Fable 5.1 hand-writes 100k-param classifier weights, beating backprop training
AccBalanced · x · 2026-11-04
X user @jacsebl demonstrated that Fable 5.1 hand-wrote the individual weights of a 100k-parameter neural network to distinguish real from fake legal citations — essentially one-shotting a full NN that outperforms a backprop-trained counterpart.
@danielmac8, resharing the demo, called weight-space programming the next inference-time reasoning breakthrough, arguing it will change the world. Note this is a single-user demo and personal claim, not independently verified.
Related event: Fable 5.1 Hand-Writes 100K-Parameter Network, Beating Trained Baseline(3 posts)→
More from Models
- Tencent Hunyuan ships Hy4 with vision, touting top-tier open-model status — mishig25 · 2026-11-04
- Dev's 4 days with Codex: tool admits 16 failures, Claude had to finish the job — v0id_user · 2026-11-04
- shadcn: Opus's 'taste' can't be benchmarked overnight, it takes real product building — shadcn · 2026-11-04
- Users say Claude's Ultrafast mode burns limits 8x faster — a 'flex' speed tier would make more sense — Lucky_Creme_5208 · 2026-11-04
- OpenAI quietly reverses its reverse engineering ban, diverging from Anthropic's stricter stance — AccBalanced · 2026-11-04
- Built with Claude Haiku 5.5, not Opus: pixel-art animation shows small models often suffice — Lazar988 · 2026-11-04