15,000 submissions to One Layer Deeper, and nobody solved the problem as intended

marksaroufim · x · 2026-09-13

marksaroufim reports that the One Layer Deeper challenge drew over 15,000 submissions, yet nobody solved the problem as intended. He highlights some interesting ideas from the community around learning reusable operations and composing them into deeper computations, concluding that "deep learning is still not solved." coreauto's repost echoes that the community showed real creativity but there is much more work to do.

Related event: One Layer Deeper challenge gets 15,000+ submissions, none as intended(3 posts)→

Original post →

More from Research

Research channel →