Konstantin Dobler

AweDist: Attention-aware Embedding Distillation for New Input Token Embeddings

arXiv 2025

I Don't Know: Explicit Modeling of Uncertainty with an [IDK] Token

arXiv 2024

Language Adaptation on a Tight Academic Compute Budget: Tokenizer Swapping Works and Pure bfloat16 Is Enough

arXiv 2024

FOCUS: Effective Embedding Initialization for Monolingual Specialization of Multilingual Models

arXiv 2023

Efficient Parallelization Layouts for Large-Scale Distributed Model Training

arXiv 2023

Art Creation with Multi-Conditional StyleGANs

arXiv 2022

Gerard de Melo

Alejandro Sierra-Múnera

Desmond Elliott

Eden Biran

Florian Hübscher

Jan Westphal

Johannes Hagemann

co-founder / CTO

Maximilian Schall

Ralf Krestel

Roi Cohen