CoCo

Headlines · 2

1. 3d
   JEPA Paper Proposes a Platonic Representation Theorem

   Randall Balestriero and Yann LeCun submitted “Proving and Improving the Platonic Representation Hypothesis with JEPAs” to arXiv on October 29, 2026. The authors say their theorem applies to separately trained JEPA encoders whose positive pairs are two observations of the same scene, with misalignment bounded otherwise, and that the encoders learn equivalent representations up to one rotation. They report experiments ranging from 145,000 captions to text associated with 940 million web images. Their reported results include requiring 42 times fewer training samples than next-token prediction for the same alignment, 64% top-5 accuracy on ImageNet-100 for a 109-million-parameter text JEPA, and a 0.838 CIDEr score on COCO after applying one rotation.

   Hugging Face · Papers · 🔥 0
2. 3d
   Snowflake Adds Governance Controls for Enterprise AI Agents

   Snowflake introduced controls for identifying agents, capping their permissions, masking sensitive values, limiting AI credit consumption, and auditing their activity, according to InfoQ 中文 AI&大模型. Its agent-restricted session scopes and per-user AI credit quotas are generally available, while CoCo automation remains in public preview and runs unattended in a Snowflake-managed sandbox. The report notes that audit views can only reveal activity by agents Snowflake has recognized and recorded.

   InfoQ 中文 AI&大模型 · 🔥 0

Experience and discussion from the community

Share my CoCo experienceAsk about CoCo