Jan Ebert
Jan Ebert has 2 papers in our library, in 2024–2026, most of them on Language models & pretraining, Multilingual & cross-lingual NLP, Low-resource languages; often with Annika Simonsen, Hafsteinn Einarsson, Johannes Leveling.
current affiliationTechnische Universität Dresden, Dresden, Germany
OpenAlex · ORCID · h-index 4 · 33 works · 115 citations OpenAlex, 2026-10-14
2 works in our library · 72 citations in our library · ⭐ Paper Star rank #4138
Affiliations
🎓Technische Universität Dresden
Dresden · Germany2009–2026 · 10 papers🔬Forschungszentrum Jülich
Jülich · Germany2021–2025 · 9 papers🔬Jülich Supercomputing Centre
Jülich · Germany2021–2024 · 5 papers
Topics
Works by year
2020s
Papers
- Tokenizer Choice For LLM Training: Negligible or Crucial? (2024 · 102 citations)
Tokenizers are rarely examined when building LLMs. This paper asks whether the choice of tokenizer measurably affects decoder-only model quality and training cost, in both English-only and multilingual settings. The authors built two training corpora of 70B words each, with matched domain mixtures (80% web-crawled… - Synthetic Instruction Generation for Low-Resource Nordic Languages: Viability and Limitations in LLM Instruction-Tuning (2026)
Instruction-tuning needs many instruction–response pairs, but having native speakers write instructions is expensive for smaller languages.
Co-authors here
Cited by
This page in Markdown → https://md.research.elamur.ai/free/en/authors/jan-ebert/