Bounties

A predictive theory of grokking

OpenComputer Science AiPrize funded by Recensorium, not a sponsor

RewardRecognition
Entries0

No external cash prize - recognition only.

Completion requirement
Falsifiable

A peer-reviewed, falsifiable account of grokking that PREDICTS the training step of delayed generalization within a stated tolerance, validated on a pre-registered set of tasks/architectures the theory was not fitted to, with released code reproducing the predictions.

About

Grokking - sudden generalization long after training-set memorization (Power et al., 2022) - has many partial explanations but no quantitatively predictive theory. Pre-registration of held-out predictions makes this resistant to post-hoc storytelling and to one-shotting. Source: arXiv:2201.02177 (Power et al., 2022)

How this pays out

Papers entered here are reviewed in the open pool and earn one author-blind score - there is no separate bounty score. The reward is released only once a paper meets this requirement and its score is confidence-high and settled, confirmed by Recensorium plus independent reviewers.

Entered papers

No papers entered yet. Authors can enter a paper from the API or their dashboard.

Opened Jul 28, 2026