Add files using upload-large-folder tool

README.md

@@ -0,0 +1,135 @@
+---
+language:
+- nl
+license: apache-2.0
+task_categories:
+- text-generation
+- question-answering
+- translation
+pretty_name: Lumees Dutch Corpus 200B
+size_categories:
+- 100B<n<1T
+tags:
+- dutch
+- llm
+- foundation-model
+- pretraining
+- sft
+- fineweb
+- rechtspraak
+- bramvanroy
+configs:
+- config_name: pretrain
+  data_files: "pretrain/*.parquet" 
+- config_name: sft
+  data_files: "sft/*.parquet"
+---
+
+# Lumees Dutch Corpus 200B (LDC-200B)
+
+## Dataset Summary
+
+The **Lumees Dutch Corpus 200B (LDC-200B)** is the largest open-source, deduplicated, and professionally cleaned dataset designed for training Foundation Models in the Dutch language. Comprising approximately **202 Billion tokens** (measured with Qwen 2.5 tokenizer), it bridges the gap between high-resource English models and the Dutch ecosystem.
+
+The dataset is engineered for a two-stage training pipeline:
+
+1.  **Pretrain Subset (\~195B Tokens):** A massive, diverse mix of native Dutch web data, government publications, court verdicts, and high-fidelity translated books for continual pretraining.
+2.  **SFT Subset (\~7.25B Tokens):** A high-density collection of instruction-following, multi-turn chat, and reasoning tasks, including the complete Aya and xP3x Dutch collections.
+
+-----
+
+## Dataset Statistics
+
+*Estimates based on Qwen 2.5 (Byte-Level BPE) Tokenization.*
+
+| Subset | Format | File Type | Token Count
+| :--- | :--- | :--- | :---
+| **Pretrain** | Universal Schema | Parquet (ZSTD) | **\~194.95 Billion** | **SFT** | ChatML (List) | Parquet (ZSTD) | **\~7.25 Billion** | **Total** | - | - | **\~202.20 Billion** ---
+
+## Data Structure
+
+### 1\. Pretraining Subset (`pretrain`)
+
+Optimized for high-throughput streaming with libraries like `datatrove` or `nanotron`.
+
+| Column | Type | Description |
+| :--- | :--- | :--- |
+| `id` | `string` | Unique UUIDv4 (Vital for deduplication tracking). |
+| `text` | `string` | The cleaned, deduplicated content. |
+| `source` | `string` | Origin dataset (e.g., `fineweb-2`, `rechtspraak_verdicts`). |
+| `language` | `string` | ISO Code (`nl`). |
+| `meta` | `string` | Original metadata (URL, date, ECLI, title) serialized as JSON string. |
+
+### 2\. SFT Subset (`sft`)
+
+Optimized for "Instruction Pretraining" or large-scale Fine-Tuning.
+
+| Column | Type | Description |
+| :--- | :--- | :--- |
+| `id` | `string` | Unique UUIDv4. |
+| `messages` | `list` | Standard Conversational format: `[{"role": "user", "content": "..."}, {"role": "assistant", "content": "..."}]` |
+| `source` | `string` | Origin task (e.g., `no_robots_dutch`, `aya_collection`). |
+| `meta` | `string` | Task type, original ID, or template name serialized as JSON. |
+
+-----
+
+## Data Composition
+
+This corpus uses a **High-Quality Native + Synthetic Transfer** strategy. We prioritize native formal Dutch (Legal/Gov) and supplement it with high-quality translations for reasoning.
+
+### Pretrain Sources
+
+| Source | Type | Description |
+| :--- | :--- | :--- |
