Mechanica Labs

透明地解决复杂问题

发表论文

  1. Uncovering a Universal Abstract Algorithm for Modular Addition in Neural Networks NeurIPS 2025 (2025)
    📄
  2. Interpreting deep neural networks trained on elementary p groups reveals algorithmic structure Topology, Algebra, and Geometry in Data Science (2025)
    📄
  3. The Geometry and Topology of Modular Addition Representations Topology, Algebra, and Geometry in Data Science (2025)
    📄
  4. The Representations of Deep Neural Networks Trained on Dihedral Group Multiplication NeurIPS 2025 Workshop on Symmetry and Geometry in Neural Representations (2025)
    📄