Kaons

Research in learning, language, and computation.

↓
  1. 1.

    LLM Reasoning

    Investigating failure modes in multi-turn constraint satisfaction and logical consistency across large language models.

  2. 2.

    Operator Learning Theory

    Finite-sample generalization bounds for neural operators under distribution shift.

  3. 3.

    Tokenization

    Studying whether relative representations can enable cross-tokenizer transfer in language models.

  4. 4.

    Connectomics

    Scalable methods for comparative analysis of neural connectivity graphs.

  5. 5.

    NLP Evaluation

    Building robust evaluation systems for detecting evasion and measuring semantic similarity in political and narrative text.