World Models with V-JEPA 2: prediction in representation space

Welcome to AI Build & Learn, a weekly AI engineering stream where we pick a new topic and learn by building together.

​This event is about world models with V-JEPA 2, Meta's self-supervised video model that learns by predicting in representation space instead of generating pixels. We'll load the model, hide parts of a video clip, watch it predict the missing pieces as embeddings rather than images, and measure how good those predictions actually are.

​V-JEPA 2 is trained on internet-scale video with no labels, and it has no decoder at all: the predictor emits vectors, not frames. That constraint makes it a great topic to build around, because "show me what it predicted" stops being a screenshot and starts being a measurement. We'll explore how the model represents video, how to probe those representations with a single frozen linear layer, why cosine similarity can quietly mislead you, and where the pretrained predictor stops behaving like a world model.

​Depending on where people want to go, we can also compare it against pixel-space world models like NVIDIA Cosmos and latent world models like DreamerV3, and look at how JEPA-style representations get used for robotics, planning, and perception.

​Some things to look up to get started:

• ​V-JEPA 2 (Meta): https://github.com/facebookresearch/vjepa2
• ​V-JEPA 2 in Transformers: https://huggingface.co/docs/transformers/model_doc/vjepa2
• ​V-JEPA 2 paper: https://huggingface.co/papers/2506.09985
• ​Meta AI research overview: https://ai.meta.com/research/vjepa/

​​​Resources

• ​​​GitHub: https://github.com/sagecodes/ai-build-and-learn
• ​​​Events Calendar: https://luma.com/ai-builders-and-learners
• ​​​Slack (Discuss during the week): https://slack.flyte.org/
• ​​​Hosted by Sage Elliott: https://www.linkedin.com/in/sageelliott/

​​In this stream

• Intro to topic
• ​​​​Community Discussion
• Practical examples

​​​Community challenge (optional)
​​​Try spending 30–90 minutes during the week learning or building something related to the topic, then share what you’re working on in Slack.

​​​Note on Flyte / Union
​​​You may see Flyte used in some demos. Flyte is an open-source AI orchestration platform maintained by Union (where I work) for building scalable, durable, and observable AI workflows. You do not need to use Flyte to participate.

• ​​​Union: https://www.union.ai/
• ​​​Flyte: https://flyte.org/

​​​Drop a comment with ideas for future topics (agents, RAG, MLOps, robotics, frameworks, and more).