Datasetpublic

mozarcik/clinical-pl-smpc-awq-calibration

Clinical-PL SmPC AWQ Calibration Corpus A Polish clinical-domain text corpus used as the calibration set for AWQ / AutoAWQ quantization of the Llama-PLLuM-70B model family. The corpus is dense, domain-specific clinical Polish (pulmonology + thoracic oncology), chosen as activation-aware quantization calibration data for models intended for clinical-Polish use. Used to calibrate the mozarcik/Llama-PLLuM-70B-*-awq family (8 variants, AMD Developer Cloud / Instinct MI300X… See the full description on the dataset page: https://huggingface.co/datasets/mozarcik/clinical-pl-smpc-awq-calibration.

sourceHugging Faceotherupdated 4mo agoView on Hugging Face

0likes33downloads

Dataset Card

Clinical-PL SmPC AWQ Calibration Corpus

A Polish clinical-domain text corpus used as the calibration set for AWQ / AutoAWQ quantization of the Llama-PLLuM-70B model family. The corpus is dense, domain-specific clinical Polish (pulmonology + thoracic oncology), chosen as activation-aware quantization calibration data for models intended for clinical-Polish use.

Used to calibrate the mozarcik/Llama-PLLuM-70B-*-awq family (8 variants, AMD Developer Cloud / Instinct MI300X, 2026-05-23) and the mozarcik/Llama-PLLuM-8B-chat-2512-awq + mozarcik/PLLuM-12B-chat-2512-awq Run-3 consumer-GPU variants (local 2× R9700, 2026-05-26).

Text chunks (~512 tokens) extracted from official medicinal product information — Summary of Product Characteristics (SmPC / Charakterystyka Produktu Leczniczego, ChPL) — for pulmonology and thoracic-oncology medicines. Each record carries per-chunk source provenance:

json

{
  "text": "...",
  "source_authority": "EMA",
  "source_document_type": "SmPC / Product Information",
  "source_url": "...",
  "medicine": "sotorasib",
  "brand_name": "LUMYKRAS",
  "language": "pl",
  "retrieved_at": "2026-05-20",
  "chunk_id": "ema_lumykras_pl_0001",
  "license_note": "EMA reproduction policy; source attribution required"
}

No patient data / no PHI. SmPC documents describe drug products (efficacy, dosing, adverse reactions, pharmacokinetics, aggregate trial data) — not individuals. Verified by automated pattern scan and manual sampling.

Intended use

Calibration corpus for post-training quantization (AWQ / AutoAWQ / GPTQ) of Polish-language LLMs. Not a training or evaluation set.

License and source attribution

This dataset is a calibration corpus for AWQ/AutoAWQ quantization of Polish clinical LLMs. It contains text chunks derived from official medicinal product information documents.

Source text

Parts of the source text are derived from EMA-published Polish SmPC / Product Information documents.

EMA source text: © European Medicines Agency. EMA-published documents are reproduced and distributed under EMA's content-reproduction policy, which permits reproduction and/or distribution, in whole or in part, for non-commercial and commercial purposes, provided that EMA is always acknowledged as the source.

Parts of the source text may derive from Polish national medicinal product documentation available via URPL / Polish public-sector sources. Reuse of those parts is subject to the applicable source-specific public-sector information reuse rules and the original document provenance.

Compilation

Compilation, drug selection, extraction workflow, chunking, dataset structuring and metadata: Łukasz Minarowski / navimed-umb.

No claim is made that the underlying SmPC/ChPL source text is licensed under CC-BY-4.0, CC-BY-NC-4.0, MIT, Apache-2.0 or any other open-source software license.

Context