The German Commons - 154 Billion Tokens of Openly Licensed Text for German Language Models
Paper
•
2510.13996
•
Published
•
6
AI & ML interests
A Tour through the 🤗 Hub • Build and Host ML Demos with Gradio & 🤗 • Getting Started with Transformers
Recent Activity
View all activity
christopher
authored
a
paper
22 days ago
BramVanroy
posted
an
update
30 days ago
Post
259
What are currently the best multilingual models with at most 72B parameters? Are Llama 3.3 70B and Qwen 2.5 72B still king?
- 1 reply
christopher
posted
an
update
about 1 month ago
BramVanroy
posted
an
update
3 months ago
Post
738
Thanks to popular request, I've just added two subsets to the CommonCrawl-Creative Commons Corpus (C5;
BramVanroy/CommonCrawl-CreativeCommons) so that you do not have to do filtering manually
- C5f ( BramVanroy/CommonCrawl-CreativeCommons-fine): only retains high-quality samples that are also present in FineWeb or FineWeb-2;
- C5r (https://huggingface.co/datasets/BramVanroy/CommonCrawl-CreativeCommons-recommended): additional strict filtering that removes samples with license disagreement, non-commercial licenses, and Wikipedia samples. The latter because you should probably get those from a more reliable source that provides better parsed content.
It goes without saying that these filters lead to a massive reduction in quantity. Doc and token counts are given on the dataset pages.
- C5f ( BramVanroy/CommonCrawl-CreativeCommons-fine): only retains high-quality samples that are also present in FineWeb or FineWeb-2;
- C5r (https://huggingface.co/datasets/BramVanroy/CommonCrawl-CreativeCommons-recommended): additional strict filtering that removes samples with license disagreement, non-commercial licenses, and Wikipedia samples. The latter because you should probably get those from a more reliable source that provides better parsed content.
It goes without saying that these filters lead to a massive reduction in quantity. Doc and token counts are given on the dataset pages.
lewtun
authored
a
paper
6 months ago
Post
936
Aspects of consciousness by Murray Shanahan:
- Awareness of the world (Perception)
- Self-awareness (own body, where it is in the space)
- Imagination / Stream of Consciousness (William James)
- Metacognition (Ability to think what we know)
- Emotions (Feel/Suffer, Sentient)
Source:
https://www.youtube.com/watch?v=v1Py_hWcmkU
- Awareness of the world (Perception)
- Self-awareness (own body, where it is in the space)
- Imagination / Stream of Consciousness (William James)
- Metacognition (Ability to think what we know)
- Emotions (Feel/Suffer, Sentient)
Source:
https://www.youtube.com/watch?v=v1Py_hWcmkU
BramVanroy
posted
an
update
6 months ago
Post
3613
📢💾 Introducing the Common Crawl Creative Commons Corpus (C5)!
C5 is a large-scale effort to heavily filter web-crawled data, as collected by the non-profit Common Crawl, to only documents that are Creative Commons-licensed such as cc-by-4.0 or public domain cc0. At this stage 150 billion tokens have been collected.
---
📄 data: BramVanroy/CommonCrawl-CreativeCommons
🧰 software: https://github.com/BramVanroy/CommonCrawl-CreativeCommons
---
</> To build C5, HTML pages are scrutinized and all links (if any) to CC licenses are collected, both in regular hyperlinks as well as in metadata. Additional data fields are included such as "was the license found in the
🌐 In this first version of C5, 8 languages are included (Afrikaans, German, English, French, Frysian, Italian, Dutch and Spanish). The language set was limited for two reasons: computational and storage limitations, and a collaboration with GPT-NL, which requested CC data for these languages to train a Dutch-focused, copyright-conscious LLM. In total, this V1 release contains almost 150 thousand documents and 150 billion tokens. This data was not filtered on quality nor deduplicated so that you can decide for yourself how much data to keep. To give some quality indication, a dataset field is present to describe whether a document is included in the FineWeb(-2) datasets, which are of high quality.
🔍 More work needs to be done! Only 7 out of 100+ Common Crawl crawls have been processed so far. That's encouraging because it means there is a lot more Creative Commons data to be collected! But to get there I need help in terms of compute. The current processing was already heavily sponsored by the Flemish Supercomputer but more is needed. If you have the compute available and which to collaborate in an open and transparent manner, please get in touch!
C5 is a large-scale effort to heavily filter web-crawled data, as collected by the non-profit Common Crawl, to only documents that are Creative Commons-licensed such as cc-by-4.0 or public domain cc0. At this stage 150 billion tokens have been collected.
