Cite this work

NaviMed-UMB is archived on Zenodo with a versioned DOI chain. It ships a CITATION.cff for GitHub's "Cite this repository" button.

DOIs

ScopeDOI
Concept (always resolves to latest version)10.5281/zenodo.19851346
Version v0.5.210.5281/zenodo.24206670
Version v0.5.110.5281/zenodo.22070559
Version v0.5.010.5281/zenodo.21207763
Version v0.4.010.5281/zenodo.20364953
Version v0.3.010.5281/zenodo.20317011

For most purposes, cite the concept DOI. It auto-resolves to the latest release. Cite a specific version DOI when reproducibility against an exact snapshot matters.

Canonical citation

From CITATION.cff (CFF 1.2.0):

Minarowski, Ł. (2026). NaviMed-UMB: hardware envelope studies for local AI deployment on consumer RDNA 4 GPUs (Version 0.5.1) [Software]. Zenodo. https://doi.org/10.5281/zenodo.19851346

BibTeX

@software{minarowski_navimed_umb_2026,
  author    = {Minarowski, {\L}ukasz},
  title     = {{NaviMed-UMB: hardware envelope studies for local
               AI deployment on consumer RDNA 4 GPUs}},
  year      = {2026},
  version   = {0.5.1},
  publisher = {Zenodo},
  doi       = {10.5281/zenodo.19851346},
  url       = {https://github.com/kicrazom/navimed-umb},
  license   = {CC-BY-4.0}
}

Author

Łukasz Minarowski, MD, PhD
Department of Respiratory Physiopathology, Medical University of Białystok, Poland
ORCID: 0000-0002-2536-3508
Email: lukasz.minarowski@umb.edu.pl

Licensing

The repository uses dual licensing:

  • Code: MIT.
  • Documentation, methodology, lab logs, benchmark findings: CC BY 4.0.
  • The calibration dataset: derived from third-party EMA-published regulatory documents (SmPC). It is governed by its own license, not the root CC-BY-4.0 / MIT.

Released model checkpoints carry the license of their base family. llama3.1 for the Llama-PLLuM derivatives, apache-2.0 for the Mistral-Nemo-derived 12B. See Models.

AI-usage disclosure

Transparency about AI assistance is a default policy for this project. Cloud language models (Anthropic Claude, OpenAI GPT, Google Gemini) were used for documentation, code, analysis, debugging and review, and open-weight models served locally on the R9700 workstation for Polish-language proofreading. Claude Code, an agentic command-line tool, runs on the workstation itself: it edits files, runs commands and, in some sessions, launched author-approved scripts and benchmark runs, always under the author's supervision.

All experimental design, hardware configuration, empirical measurements, and scientific claims are the sole responsibility of the author. No AI tool generated, estimated or interpolated a measurement: every figure is read from an instrument's raw log. The work has a single author. Consistent with COPE and ICMJE guidance, AI tools are documented as research aids, not co-authors. Which tools, for what and when: AI_USAGE_DISCLOSURE.md; per-change attribution: Co-Authored-By trailers in the Git history.

Inline disclosure for citing work. "This work used generative-AI assistants (Anthropic Claude via claude.ai and the Claude Code agentic command-line tool; OpenAI GPT and Google Gemini via web apps) for documentation, code, analysis, debugging and review. Claude Code operated on the measurement workstation under the author's supervision and, in some sessions, launched author-approved scripts and benchmark runs; it did not generate any measurement. All experimental design, hardware configuration, empirical results and scientific claims are the sole responsibility of the author. Full disclosure: AI_USAGE_DISCLOSURE.md (DOI: 10.5281/zenodo.19851346)."

Acknowledgements

Quantization compute for the 70B family ran on an AMD Instinct MI300X through the AMD Developer Cloud, powered by DigitalOcean, provided as a promotional credit under the AMD Developer Program. The base models come from the PLLuM consortium (SpeakLeash, OPI-PIB, NASK, Wrocław University of Science and Technology) and the Polish Ministry of Digital Affairs, published via CYFRAGOVPL. The calibration corpus derives from SmPC text published by the European Medicines Agency.