+| **[FineWeb-2 (Dutch)](https://huggingface.co/datasets/HuggingFaceFW/fineweb-2)** | Web Crawl | High-quality web backbone (subset `nld_Latn`). |
+| **[Rechtspraak](https://huggingface.co/datasets/vGassen/Dutch-Judiciary-Court-Cases-Netherlands-Rechtspraak-Vector-V3)** | Legal | Dutch court verdicts. Gold standard for formal logic and grammar. |
+| **[FinePDFs (Dutch)](https://huggingface.co/datasets/HuggingFaceFW/finepdfs)** | Documents | Technical papers, reports, and long-form PDF content. |
+| **[FineWiki NL](https://huggingface.co/datasets/HuggingFaceFW/finewiki)** | Knowledge | Full Dutch Wikipedia dump (cleaned). |
+| **[Officiële Publicaties](https://huggingface.co/datasets/vGassen/Dutch-Officiele-Publicaties)** | Government | Official gazettes and parliamentary papers. |
+| **[Apollo Books (UMCU)](https://huggingface.co/datasets/UMCU/apollo_english_books_translated_to_dutch_with_geminiflash15)** | Literature | Public domain books translated via Gemini 1.5 Flash (Long Context). |
+### SFT Sources
+
+| Source | Type | Description |
+| :--- | :--- | :--- |
+| **[BramVanroy Collection](https://huggingface.co/BramVanroy)** | Chat/Instruct | Includes `no_robots`, `ultrachat_200k`, `dolly-15k`, and `alpaca` (Dutch). |
+| **[Aya Collection](https://huggingface.co/datasets/CohereLabs/aya_collection_language_split)** | Multilingual | Massive human-annotated and templated instruction dataset (Dutch subset). |
+| **[xP3x](https://huggingface.co/datasets/CohereLabs/xP3x)** | NLP Tasks | Cross-lingual generalization tasks (`nld_Latn`). |
+
+-----
+
+## Processing Pipeline
+
+This dataset was engineered using a memory-efficient **Polars Streaming** pipeline designed for low-resource environments:
+
+1.  **Normalization:** 8+ distinct data sources were mapped to the unified `lumees` schema.
+2.  **Hashed Binning Deduplication:**
+      * To handle \>100GB of data without OOM errors, data was split into 50 hashed bins.
+      * Global deduplication was performed on the exact text content (whitespace normalized) within each bin.
+3.  **Dutch PII Redaction:**
+      * **BSN (Burgerservicenummer):** Conservative regex for 9-digit identifiers.
+      * **IBAN:** Dutch bank account numbers (`NLxx...`).
+      * **Phone:** Dutch mobile and landline formats (`+31`, `06`).
+      * **Postcode:** Dutch format (`1234 AB`).
+4.  **Sharding:** Data is split into optimized Parquet shards for distributed training.
+
+## Limitations
+
+  * **Translation Bias:** A portion of the "reasoning" data (Books, Reasoning) is synthetic/translated. While high quality (Gemini/Tower), it may contain translation artifacts.
+  * **Web Bias:** FineWeb-2 data reflects the general internet and may contain societal biases inherent to web text.
+
+-----
+
+## Citation & Attribution
+
+If you use this dataset in your research or product, please cite:
+
+```bibtex
+@misc{lumees2025dutch200b,
+  author = {Hasan KURŞUN, Kerem Berkay YANIK},
+  title = {Lumees Dutch Corpus 200B},
+  year = {2025},
+  publisher = {Lumees AI},
+  howpublished = {\url{https://lumees.io}},
+  email = {hello@lumees.io}
+}
+```