---
📄 data: BramVanroy/CommonCrawl-CreativeCommons
🧰 software: https://github.com/BramVanroy/CommonCrawl-CreativeCommons
---
</> To build C5, HTML pages are scrutinized and all links (if any) to CC licenses are collected, both in regular hyperlinks as well as in metadata. Additional data fields are included such as "was the license found in the
head?" or "if multiple licenses were found, do they contradict each other?", which makes further filtering a breeze. 🌐 In this first version of C5, 8 languages are included (Afrikaans, German, English, French, Frysian, Italian, Dutch and Spanish). The language set was limited for two reasons: computational and storage limitations, and a collaboration with GPT-NL, which requested CC data for these languages to train a Dutch-focused, copyright-conscious LLM. In total, this V1 release contains almost 150 thousand documents and 150 billion tokens. This data was not filtered on quality nor deduplicated so that you can decide for yourself how much data to keep. To give some quality indication, a dataset field is present to describe whether a document is included in the FineWeb(-2) datasets, which are of high quality.
🔍 More work needs to be done! Only 7 out of 100+ Common Crawl crawls have been processed so far. That's encouraging because it means there is a lot more Creative Commons data to be collected! But to get there I need help in terms of compute. The current processing was already heavily sponsored by the Flemish Supercomputer but more is needed. If you have the compute available and which to collaborate in an open and transparent manner, please get in touch!
-
- 1 reply
lewtun
authored
a
paper
7 months ago
lewtun
authored
a
paper
8 months ago
Post
3835
Introducing OlympicCoder: a series of open reasoning models that can solve olympiad-level programming problems 🧑💻
- 7B open-r1/OlympicCoder-7B
- 32B open-r1/OlympicCoder-32B
We find that OlympicCoder models outperform Claude 3.7 Sonnet, as well as others over 100x larger 💪
Together with the models, we are releasing:
📊CodeForces-CoTs: new dataset of code problems from the most popular competitive coding platform, with R1 traces in C++ and Python open-r1/codeforces-cots
🏆 IOI'2024: a new benchmark of VERY hard programming problems where even frontier models struggle to match human performance open-r1/ioi
For links to the models and datasets, check out our latest progress report from Open R1: https://huggingface.co/blog/open-r1/update-3
- 7B open-r1/OlympicCoder-7B
- 32B open-r1/OlympicCoder-32B
We find that OlympicCoder models outperform Claude 3.7 Sonnet, as well as others over 100x larger 💪
Together with the models, we are releasing:
📊CodeForces-CoTs: new dataset of code problems from the most popular competitive coding platform, with R1 traces in C++ and Python open-r1/codeforces-cots
🏆 IOI'2024: a new benchmark of VERY hard programming problems where even frontier models struggle to match human performance open-r1/ioi
For links to the models and datasets, check out our latest progress report from Open R1: https://huggingface.co/blog/open-r1/update-3
- 1 reply
gsarti
authored
a
paper
8 months ago
Post
5486
Introducing OpenR1-Math-220k!
open-r1/OpenR1-Math-220k
The community has been busy distilling DeepSeek-R1 from inference providers, but we decided to have a go at doing it ourselves from scratch 💪
What’s new compared to existing reasoning datasets?
♾ Based on AI-MO/NuminaMath-1.5: we focus on math reasoning traces and generate answers for problems in NuminaMath 1.5, an improved version of the popular NuminaMath-CoT dataset.
🐳 800k R1 reasoning traces: We generate two answers for 400k problems using DeepSeek R1. The filtered dataset contains 220k problems with correct reasoning traces.
📀 512 H100s running locally: Instead of relying on an API, we leverage vLLM and SGLang to run generations locally on our science cluster, generating 180k reasoning traces per day.
⏳ Automated filtering: We apply Math Verify to only retain problems with at least one correct answer. We also leverage Llama3.3-70B-Instruct as a judge to retrieve more correct examples (e.g for cases with malformed answers that can’t be verified with a rules-based parser)
📊 We match the performance of DeepSeek-Distill-Qwen-7B by finetuning Qwen-7B-Math-Instruct on our dataset.
🔎 Read our blog post for all the nitty gritty details: https://huggingface.co/blog/open-r1/update-2
open-r1/OpenR1-Math-220k
The community has been busy distilling DeepSeek-R1 from inference providers, but we decided to have a go at doing it ourselves from scratch 💪
What’s new compared to existing reasoning datasets?
♾ Based on AI-MO/NuminaMath-1.5: we focus on math reasoning traces and generate answers for problems in NuminaMath 1.5, an improved version of the popular NuminaMath-CoT dataset.
