Close Menu
    DevStackTipsDevStackTips
    • Home
    • News & Updates
      1. Tech & Work
      2. View All

      How To Prevent WordPress SQL Injection Attacks

      June 14, 2025

      This week in AI dev tools: Apple’s Foundations Model framework, Mistral’s first reasoning model, and more (June 13, 2025)

      June 13, 2025

      Open Talent platforms emerging to match skilled workers to needs, study finds

      June 13, 2025

      Java never goes out of style: Celebrating 30 years of the language

      June 12, 2025

      6 registry tweaks every tech-savvy user must apply on Windows 11

      June 14, 2025

      Here’s why network infrastructure is vital to maximizing your company’s AI adoption

      June 14, 2025

      The AI video tool behind the most viral social trends right now

      June 14, 2025

      Got a new password manager? How to clean up the password mess you left in the cloud

      June 14, 2025
    • Development
      1. Algorithms & Data Structures
      2. Artificial Intelligence
      3. Back-End Development
      4. Databases
      5. Front-End Development
      6. Libraries & Frameworks
      7. Machine Learning
      8. Security
      9. Software Engineering
      10. Tools & IDEs
      11. Web Design
      12. Web Development
      13. Web Security
      14. Programming Languages
        • PHP
        • JavaScript
      Featured

      Right Invoicing App for iPhone: InvoiceTemple

      June 14, 2025
      Recent

      Right Invoicing App for iPhone: InvoiceTemple

      June 14, 2025

      Tunnel Run game in 170 lines of pure JS

      June 14, 2025

      Integrating Drupal with Salesforce SSO via SAML and Dynamic User Sync

      June 14, 2025
    • Operating Systems
      1. Windows
      2. Linux
      3. macOS
      Featured

      Windows 11 24H2 tests toggle to turn off Recommended feed in the Start menu

      June 14, 2025
      Recent

      Windows 11 24H2 tests toggle to turn off Recommended feed in the Start menu

      June 14, 2025

      User calls Windows 11 “pure horror,” Microsoft says it’s listening to feedback

      June 14, 2025

      John the Ripper is an advanced offline password cracker

      June 14, 2025
    • Learning Resources
      • Books
      • Cheatsheets
      • Tutorials & Guides
    Home»Development»Machine Learning»Meta AI Releases V-JEPA 2: Open-Source Self-Supervised World Models for Understanding, Prediction, and Planning

    Meta AI Releases V-JEPA 2: Open-Source Self-Supervised World Models for Understanding, Prediction, and Planning

    June 12, 2025

    Meta AI has introduced V-JEPA 2, a scalable open-source world model designed to learn from video at internet scale and enable robust visual understanding, future state prediction, and zero-shot planning. Building upon the joint-embedding predictive architecture (JEPA), V-JEPA 2 demonstrates how self-supervised learning from passive internet video, combined with minimal robot interaction data, can yield a modular foundation for intelligent physical agents.

    Scalable Self-Supervised Pretraining from 1M Hours of Video

    V-JEPA 2 is pretrained on over 1 million hours of internet-scale video combined with 1 million images. Using a visual mask denoising objective, the model learns to reconstruct masked spatiotemporal patches in a latent representation space. This approach avoids the inefficiencies of pixel-level prediction by focusing on predictable scene dynamics while disregarding irrelevant noise.

    To scale JEPA pretraining to this level, Meta researchers introduced four key techniques:

    • Data scaling: Constructed a 22M-sample dataset (VideoMix22M) from public sources like SSv2, Kinetics, HowTo100M, YT-Temporal-1B, and ImageNet.
    • Model scaling: Expanded the encoder capacity to over 1B parameters using ViT-g.
    • Training schedule: Adopted a progressive resolution strategy and extended pretraining to 252K iterations.
    • Spatial-temporal augmentation: Trained on progressively longer and higher-resolution clips, reaching 64 frames at 384×384 resolution.

    These design choices led to an 88.2% average accuracy across six benchmark tasks—including SSv2, Diving-48, Jester, Kinetics, COIN, and ImageNet—surpassing previous baselines.

    Understanding via Masked Representation Learning

    V-JEPA 2 exhibits strong motion understanding capabilities. On the Something-Something v2 benchmark, it achieves 77.3% top-1 accuracy, outperforming models like InternVideo and VideoMAEv2. For appearance understanding, it remains competitive with state-of-the-art image-text pretraining models like DINOv2 and PEcoreG.

    The encoder’s representations were evaluated using attentive probes, verifying that self-supervised learning alone can yield transferable and domain-agnostic visual features applicable across diverse classification tasks.

