-Source Vision-Language-Action Modelcited by 2,887
  3. 2023Diffusion Policy: Visuomotor Policy Learning via Action Diffusioncited by 2,838
  4. 2024π0: A Vision-Language-Action Flow Model for General Robot Controlcited by 2,407
  5. 2025π0.5: a Vision-Language-Action Model with Open-World Generalizationcited by 1,761
  6. 2017Understanding Black-box Predictions via Influence Functionscited by 1,741
  7. 2023Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardwarecited by 1,568
  8. 2024DROID: A Large-Scale In-The-Wild Robot Manipulation Datasetcited by 852
  9. 2021What Matters in Learning from Offline Human Demonstrations for Robot Manipulationcited by 725
  10. 2019Data Shapley: Equitable Valuation of Data for Machine Learningcited by 629
  11. 2022BC-Z: Zero-Shot Task Generalization with Robotic Imitation Learningcited by 523
  12. 2025RoboTwin 2.0: A Scalable Data Generator and Benchmark with Strong Domain Randomization for Robust Bimanual Robotic Manipulationcited by 518
36 more
  1. 2020Dataset Condensation with Gradient Matchingcited by 516
  2. 2021Implicit Behavioral Cloningcited by 452
  3. 2020Transporter Networks: Rearranging the Visual World for Robotic Manipulationcited by 344
  4. 2023MimicGen: A Data Generation System for Scalable Robot Learning using Human Demonstrationscited by 327
  5. 2023TRAK: Attributing Model Behavior at Scalecited by 283
  6. 2018Dataset Distillationcited by 230
  7. 2025DexVLA: Vision-Language Model with Plug-In Diffusion Expert for General Robot Controlcited by 195
  8. 2023Scaling Robot Learning with Semantically Imagined Experiencecited by 177
  9. 2019Efficient Task-Specific Data Valuation for Nearest Neighbor Algorithmscited by 148
  10. 2019On the Accuracy of Influence Functions for Measuring Group Effectscited by 147
  11. 2024Data Scaling Laws in Imitation Learning for Robotic Manipulationcited by 145
  12. 2025Unified Vision-Language-Action Modelcited by 134
  13. 2026A Pragmatic VLA Foundation Modelcited by 103
  14. 2024What is Your Data Worth to GPT? LLM-Scale Data Valuation with Influence Functionscited by 94
  15. 2022Learning and Retrieval from Prior Data for Skill-based Imitation Learningcited by 77
  16. 2023Data Quality in Imitation Learningcited by 75
  17. 2023Behavior Retrieval: Few-Shot Imitation Learning by Querying Unlabeled Datasetscited by 58
  18. 2024SkillMimicGen: Automated Demonstration Generation for Efficient Skill Learning and Deploymentcited by 52
  19. 2025Long-VLA: Unleashing Long-Horizon Capability of Vision Language Action Model for Robot Manipulationcited by 48
  20. 2024Re-Mix: Optimizing Data Mixtures for Large Scale Imitation Learningcited by 47
  21. 2025Robot Data Curation with Mutual Information Estimatorscited by 37
  22. 2024FlowRetrieval: Flow-Guided Data Retrieval for Few-Shot Imitation Learningcited by 33
  23. 2024Influence Functions for Scalable Data Attribution in Diffusion Modelscited by 33
  24. 2025DataMIL: Selecting Data for Robot Imitation Learning with Datamodelscited by 28
  25. 2022Eliciting Compatible Demonstrations for Multi-Human Imitation Learningcited by 23
  26. 2025CUPID: Curating Data your Robot Loves with Influence Functionscited by 23
  27. 2023Learning to Discern: Imitating Heterogeneous Human Demonstrations with Preference and Representation Learningcited by 15
  28. 2024Delta-Influence: Unlearning Poisons via Influence Functionscited by 13
  29. 2024TAROT: Targeted Data Selection via Optimal Transportcited by 9
  30. 2025Fast Data Attribution for Text-to-Image Modelscited by 7
  31. 2026FrameSkip: Learning from Fewer but More Informative Frames in VLA Trainingcited by 6
  32. 2025Where Did It Go Wrong? Attributing Undesirable LLM Behaviors via Representation Gradient Tracingcited by 4
  33. 2025Toward Efficient Influence Function: Dropout as a Compression Toolcited by 2
  34. 2025Measuring Fine-Grained Relatedness in Multitask Learning via Data Attributioncited by 1
  35. 2025FT-NCFM: An Influence-Aware Data Distillation Framework for Efficient VLA Modelscited by 1
  36. 2026Quality over Quantity: Demonstration Curation via Influence Functions for Data-Centric Robot Learningcited by 1

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026SIEVE: Structure-Aware Data Selection for Imitation Learning with VLA Modelscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025SCIZOR: A Self-Supervised Approach to Data Curation for Large-Scale Imitation Learning
  2. 2020Scalable Multi-Task Imitation Learning with Autonomous Improvement
  3. 2026RoboGene: Boosting VLA Pre-training via Diversity-Driven Agentic Framework for Real-World Task Generation
  4. 2025What Matters in Learning from Large-Scale Datasets for Robot Manipulation
  5. 2024Leveraging Locality to Boost Sample Efficiency in Robotic Manipulation
  6. 2026AXIS: A Growable Community-Driven Data Engine for Scalable Robot Manipulation
  7. 2026A Systematic Study of Data Modalities and Strategies for Co-training Large Behavior Models for Robot Manipulation
  8. 2026ReGIL: Retrieval-Guided Imitation Learning from a Single Demonstration