🐳 800k R1 reasoning traces: We generate two answers for 400k problems using DeepSeek R1. The filtered dataset contains 220k problems with correct reasoning traces.
📀 512 H100s running locally: Instead of relying on an API, we leverage vLLM and SGLang to run generations locally on our science cluster, generating 180k reasoning traces per day.
⏳ Automated filtering: We apply Math Verify to only retain problems with at least one correct answer. We also leverage Llama3.3-70B-Instruct as a judge to retrieve more correct examples (e.g for cases with malformed answers that can’t be verified with a rules-based parser)
📊 We match the performance of DeepSeek-Distill-Qwen-7B by finetuning Qwen-7B-Math-Instruct on our dataset.
🔎 Read our blog post for all the nitty gritty details: https://huggingface.co/blog/open-r1/update-2
lewtun
authored
a
paper
9 months ago
Post
10489
We are reproducing the full DeepSeek R1 data and training pipeline so everybody can use their recipe. Instead of doing it in secret we can do it together in the open!
🧪 Step 1: replicate the R1-Distill models by distilling a high-quality reasoning corpus from DeepSeek-R1.
🧠 Step 2: replicate the pure RL pipeline that DeepSeek used to create R1-Zero. This will involve curating new, large-scale datasets for math, reasoning, and code.
🔥 Step 3: show we can go from base model -> SFT -> RL via multi-stage training.
Follow along: https://github.com/huggingface/open-r1
🧪 Step 1: replicate the R1-Distill models by distilling a high-quality reasoning corpus from DeepSeek-R1.
🧠 Step 2: replicate the pure RL pipeline that DeepSeek used to create R1-Zero. This will involve curating new, large-scale datasets for math, reasoning, and code.
🔥 Step 3: show we can go from base model -> SFT -> RL via multi-stage training.
Follow along: https://github.com/huggingface/open-r1
- 5 replies
Post
4017
I was initially pretty sceptical about Meta's Coconut paper [1] because the largest perf gains were reported on toy linguistic problems. However, these results on machine translation are pretty impressive!
https://x.com/casper_hansen_/status/1875872309996855343
Together with the recent PRIME method [2] for scaling RL, reasoning for open models is looking pretty exciting for 2025!
[1] Training Large Language Models to Reason in a Continuous Latent Space (2412.06769)
[2] https://huggingface.co/blog/ganqu/prime
https://x.com/casper_hansen_/status/1875872309996855343
Together with the recent PRIME method [2] for scaling RL, reasoning for open models is looking pretty exciting for 2025!
[1] Training Large Language Models to Reason in a Continuous Latent Space (2412.06769)
[2] https://huggingface.co/blog/ganqu/prime
Post
2378
This paper (
HuatuoGPT-o1, Towards Medical Complex Reasoning with LLMs (2412.18925)) has a really interesting recipe for inducing o1-like behaviour in Llama models:
* Iteratively sample CoTs from the model, using a mix of different search strategies. This gives you something like Stream of Search via prompting.
* Verify correctness of each CoT using GPT-4o (needed because exact match doesn't work well in medicine where there are lots of aliases)
* Use GPT-4o to reformat the concatenated CoTs into a single stream that includes smooth transitions like "hmm, wait" etc that one sees in o1
* Use the resulting data for SFT & RL
* Use sparse rewards from GPT-4o to guide RL training. They find RL gives an average ~3 point boost across medical benchmarks and SFT on this data already gives a strong improvement.
Applying this strategy to other domains could be quite promising, provided the training data can be formulated with verifiable problems!
* Iteratively sample CoTs from the model, using a mix of different search strategies. This gives you something like Stream of Search via prompting.
* Verify correctness of each CoT using GPT-4o (needed because exact match doesn't work well in medicine where there are lots of aliases)
* Use GPT-4o to reformat the concatenated CoTs into a single stream that includes smooth transitions like "hmm, wait" etc that one sees in o1
* Use the resulting data for SFT & RL
* Use sparse rewards from GPT-4o to guide RL training. They find RL gives an average ~3 point boost across medical benchmarks and SFT on this data already gives a strong improvement.
Applying this strategy to other domains could be quite promising, provided the training data can be formulated with verifiable problems!
- 1 reply
Post
7067
We outperform Llama 70B with Llama 3B on hard math by scaling test-time compute 🔥
How? By combining step-wise reward models with tree search algorithms :)
We show that smol models can match or exceed the performance of their much larger siblings when given enough "time to think"
We're open sourcing the full recipe and sharing a detailed blog post.
