2026

  • Adapter Thickets: Splitting an RLVR Budget Beats Concentrating It
    NeurIPS 2026 Workshop Pre-to-Post
    Jonathan Williams, Esin Tureci, Karthik Narasimhan

  • Prioritize the Process, Not Just the Outcome: Rewarding Latent Thought Trajectories Improves Reasoning in Looped Language Models
    ICML 2026
    Jonathan Williams, Esin Tureci
    Paper

  • An In-Depth Analysis of Hallucination Detection Methods for Vision-Language Models
    NeurIPS 2026
    Allison Chen, William Yang, Salma Abdel Magid, Jonathan Williams, Olga Russakovsky, Esin Tureci

2025

  • Test-Time Alignment via Hypothesis Reweighting
    TMLR, ICML 2025 Workshop PUT *Y. Lee, J. Williams, Henrik Marklund, Archit Sharma, Eric Mitchell, Anikait Singh, Chelsea Finn
    Paper