NeurIPS 2025 Explorer
Concepts
Authors
Glossary
johnsanterre.github.io
vocabulary size
4 papers
Discrete Diffusion Models: Novel Analysis and New Sampler Guarantees
Explaining and Mitigating Crosslingual Tokenizer Inequities
Scaling Laws for Gradient Descent and Sign Descent for Linear Bigram Models under Zipf’s Law
UniTok: a Unified Tokenizer for Visual Generation and Understanding