Martin Pawelczyk

Papers that credit Martin Pawelczyk in their source bylines. Matching names may refer to different researchers.

1 results

1. PaperMachine Learning

   Vertex sampling helps mix many language-model data domains

   

   LinMix uses varied proxy data mixtures to estimate domain value; its linear fit holds up at 288 domains, where tuned Dirichlet fits struggle.

   Added to Chiosa Oct 28, 2026