Polish Medical Leaderboard

Leaderboard was created as part of an open-science project SpeakLeash.org

The leaderboard evaluates language models on Polish Board Certification Examinations (Państwowy Egzamin Specjalizacyjny) from years 2018-2022.

Average columns are normalized against scores by "Baseline (majority class)".

* `,chat` suffix means that a model is tested using chat templates
* `,chat,multiturn` suffix means that a model is tested using chat templates and fewshot examples are treated as a multi-turn conversation
* 🚧 prefix means that not all tasks were calculated and the average scores are not accurate

We gratefully acknowledge Polish high-performance computing infrastructure PLGrid (HPC Centers: ACK Cyfronet AGH) for providing computer facilities and support within computational grant no. PLG/2024/016951.

Model types

Model sizes (billions of parameters)

N-shot

Show private/deleted models

Select columns to show

Generated 2026-09-19 22:16 UTC

The benchmark uses datasets:
* [speakleash/PES-2018-2022](https://huggingface.co/datasets/speakleash/PES-2018-2022), which is based on [amu-cai/PES-2018-2022](https://huggingface.co/datasets/amu-cai/PES-2018-2022) ([J. Pokrywka, J. Kaczmarek, E. Gorzelańczyk, GPT-4 passes most of the 297 written Polish Board Certification Examinations, 2024](https://arxiv.org/abs/2405.01589)).

## Do you want to add your model to the leaderboard?

Contact with me: [LinkedIn](https://www.linkedin.com/in/wrobelkrzysztof/)

or join our [Discord SpeakLeash](https://discord.gg/FfYp4V6y3R)

## TODO

* fix long model names
* add inference time
* add more tasks
* fix scrolling on Firefox

## Tasks

Tasks taken into account while calculating averages:
* Average: polish_pes_medycyna_rodzinna, polish_pes_pediatria, polish_pes_alergologia, polish_pes_anestezjologia, polish_pes_angiologia, polish_pes_balneologia_i_medycyna_fizykalna, polish_pes_chirurgia_dziecieca, polish_pes_chirurgia_naczyniowa, polish_pes_chirurgia_ogolna, polish_pes_chirurgia_onkologiczna, polish_pes_chirurgia_stomatologiczna, polish_pes_chirurgia_szczekowo-twarzowa, polish_pes_choroby_pluc, polish_pes_choroby_pluc_dzieci, polish_pes_choroby_wewnetrzne, polish_pes_choroby_zakazne, polish_pes_dermatologia_i_wenerologia, polish_pes_diabetologia, polish_pes_endokrynologia, polish_pes_endokrynologia_ginekologiczna_i_rozrodczosc, polish_pes_endokrynologia_i_diabetologia_dziecieca, polish_pes_gastroenterologia, polish_pes_gastroenterologia_dziecieca, polish_pes_geriatria, polish_pes_ginekologia_onkologiczna, polish_pes_hematologia, polish_pes_hipertensjologia, polish_pes_kardiochirurgia, polish_pes_kardiologia, polish_pes_medycyna_pracy, polish_pes_medycyna_paliatywna, polish_pes_medycyna_ratunkowa, polish_pes_medycyna_sportowa, polish_pes_nefrologia, polish_pes_neonatologia, polish_pes_neurochirurgia, polish_pes_neurologia, polish_pes_neurologia_dziecieca, polish_pes_okulistyka, polish_pes_onkologia_kliniczna, polish_pes_ortodoncja, polish_pes_ortopedia, polish_pes_otolaryngologia, polish_pes_patomorfologia, polish_pes_perinatologia, polish_pes_periodontologia, polish_pes_poloznictwo_i_ginekologia, polish_pes_protetyka_stomatologiczna, polish_pes_psychiatria, polish_pes_psychiatria_dzieci_i_mlodziezy, polish_pes_radiologia_i_diagnostyka_obrazowa, polish_pes_radioterapia_onkologiczna, polish_pes_rehabilitacja_medyczna, polish_pes_reumatologia, polish_pes_stomatologia_dziecieca, polish_pes_stomatologia_zachowawcza, polish_pes_transplantologia_kliniczna


## Reproducibility
To reproduce our results, you need to clone the repository:

```
git clone https://github.com/speakleash/lm-evaluation-harness.git -b polish4
cd lm-evaluation-harness
pip install -e .
```

and run benchmark for 0-shot and 5-shot:

```
lm_eval --model hf --model_args pretrained=speakleash/Bielik-7B-Instruct-v0.1 --tasks polish_pes --num_fewshot 0 --output_path results/ --log_samples
lm_eval --model hf --model_args pretrained=speakleash/Bielik-7B-Instruct-v0.1 --tasks polish_pes --num_fewshot 5 --output_path results/ --log_samples
```