    Temporal Reasoning via Video Question Answering

    To assess temporal reasoning, the V-JEPA 2 encoder is aligned with a multimodal large language model and evaluated on multiple video question-answering tasks. Despite lacking language supervision during pretraining, the model achieves:

    • 84.0% on PerceptionTest
    • 76.9% on TempCompass
    • 44.5% on MVP
    • 36.7% on TemporalBench
    • 40.3% on TOMATO

    These results challenge the assumption that visual-language alignment requires co-training from the start, demonstrating that a pretrained video encoder can be aligned post hoc with strong generalization.

    V-JEPA 2-AC: Learning Latent World Models for Robotic Planning

    A key innovation in this release is V-JEPA 2-AC, an action-conditioned variant of the pretrained encoder. Fine-tuned using only 62 hours of unlabeled robot video from the Droid dataset, V-JEPA 2-AC learns to predict future video embeddings conditioned on robot actions and poses. The architecture is a 300M parameter transformer with block-causal attention, trained using a teacher-forcing and rollout objective.

    This allows zero-shot planning through model-predictive control. The model infers action sequences by minimizing the distance between imagined future states and visual goals using the Cross-Entropy Method (CEM). It achieves high success in tasks such as reaching, grasping, and pick-and-place on unseen robot arms in different labs—without any reward supervision or additional data collection.

    Benchmarks: Robust Performance and Planning Efficiency

    Compared to baselines like Octo (behavior cloning) and Cosmos (latent diffusion world models), V-JEPA 2-AC:

    • Executes plans in ~16 seconds per step (versus 4 minutes for Cosmos).
    • Reaches a 100% success rate on reach tasks.
    • Outperforms others in grasp and manipulation tasks across object types.

    Notably, it operates using a monocular RGB camera without calibration or environment-specific fine-tuning, reinforcing the generalization capability of the learned world model.

    Conclusion

    Meta’s V-JEPA 2 represents a significant advancement in scalable self-supervised learning for physical intelligence. By decoupling observation learning from action conditioning and leveraging large-scale passive video, V-JEPA 2 demonstrates that general-purpose visual representations can be harnessed for both perception and control in the real world.


    Check out the Paper, Models on Hugging Face and GitHub Page. All credit for this research goes to the researchers of this project. Also, feel free to follow us on Twitter and don’t forget to join our 99k+ ML SubReddit and Subscribe to our Newsletter.

    The post Meta AI Releases V-JEPA 2: Open-Source Self-Supervised World Models for Understanding, Prediction, and Planning appeared first on MarkTechPost.

    Source: Read More 

    Facebook Twitter Reddit Email Copy Link
    Previous ArticleDeveloper Spotlight: Robin Payot
    Next Article Run Multiple AI Coding Agents in Parallel with Container-Use from Dagger

    Related Posts

    Machine Learning

    How to Evaluate Jailbreak Methods: A Case Study with the StrongREJECT Benchmark

    June 14, 2025
    Machine Learning

    MemOS: A Memory-Centric Operating System for Evolving and Adaptive Large Language Models

    June 14, 2025
    Leave A Reply Cancel Reply

    For security, use of Google's reCAPTCHA service is required which is subject to the Google Privacy Policy and Terms of Use.

    Continue Reading

    Oblivion Remastered runs a lot better with DLSS 4 enabled — Here’s how to force the override

    News & Updates

    Scaling Reinforcement Learning Beyond Math: Researchers from NVIDIA AI and CMU Propose Nemotron-CrossThink for Multi-Domain Reasoning with Verifiable Reward Modeling

    Machine Learning

    Omni-R1: Advancing Audio Question Answering with Text-Driven Reinforcement Learning and Auto-Generated Data

    Machine Learning

    Obama to propose legislation that protects firms sharing cyberthreat data

    Development

    Highlights

    CVE-2024-55567 – Insyde H2O UsbCoreDxe SMM Call Out Vulnerability

    June 12, 2025

    CVE ID : CVE-2024-55567

    Published : June 12, 2025, 5:15 p.m. | 50 minutes ago

    Description : Improper input validation was discovered in UsbCoreDxe in Insyde InsydeH2O kernel 5.4 before 05.47.01, 5.5 before 05.55.01, 5.6 before 05.62.01, and 5.7 before 05.71.01. The SMM module has an SMM call out vulnerability which can be used to write arbitrary memory inside SMRAM and execute arbitrary code at SMM level.

    Severity: 7.5 | HIGH

    Visit the link for more details, such as CVSS details, affected products, timeline, and more…

    CVE-2025-43922 – FileWave Windows Privilege Escalation

    April 21, 2025

    What’s the point of Headless?

    April 23, 2025

    GenCast predicts weather and the risks of extreme conditions with state-of-the-art accuracy

    May 29, 2025
    © DevStackTips 2025. All rights reserved.
    • Contact
    • Privacy Policy

    Type above and press Enter to search. Press Esc to cancel.