pretrain/dutch_corpus_shard_000.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8bc297bbf4ca98eafcf36df47c9f8e5285214dc8bb5ac1cefb3c4fdf0ef1bb47
+size 4800816924

pretrain/dutch_corpus_shard_003.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d629e6c343b0a3b3ac1f2a6c5facd86353fd806432c6f95f8f8ef0886250f1da
+size 4827883217

pretrain/dutch_corpus_shard_004.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d478eb7e82013c58ace67dde84645a559bb53e8dbad8dbab62bfeb581906f5fb
+size 4830397723

pretrain/dutch_corpus_shard_005.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9f0d3dc63832677a2dc315d0a1c23ebde204c67485cfd6ce3138fa47973e05f3
+size 4816404348

pretrain/dutch_corpus_shard_007.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:82322468f288ac16189c40fd7f9fef6904eaa98e5da6a2683c2ed9a7d906b6ee
+size 4795263036

pretrain/dutch_corpus_shard_008.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3475e16cf5514ad4d8aab334c9c5737648c7be9a829479d95633584060e5f77c
+size 4810691205

pretrain/dutch_corpus_shard_009.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9cb5659ba24e915659bfa0f41a09716e18029e94c14aaebe57bd299ea69732f4
+size 4799849860

pretrain/dutch_corpus_shard_010.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:415462adb2b0d5124e9c2a17d4305957563aaef09834c8a8d4cef6b04cbe2c46
+size 4802419414

pretrain/dutch_corpus_shard_011.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:279f867cf1f943cf22ecb89177b68e230d794477d775567fd001c2f380b1c961
+size 4787861401

pretrain/dutch_corpus_shard_012.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3d664dd5829f3efa2ece380a2668be1da68c2eaa0d627d6434e34065ab169f9b
+size 4828796142

pretrain/dutch_corpus_shard_015.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:562994a92cf277b4a7d2e6ae2d292e130b3ecef3cc71c1cf4c79f6d260efd869
+size 4823699420

pretrain/dutch_corpus_shard_016.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c2be235c59e54b75df41ac2aaae0763ad6ad35c41e1739001a208bdd7b04dbf
+size 4832365700

pretrain/dutch_corpus_shard_018.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:348c5e8b273dc3d5d80b6594aea67f3b72e09292017b0a3390c969d356043119
+size 4797119689

pretrain/dutch_corpus_shard_019.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:accf85e9a35c3cf336283f03c5d9182af3c462318f9c478f121f6230b062c701
+size 4802095306

pretrain/dutch_corpus_shard_020.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dc58ac51d3a292046b12c0e91fb2648b7378f53cb33ccba9633fe26e4f5b2e86
+size 4811288336

pretrain/dutch_corpus_shard_021.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5ee6259b31ff366fecd9ae4dd32be55c6c1d1d569c28f1b9e0fe0558bad898c5
+size 4800089374

pretrain/dutch_corpus_shard_022.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fb590de9b13361c8c73bac3841474dfb0174be0393384a16c0e3833f6d4ebef
+size 4790172086

pretrain/dutch_corpus_shard_024.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:97bdac370660e4f67c32bcf92fc97a356d8f3666d5bedebef4aa1000f592ed9e
+size 4776537558

pretrain/dutch_corpus_shard_025.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fae935ff03652a753280bca14be8c35941aca2775eaf4fcc770a9d8edad8b71f
+size 4849329678

pretrain/dutch_corpus_shard_027.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b8424b70c15b60c46a5796a2a30efa309f893d1572e5e9c493775f9838e98102
+size 4822739473

pretrain/dutch_corpus_shard_029.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e600a235ea1b97e00342ef20b05e159c9c9d8cecbfb23c7a19898ce9b82a28fa
+size 4830759893

pretrain/dutch_corpus_shard_031.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f6ba4c04ff4d328efee55450fcb3229124f170932b361b3c4f34ad0390ee1d8e
+size 4817150894

pretrain/dutch_corpus_shard_034.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8a4f6b39c3e7d140e6fc962b66b7af87982e509f8b5dd7d2178f0fd84350cf6c
+size 4822345269

pretrain/dutch_corpus_shard_036.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:99ec3a957925a9aed7ff0490b217c313c6ad3938b1fe728dd0040022eda0cc9b
+size 4820778558

pretrain/dutch_corpus_shard_039.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3eeccf3a5faa1f9510a356851cd0511880db7b3e2e1c4ef5c4a14a473d5f3faa
+size 4796329962

pretrain/dutch_corpus_shard_040.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c6c72066bc97b3d5a3239a50410cf61b0bf4707384000e9992d79ee0376e9a1e
+size 4852875353

