eran malach

Mixture of Parrots: Experts improve memorization more than reasoning

DON’T STOP ME NOW: EMBEDDING BASED SCHEDULING FOR LLMS

A New Perspective on Shampoo's Preconditioner

Universal length generalization with Turing Programs

MLPs for NLP: Towards Discovering Inductive Bias From Scratch

Transcendence: Generative Models Can Outperform The Experts That Train Them

Auto-Regressive Next-Token Predictors are Universal Learners

On the Power of Decision Trees in Auto-Regressive Language Modeling

The Evolution of Statistical Induction Heads: In-Context Learning Markov Chains

Implicit Intermediate Supervision for Learning Complex Functions

Corgi$^2$: A Hybrid Offline-Online Approach To Storage-Aware Data Shuffling For SGD

Less is More: Selective Layer Finetuning with SubTuning