publications

2026

  1. AutoformBench: A Lean Formalization Benchmark for Advanced Mathematics
    Charles Arnal, Ahmad Rammal, Niket Patel, and 6 more authors
    In NeurIPS MathAI Workshop, 2026
    PDF Code
  2. h1: A Taste of the Bitter Lesson in Automated Theorem Proving
    Skander Moalla, Ahmad Rammal, Niket Patel, and 2 more authors
    Preprint, 2026
    PDF
  3. Learning to Discover Interesting Mathematics
    Niket Patel, Ahmad Rammal, Amaury Hayat, and 2 more authors
    In NeurIPS MathAI Workshop, 2026
    Oral Presentation
    arXiv PDF
  4. Formalizing Mathematics at Scale
    Ahmad Rammal, Niket Patel, Fabian Gloeckle, and 5 more authors
    arXiv preprint, 2026
    arXiv Code

2025

  1. Task Priors: Enhancing Model Evaluation by Considering the Entire Space of Downstream Tasks
    Niket Patel, and Randall Balestriero
    In NeurIPS UniReps Workshop, 2025
    arXiv Code
  2. Layer by Layer: Uncovering Hidden Representations in Language Models
    Oscar Skean, Md Rifat Arefin, Dan Zhao, and 4 more authors
    In International Conference on Machine Learning (ICML), 2025
    Spotlight Paper, Oral Presentation
    arXiv Code
  3. On the Local Complexity of Linear Regions in Deep ReLU Networks
    Niket Patel, and Guido Montúfar
    In International Conference on Machine Learning (ICML), 2025
    arXiv Code

2024

  1. Learning to Compress: Local Rank and Information Compression in Deep Neural Networks
    Niket Patel, and Ravid Shwartz-Ziv
    In NeurIPS Workshop on Compression, 2024
    arXiv

2023

  1. Bridging the Gap: Addressing Discrepancies in Diffusion Model Training for Classifier-Free Guidance
    Niket Patel, Luis Salamanca, and Luis Barba
    In NeurIPS Workshop on Diffusion Models, 2023
    arXiv