In our blog post we cover:
📈 Compute-optimal scaling: How we implemented DeepMind's recipe to boost the mathematical capabilities of open models at test-time.
🎄 Diverse Verifier Tree Search (DVTS): An unpublished extension we developed to the verifier-guided tree search technique. This simple yet effective method improves diversity and delivers better performance, particularly at large test-time compute budgets.
🧭 Search and Learn: A lightweight toolkit for implementing search strategies with LLMs and built for speed with vLLM
Here's the links:
- Blog post: HuggingFaceH4/blogpost-scaling-test-time-compute
- Code: https://github.com/huggingface/search-and-learn
Enjoy!
How? By combining step-wise reward models with tree search algorithms :)
We show that smol models can match or exceed the performance of their much larger siblings when given enough "time to think"
We're open sourcing the full recipe and sharing a detailed blog post.
In our blog post we cover:
📈 Compute-optimal scaling: How we implemented DeepMind's recipe to boost the mathematical capabilities of open models at test-time.
🎄 Diverse Verifier Tree Search (DVTS): An unpublished extension we developed to the verifier-guided tree search technique. This simple yet effective method improves diversity and delivers better performance, particularly at large test-time compute budgets.
🧭 Search and Learn: A lightweight toolkit for implementing search strategies with LLMs and built for speed with vLLM
Here's the links:
- Blog post: HuggingFaceH4/blogpost-scaling-test-time-compute
- Code: https://github.com/huggingface/search-and-learn
Enjoy!
- 2 replies
christopher
posted
an
update
11 months ago
Post
2104
The folks at Foursquare released a dataset of 104.5 million places of interest (
foursquare/fsq-os-places) and here's all of them on a plot
-
- 4 replies
christopher
posted
an
update
11 months ago
Post
2441
The Lichess database of games, puzzles, and engine evaluations is now on the Hub:
Lichess
Billions of chess data points to download, query, and stream and we're excited to see what you'll build with it! ♟️ 🤗
- https://huggingface.co/collections/Lichess/positions-datasets-66f50837db5cd3287d60d489
- https://huggingface.co/collections/Lichess/games-datasets-66f508df78f4b43e1bb2d353
Lichess
Billions of chess data points to download, query, and stream and we're excited to see what you'll build with it! ♟️ 🤗
- https://huggingface.co/collections/Lichess/positions-datasets-66f50837db5cd3287d60d489
- https://huggingface.co/collections/Lichess/games-datasets-66f508df78f4b43e1bb2d353
BramVanroy
posted
an
update
11 months ago
Post
1257
In the spirit of "Better late than never", I've finally written a brief overview paper for GEITje 7B Ultra. Initially released 10 months ago (oops), but still reaching around 1300 monthly downloads across the HF ecosystem (not including ollama).
GEITje 7B Ultra: A Conversational Model for Dutch (2412.04092)
While the paper discusses the model a little bit, I especially wanted to write about the datasets, which to this day seem an important asset for Dutch LLM training (SFT and preference tuning). We have a long way to go for Dutch, but publishing transparent and reproducible artefacts seems an important step to me, alongside having open discussions about data, bias, architectures.
In that spirit, thanks are in order for the creation of GEITje 7B Ultra and all related datasets:
- Michiel Buisman and UWV for providing the means to create the datasets
- Flemish Supercomputer Center (VSC) for the compute
- The Hugging Face Fellows and rest of the team for their discussions and insights
- The Dutch NLP community, notably @Rijgersberg for building the base GEITje model and the fruitful discussions we've had
More to come, step by step!
BramVanroy/geitje-7b-ultra-65c1ee010ad80fd1f6a8f208
GEITje 7B Ultra: A Conversational Model for Dutch (2412.04092)
While the paper discusses the model a little bit, I especially wanted to write about the datasets, which to this day seem an important asset for Dutch LLM training (SFT and preference tuning). We have a long way to go for Dutch, but publishing transparent and reproducible artefacts seems an important step to me, alongside having open discussions about data, bias, architectures.
In that spirit, thanks are in order for the creation of GEITje 7B Ultra and all related datasets:
- Michiel Buisman and UWV for providing the means to create the datasets
- Flemish Supercomputer Center (VSC) for the compute
- The Hugging Face Fellows and rest of the team for their discussions and insights
- The Dutch NLP community, notably @Rijgersberg for building the base GEITje model and the fruitful discussions we've had
More to come, step by step!
BramVanroy/geitje-7b-ultra-65c1ee010ad80fd1f6a8f208