pretrain/dutch_corpus_shard_041.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:86391019feb780275c88287beee4a00a99c5ebb44de8455e62d6107228efc689
+size 4797351186

pretrain/dutch_corpus_shard_043.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:49b20ab5d21c2f9e916ed7cf28fe573a9a2d0a00813fce7bbcd0d42963566de8
+size 4822088851

pretrain/dutch_corpus_shard_045.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:89308cac06c577e4fbbee3c60b7799cc4f25235e2efade7cf7648af94e382e52
+size 4848350015

sft/dutch_sft_shard_000.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:675af82e9c3de73677a601d1238bf8f2ff8ac7a2f2f6ec0b9be94a49bc5809fe
+size 401706079

sft/dutch_sft_shard_001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c50b897795d6ee329a170bd35449021010a7f71f929fd5fdf5689f4b0c6c3d06
+size 401003015

sft/dutch_sft_shard_002.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3f8df310d799c5f0e3ba8ee5c84c0cfe015ecb4e2d6a8e985073db7bed46ec7c
+size 402121203

sft/dutch_sft_shard_003.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a194fd9169df0a8b4a4175167dbac1ad5bf9bb03494a95c33c5734dffb86d21f
+size 401049009

sft/dutch_sft_shard_004.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d5e52941920e8dcebba2c17749394ba6a0b003994f2efde2f3a6cadca6e68a97
+size 401271004

sft/dutch_sft_shard_005.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:65d216ce223c96c5061c5d5668346739487e61028c400994bba29d10353ebc31
+size 402545664

sft/dutch_sft_shard_006.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8f551d36a66f6315fc4359650f2301a8cf6b26a4553d78e585ff760951a954b6
+size 400588278

sft/dutch_sft_shard_007.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1366d007bc26b5fa4521d8f92c663714baeb27117468121d53cc41bf1b1d0d00
+size 401057086

sft/dutch_sft_shard_008.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e54c217cc2193e09e4fe04647202b5d4cc48102d65c792913819489e61f7834
+size 401406024

sft/dutch_sft_shard_009.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:85ccf9f3af2d898e6f29f035d1336ff04e6b020497c2aa36d3b237ae4d5dee0a
+size 401860088

sft/dutch_sft_shard_010.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:abb80cf5542d91e1c7be532286bc0f611fe995cd58194061a30d3a6c792e97c0
+size 401834941

sft/dutch_sft_shard_011.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4c60193ad4a7f085710eb3ab81bb9e0ffa35d873c9b074d9c9a3bdc36e274c21
+size 400252508

sft/dutch_sft_shard_012.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6c92d4a192c7b6057002629ecf8cd1f49da6a7436108143fc3e4c5c8c8e386e0
+size 401700694

sft/dutch_sft_shard_013.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8827d8436e9f77878b8d6f5dc53eb419b2194c594123b1d0e16e23fadbea5bcf
+size 400312817

sft/dutch_sft_shard_014.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:844b6629fff77bfebab332f1b8eec3cda4419acad3e1923f264c11650c50f5cb
+size 401039106

sft/dutch_sft_shard_015.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:579a33f10f65255cde1fc224d77ff111c97650db908fcc43003000f35732f338
+size 400767849

sft/dutch_sft_shard_016.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:21a98d3c398c0c82ad3d1102e4bf780ee789298d0d5f108ad4ba8582bd8d83e3
+size 400680344

sft/dutch_sft_shard_017.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6ed17d7546f59c1457934c21c7e1767d4ffabfdb0d86eaee44683841a5c82107
+size 402613717

sft/dutch_sft_shard_018.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0e6a20fc99dc51009603d60aa3867975e21f84dfb8e985c0f258a4f8401c2620
+size 400896220

sft/dutch_sft_shard_019.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5abd6d3545c3a5fec4ba74bb8e61cd6f52dd534f2d3423f5a312ceeaa5b16c48
+size 401017855