Poodle: Seamlessly Scaling Down Large Language Models with Just-in-Time Model Replacement

Authors

• Nils Strassenburg
• Boris Glavic
• Tilmann Rabl

Materials

bibtex

@techreport{SG25b,
  author = {Strassenburg, Nils and Glavic, Boris and Rabl, Tilmann},
  title = {Poodle: Seamlessly Scaling Down Large Language Models with Just-in-Time Model Replacement},
  institution = {CoRR},
  venueshort = {Techreport},
  year = {2025},
  url = {https://doi.org/10.48550/arXiv.2512.05525},
  pdfurl = {https://arxiv.org/pdf/2512.05525},
  doi = {10.48550/ARXIV.2512.05525},
  eprinttype = {arXiv},
  eprint = {2512.05525},
  timestamp = {Sun, 25 May 2025 20:50:31 +0200},
  keywords = {Provenance; Data Skipping; Relevance-based Data Management},
  projects = {Relevance-based Data Management},
  biburl = {https://dblp.org/rec/journals/corr/abs-2512-05525.bib},
  bibsource = {dblp computer science bibliography, https://dblp.org}
}

Reference

Poodle: Seamlessly Scaling Down Large Language Models with Just-in-Time Model Replacement Nils Strassenburg, Boris Glavic and Tilmann Rabl CoRR.