Samy Jelassi

Parameter-Efficient Reinforcement Learning using Prefix Optimization

Let's (not) just put things in Context: Test-time Training for Long-context LLMs

To Backtrack or Not to Backtrack: When Sequential Search Limits Model Reasoning

Mixture of Parrots: Experts improve memorization more than reasoning

The Role of Sparsity for Length Generalization in LLMs

Echo Chamber: RL Post-training Amplifies Behaviors Learned in Pretraining

Let Me Think! A Long Chain of Thought Can Be Worth Exponentially Many Short Ones

Universal Length Generalization with Turing Programs

Universal length generalization with Turing Programs

Collective Model Intelligence Requires Compatible Specialization