Logs

Peeking into AIML@K

*
All AI4Math Math4AI LM AI4Sci Conj Diary
Embedding of numbers

Why LLMs make silly math mistakes? This may be the biggest cause.

Sep 30, 2025
Linear Separability of NN-induced Embeddings

A particular case of implicit margin maximization by cross-entropy loss in supervised learning.

Sep 30, 2025
ICML 2025
ICML 2025, Vancouver. Three students of mine - Jaeheun, Jaehyuk, and Taehun - came along. It would have been much better had all contributors were here with me. Last NeurIPS was at the same place.
Jul 20, 2025
ICML 2025
들어가며
Gaussian process, Bayesian optimization 그리고 Knowledge gradient에 대하여 작성합니다. 내용 테스
Sep 29, 2024