With chat templates:
```
lm_eval --model hf --model_args pretrained=speakleash/Bielik-7B-Instruct-v0.1 --tasks polish_pes --num_fewshot 0 --output_path results/ --log_samples --apply_chat_template
lm_eval --model hf --model_args pretrained=speakleash/Bielik-7B-Instruct-v0.1 --tasks polish_pes --num_fewshot 5 --output_path results/ --log_samples --apply_chat_template
```

## List of Polish models

* speakleash/Bielik-11B-v2.2-Instruct
* speakleash/Bielik-11B-v2.1-Instruct
* speakleash/Bielik-11B-v2.0-Instruct
* speakleash/Bielik-7B-Instruct-v0.1

### List of multilingual models

* Meta-Llama-3.1-405B-Instruct-FP8
* Mistral-Large-Instruct-2407
* Meta-Llama-3.1-70B-Instruct
* Qwen2-72B-Instruct
* Meta-Llama-3-70B-Instruct
* glm-4-9b-chat
* Meta-Llama-3.1-8B-Instruct
* Mistral-Nemo-Instruct-2407-PL-finetuned
* Mistral-Nemo-Instruct-2407
* Qwen2-7B-Instruct
* Meditron3-8B
* SOLAR-10.7B-Instruct-v1.0
* openchat-3.5-0106-gemma
* Starling-LM-7B-beta
* Mistral-7B-Instruct-v0.3
* Phi-3.5-mini-instruct

📙 Citation

@misc{polish-medical-llm-leaderboard,
	title        = {Polish Medical Leaderboard},
	author       = {Wróbel, Krzysztof and {SpeakLeash Team} and {Cyfronet Team}},
	year         = 2024,
	publisher    = {Hugging Face},
	howpublished = "\url{https://huggingface.co/spaces/speakleash/polish_medical_leaderboard}"
}

@misc{eval-harness,
  author       = {Gao, Leo and Tow, Jonathan and Abbasi, Baber and Biderman, Stella and Black, Sid and DiPofi, Anthony and Foster, Charles and Golding, Laurence and Hsu, Jeffrey and Le Noac'h, Alain and Li, Haonan and McDonell, Kyle and Muennighoff, Niklas and Ociepa, Chris and Phang, Jason and Reynolds, Laria and Schoelkopf, Hailey and Skowron, Aviya and Sutawika, Lintang and Tang, Eric and Thite, Anish and Wang, Ben and Wang, Kevin and Zou, Andy},
  title        = {A framework for few-shot language model evaluation},
  month        = 07,
  year         = 2024,
  publisher    = {Zenodo},
  version      = {v0.4.3},
  doi          = {10.5281/zenodo.12608602},
  url          = {https://zenodo.org/records/12608602}
}

@misc{fwpes2024,
  title={speakleash/PES-2018-2022},
  author={Maria, Filipkowska and Krzyszof, Wróbel},
  year={2024},
  howpublished={\url{https://huggingface.co/datasets/speakleash/PES-2018-2022}},
}

@misc{pkgpes2024,
  title={amu-cai/PES-2018-2022},
  author={Jakub Pokrywka and Jeremi Kaczmarek and Edward Gorzelańczyk},
  year={2024},
  howpublished={\url{https://huggingface.co/datasets/amu-cai/PES-2018-2022}},
}

@misc{pokrywka2024gpt4,
      title={GPT-4 passes most of the 297 written Polish Board Certification Examinations}, 
      author={Jakub Pokrywka and Jeremi Kaczmarek and Edward Gorzelańczyk},
      year={2024},
      eprint={2405.01589},
      archivePrefix={arXiv},
      primaryClass={id='cs.CL' full_name='Computation and Language' is_active=True alt_name='cmp-lg' in_archive='cs' is_general=False description='Covers natural language processing. Roughly includes material in ACM Subject Class I.2.7. Note that work on artificial languages (programming languages, logics, formal systems) that does not explicitly address natural-language issues broadly construed (natural-language processing, computational linguistics, speech, text retrieval, etc.) is not appropriate for this area.'}
}

@misc{open-pl-llm-leaderboard,
	title        = {Open PL LLM Leaderboard},
	author       = {Wróbel, Krzysztof and {SpeakLeash Team} and {Cyfronet Team}},
	year         = 2024,
	publisher    = {Hugging Face},
	howpublished = "\url{https://huggingface.co/spaces/speakleash/open_pl_llm_leaderboard}"
}

@misc{open-llm-leaderboard-v1,
  author = {Edward Beeching and Clémentine Fourrier and Nathan Habib and Sheon Han and Nathan Lambert and Nazneen Rajani and Omar Sanseviero and Lewis Tunstall and Thomas Wolf},
  title = {Open LLM Leaderboard (2023-2024)},
  year = {2023},
  publisher = {Hugging Face},
  howpublished = "\url{https://huggingface.co/spaces/open-llm-leaderboard-old/open_llm_leaderboard}"
}