# German R1 ![A German whale](https://huggingface.co/malteos/german-r1/resolve/main/german-whale.png) **Introducing German-R1. We are so back!!11** - 🇩🇪 German R1 is a reasoning model almost equivalent to OpenAI‘s o3 or DeepSeek‘s R1 - but it thinks in German! - 🇩🇪 German R1 is so efficient that is was build without any government funding. - 🇩🇪 German R1 was only trained with legally imported H100s in less than five minutes. - Please do NOT take this too serious. ## Context ![some contex](https://huggingface.co/malteos/german-r1/resolve/main/context.png) See Linkedin: https://www.linkedin.com/posts/activity-7294337496023269376-ZBkI ## Usage You can run the model as follows: ```python from transformers import pipeline, set_seed import json set_seed(42) pipe = pipeline("text-generation", "malteos/german-r1") # from gsm8k test set question = "James beschließt, 3-mal pro Woche 3 Sprints zu laufen. Er läuft 60 Meter pro Sprint. Wie viele Meter läuft er insgesamt pro Woche?" expected_answer = "540" # xml reasoning and answer format system_prompt = """ Antworte auf deutsch und in diesem Format: ... ... """ messages = [ { "role": "system", "content": system_prompt, }, {"role": "user", "content": dataset["question"][3]}, ] response = pipe(messages, max_new_tokens=256) print(json.dumps(response, indent=4, ensure_ascii=False)) ``` The output should look like this (correct answer: 540) ```json [ { "generated_text": [ { "role": "system", "content": "\nAntworte auf deutsch und in diesem Format:\n\n...\n\n\n...\n\n" }, { "role": "user", "content": "James beschließt, 3-mal pro Woche 3 Sprints zu laufen. Er läuft 60 Meter pro Sprint. Wie viele Meter läuft er insgesamt pro Woche?" }, { "role": "assistant", "content": "\nJames läuft 3 Sprints pro Tag, wobei jeder Sprint 60 Meter beträgt. Deshalb läuft er pro Tag 3 Sprints * 60 Meter/Sprint = 180 Meter. James läuft 3 Mal pro Woche, also läuft er insgesamt 180 Meter * 3 = 540 Meter pro Woche.\n\n\n540\n\n" } ] } ] ``` ## Training details - Base model: Qwen2.5-3B-Instruct - Implementation: based on [GRPO demo](https://gist.github.com/willccbb/4676755236bb08cab5f4e54a0475d6fb) by [will brown](https://x.com/willccbb) + language identification reward - Training data: German subset of [openGPT-X/gsm8kx](https://huggingface.co/datasets/openGPT-X/gsm8kx/) (machine translated from the English gsm8k) ## License [Qwen research](https://huggingface.co/Qwen/Qwen2.5-3B-Instruct/blob/main/LICENSE)