Small-scale replication: good views beat next-token prediction

A researcher replicated Balestriero and LeCun's “good views” paper at laptop scale, using 10,000 COCO photos, human captions and 11M-parameter encoders. Good views scored 0.320 versus 0.192 for bad views and 0.148 for next-token prediction; scale limits remain.

• ORTUS AI open-sources 360-degree image rotation model RightWayUp2026-09-30
• Shelf audit: YOLO detects products, but embeddings confuse sibling SKUs2026-09-28
• Commentary: AI Has No Intent and No Motivation2026-09-24