Datasets:

sumanthd
/

c3po-mixed-pretraining-corpus

text stringlengths 1.63k 140k	doc_id stringlengths 1 45	source stringclasses 7 values	metadata dict	token_count int64 1k 32k	tokenizer stringclasses 1 value
# Canadian Taxpayers Federation \| This article is part of a series on \| \| --- \| \| Conservatism in Canada \| \| Schools * Calgary School * Christian right * Clerico-nationalism * Right-wing populism + Trumpism * Social conservatism * Social credit * Toryism + Blue + Red \| \| Principles * Canadian values * Federalism * E...	1414852	wikipedia	{ "all_licenses": null, "arxiv_id": null, "audience": null, "authors": null, "blob_id": null, "branch_name": null, "category": null, "committer_date": null, "config": null, "content_id": null, "dataset_token_count": 8227, "dataset_token_length": null, "detected_licenses": null, "directory_id...	8,255	google/gemma-4-26b-a4b-it
# Poor Family Simulator How can I help? Some charities are 1000 times more effective. The following charities might be effective: - [GiveDirectly](https://www.givewell.org/charities/give-directly) - [Against Malaria Foundation](https://www.givewell.org/charities/amf) - [Deworm the World](https://www.givewell.org/ch...	4efca7d3aad3eda0e62b865d8f8f3111668fdc2f	stackv2_md	{ "all_licenses": null, "arxiv_id": null, "audience": null, "authors": null, "blob_id": "4efca7d3aad3eda0e62b865d8f8f3111668fdc2f", "branch_name": "refs/heads/main", "category": null, "committer_date": "2020-10-27T04:10:33", "config": null, "content_id": "4a55e683360977ea0f8b19ba955bbaa5143b013f", ...	2,260	google/gemma-4-26b-a4b-it
PROBLEM STEPS AND FUNCTION CODE: Here, you'll find the Python code for the initial steps of the problem-solving process. This code is integral to building the solution. Compute the total charge (monopole moment) Q from atomic charges Background: The monopole moment Q is the sum of all atomic partial charges. It repres...	074e2649-0655-4cd8-9bc6-bc2d4aa04a02	scientific_coding	{ "all_licenses": null, "arxiv_id": null, "audience": null, "authors": null, "blob_id": null, "branch_name": null, "category": "Nemotron-Pretraining-Scientific-Coding", "committer_date": null, "config": null, "content_id": null, "dataset_token_count": null, "dataset_token_length": null, "detec...	1,236	google/gemma-4-26b-a4b-it
# Toxoplasmosis Toxoplasmosis is a parasitic disease caused by Toxoplasma gondii, an apicomplexan. Infections with toxoplasmosis are associated with a variety of neuropsychiatric and behavioral conditions. Occasionally, people may have a few weeks or months of mild, flu-like illness such as muscle aches and te...	73592	wikipedia	{ "all_licenses": null, "arxiv_id": null, "audience": null, "authors": null, "blob_id": null, "branch_name": null, "category": null, "committer_date": null, "config": null, "content_id": null, "dataset_token_count": 11340, "dataset_token_length": null, "detected_licenses": null, "directory_i...	10,830	google/gemma-4-26b-a4b-it
\section{Introduction} \label{secIntro} W UMa stars are late-type contact binaries with short periods (of about 0.35 days) whose components typically have very different masses (with mass ratios around 0.3) but nearly identical temperatures ($\Delta T \le 200 K$). This paradoxical property is caused by the exchange of...	2103.06693	proof_pile	{ "all_licenses": null, "arxiv_id": "2103.06693", "audience": null, "authors": null, "blob_id": null, "branch_name": null, "category": null, "committer_date": null, "config": null, "content_id": null, "dataset_token_count": null, "dataset_token_length": null, "detected_licenses": null, "dire...	18,499	google/gemma-4-26b-a4b-it
"\\section{Introduction}\n\nWe study $r$-nets, a powerful tool in computational and metric geometry,(...TRUNCATED)	1607.04755	proof_pile	{"all_licenses":null,"arxiv_id":"1607.04755","audience":null,"authors":null,"blob_id":null,"branch_n(...TRUNCATED)	13,430	google/gemma-4-26b-a4b-it
"Representation Theory: A Journey through Abstract Mathematics and Combinatorics\n\nIn the world of (...TRUNCATED)	4486899512941329015	cosmopedia_v2	{"all_licenses":null,"arxiv_id":null,"audience":"college_students","authors":null,"blob_id":null,"br(...TRUNCATED)	1,085	google/gemma-4-26b-a4b-it
"---\nlayout: default\n---\n\n# 1996-05-03 - proposed anti-pseudospoofing law in Georgia\n\n## Heade(...TRUNCATED)	0dc47ba9732ef945c56dab5be6d6f0bb2daace56	stackv2_md	{"all_licenses":null,"arxiv_id":null,"audience":null,"authors":null,"blob_id":"0dc47ba9732ef945c56da(...TRUNCATED)	1,770	google/gemma-4-26b-a4b-it
"# 1977 Old Dominion 500\n\n\n\nThe 1977 Old Dominion 500 was a NASCAR Winston Cup Series racing(...TRUNCATED)	31058493	wikipedia	{"all_licenses":null,"arxiv_id":null,"audience":null,"authors":null,"blob_id":null,"branch_name":nul(...TRUNCATED)	2,393	google/gemma-4-26b-a4b-it
"PROBLEM STEPS AND FUNCTION CODE:\nHere, you'll find the Python code for the initial steps of the pr(...TRUNCATED)	6f11ab2e-2456-4f48-a1c6-53340b491704	scientific_coding	{"all_licenses":null,"arxiv_id":null,"audience":null,"authors":null,"blob_id":null,"branch_name":nul(...TRUNCATED)	1,014	google/gemma-4-26b-a4b-it

End of preview. Expand in Data Studio

C3PO Mixed Pretraining Corpus

Mixed pretraining corpus sampled from seven public Hugging Face datasets, filtered by token length (no truncation). Built by the c3po project.

Stats

Documents: 25,000
Tokenizer: google/gemma-4-26b-a4b-it
Token window: 1,000 – 32,000
Built: 2026-05-23T12:11:43.084386+00:00

Sources

Source	Documents	HF repo
Clean Wikipedia	4,940	DragonLLM/Clean-Wikipedia-English-Articles
Cosmopedia v2	2,870	HuggingFaceTB/smollm-corpus
Proof-Pile-2	4,972	EleutherAI/proof-pile-2
Scientific Coding	4,316	nvidia/Nemotron-Pretraining-Specialized-v1
Stack Exchange	4,305	common-pile/stackexchange
Stack V2 Edu (Markdown)	2,839	common-pile/stackv2_edu_filtered
olmOCR Mix 1025	1,099	allenai/olmOCR-mix-1025

Token statistics

Mean: 4,820.4
Median: 2,127.0
Min / max: 1,000 / 31,976

Schema

Field	Type	Description
`text`	string	Document text
`source`	string	Source slug (wikipedia, scientific_coding, …)
`doc_id`	string	Source-specific document id
`token_count`	int	Token count (Gemma 4 tokenizer)
`tokenizer`	string	Tokenizer used for `token_count`
`metadata`	object	Source-specific metadata

Load

from datasets import load_dataset
ds = load_dataset("sumanthd/c3po-mixed-pretraining-corpus", split="train")
print(ds[0]['source'], ds[0]['token_count'])

Downloads last month: 38