Math question -> Codex generates program -> execute program -> solution! Amazing... x.com/HochreiterSepp…

.Jonathan Lorraine et al conduct an in-depth analysis of optimizers and propose a new one that incorporates a "branching" mechanic. x.com/jonLorraine9/s…

Grokking is now on ArXiv! x.com/ak92501/status…

Put your documents in a new light :) x.com/austinvhuang/s…

Put your documents in a new light :) x.com/austinvhuang/s…

.Stella Biderman evaluates how well GPT-J plagiarizes by investigating MOSS' "plagiarism score" and finds that it outperforms humans. x.com/BlancheMinerva…

.Robin Tibor Schirrmeister et al. propose a new method that "simplifies" images, allowing better dataset condensation and debugging and failures! x.com/robintibor/sta…

Multi-modality continues to advance with CM3! x.com/ArmenAgha/stat…

Controllable symbolic music generation! x.com/ak92501/status…

Hierarchical VAEs continue to improve. x.com/RayhaneMama/st…

A new approach to tabular learning: Bayesian approximation + Transformers + Causality = SOTA performance in < 1 second x.com/FrankRHutter/s…

A simple conceptual framework for improving, understanding and deriving self-supervised learning methods! x.com/yanndubs/statu…

A simple conceptual framework for improving, understanding and deriving self-supervised learning methods! x.com/yanndubs/statu…