Datasets:
Controlled Access Level (Blue – Research): authorization and an NDA with CENIA are required
This repository is NOT openly available. It holds data destined EXCLUSIVELY for scientific and academic research, managed under international custody standards analogous to PhysioNet or LDC protocols. Access is granted individually after review of each request. Submitting this form does not grant access: it opens a request that CENIA will evaluate, and approval additionally requires signing a Confidentiality Agreement (NDA) with CENIA.
This repository corresponds to the Controlled Access Level (Blue –
Research) of the LatamGPT source catalog: data destined exclusively for
scientific and academic research.
Per the project's data-governance framework, access requires three things:
(a) an accredited institutional affiliation, (b) a declaration of the
academic purpose for which the data will be used, and (c) signing a
Confidentiality Agreement (NDA) with CENIA. Submitting this form opens
the request; it does not grant access.
By submitting this request you agree to the following conditions:
Research use only. Use of this corpus is restricted to scientific
and academic research. Any other use — including commercial use, whether
direct or indirect — requires separate written authorization from CENIA.A Confidentiality Agreement (NDA) is mandatory. Approval of this
request is conditional on signing an NDA with CENIA guaranteeing
compliance with the project's ethical and legal standards. No access is
granted before that agreement is in force.Accredited institutional affiliation. You must be able to
substantiate the affiliation you declare, not merely state it. CENIA may
request supporting documentation.Authorization is individual and non-transferable. Access granted to
you does not extend to your employer, collaborators, or any third party.
You may not redistribute the corpus, in whole or in part, nor share the
credentials or files obtained through it.Licenses are managed at dataset level, not per document. Each
document retains the license recorded in its metadata, and 81.86% of them
prohibit commercial use outright. The remaining 18.14% carry CC BY-SA or
CC BY-ND licenses which, taken in isolation, would permit commercial use
— but they are held at this access level because they come from hybrid
repositories whose components cannot be segregated automatically, and
research use is the only use compatible with the most restrictive license
present in each set. Access to this repository grants no commercial
rights over any subset of it.Entities with annual revenue exceeding USD 1 billion, or that have
raised cumulative capital exceeding USD 1 billion, require CENIA's prior
written authorization before accessing, using, or redistributing this
corpus or any part of it. See "Restriction on use by large enterprises"
below for full details.The corpus is provided "as is." CENIA and the contributing institutions
do not warrant that the content is complete, accurate, current, or fit
for a particular purpose.You assume responsibility for the use of the corpus, its copies,
transformations, derived datasets, and models developed from it.Some datasets may contain technical traceability signals, such as
statistical watermarks, designed to assess whether they were used in
training an artificial-intelligence model.CENIA may revoke access at any time, in particular where these
conditions are breached, where the declared purpose changes, or where
the corpus is used or redistributed beyond the terms of the granted
authorization or of the NDA.
By submitting, you authorize the sharing of your name, username, email
address, affiliation, and declared purpose with the repository
administrators, for the purpose of evaluating the request and maintaining
an access record for the corpus.
Log in or Sign Up to review the conditions and access this dataset content.
LatamGPT-Corpus-1.0-research
🌐 Language versions: English | Español | Português
🔗 Project links: Official LatamGPT website | Corpus dashboard
🔓 Open portion: the openly released part of this corpus is published separately as LatamGPT-Corpus-1.0.
⚠️ Controlled Access Level (Blue – Research)
This repository is not openly available. It holds data destined exclusively for scientific and academic research, and is managed as a restricted repository under international custody standards analogous to PhysioNet or LDC protocols.
Access requires three things: an accredited institutional affiliation, a declaration of the academic purpose of the intended use, and signing a Confidentiality Agreement (NDA) with CENIA. Requests are submitted through the form on this page; submitting it does not grant access.
Dataset description
Summary
LatamGPT-Corpus-1.0-research is the research-access portion of the data corpus assembled for the continued pretraining of LatamGPT, a language model developed by CENIA together with a regional network of partner institutions.
It corresponds to the Controlled Access Level (Blue – Research) of the LatamGPT source catalog, and complements LatamGPT-Corpus-1.0, which holds the Open Access Level (Orange). By data volume, the research level accounts for approximately 16.2% of the complete corpus.
Two criteria place a source at this level:
- The contributing institution explicitly chose the governance modality "available and open for research purposes only."
- The source comes from a hybrid repository whose components carry heterogeneous licenses. Licenses are managed at dataset level, not per individual component, so where several licenses coexist it is not possible to satisfy all of them simultaneously — even when each is permissive on its own. Research use is the only use compatible with the most restrictive license present in the set, so the whole dataset is held at this level.
The dataset contains text in Portuguese, Spanish, and English associated with Latin America and the Caribbean.
Detailed project statistics are available in the LatamGPT corpus dashboard.
Dataset information
| Property | Value |
|---|---|
| Curated by | LatamGPT team (CENIA and partner institutions) |
| Access level | Controlled Access Level (Blue – Research) |
| Languages | Portuguese, Spanish, and English |
| License | Multiple licenses by source; see Licensing |
| Access | Research only — accredited affiliation, academic purpose, and NDA with CENIA |
| Custody standard | Restricted repository, analogous to PhysioNet / LDC protocols |
| Share of the complete corpus | ~16.2% by data volume |
| Companion open release | https://huggingface.co/datasets/latam-gpt/LatamGPT-Corpus-1.0 |
| Project dashboard | https://www.latamgpt.org/dashboard |
| Contact | latam-gpt@cenia.cl |
How to request access
- Submit the form on this page, declaring your verifiable institutional affiliation and the academic purpose of the intended use.
- CENIA reviews the request against the project's data-governance framework.
- If the request proceeds, you sign a Confidentiality Agreement (NDA) with CENIA covering the project's ethical and legal standards.
- Access is enabled only once that agreement is in force.
Questions about the process: latam-gpt@cenia.cl.
Uses
Direct use
Subject to the granted authorization, the signed NDA, and the license of each document, this dataset is intended for:
- Continued pretraining (CPT) or domain-adaptive pretraining of language models, within scientific or academic research.
- Research on the linguistic representation of Latin America.
- Studies of regional varieties and multilingual corpora.
- Corpus analysis and dataset-curation research.
Out-of-scope use
- Any use outside scientific and academic research, absent separate written authorization from CENIA.
- Any use directed at economic exploitation, direct or indirect, of the corpus or of models derived from it. This applies to the entire repository, including the CC BY-SA and CC BY-ND subsets whose licenses would permit commercial use in isolation — see Licensing.
- Redistribution of the corpus, in whole or in substantial part, to third parties.
- Sharing the files or access credentials obtained under a granted authorization.
⚠️ Restriction on use by large enterprises
Regardless of any authorization granted under this repository's access terms, access, use, or redistribution of any data released by the project by a large enterprise is strictly conditioned on obtaining prior written permission from CENIA. Failure to comply with this policy will be considered a violation of the governance terms.
For the purposes of this section, a large enterprise is understood to be any entity that, considered together with its parent companies, subsidiaries, and other related entities, records annual revenue exceeding USD 1 billion or has raised cumulative capital exceeding USD 1 billion, considering capital contributions, capital-convertible instruments, and comparable financing. Entities that do not exceed either of these thresholds may use the corpus for the purposes described in this section without prior permission. These entities are invited to inform CENIA of their use for traceability purposes, without this constituting a condition of access. This condition is evaluated as of the date the corpus is obtained; exceeding the threshold afterward does not affect models already trained with data obtained previously. CENIA may update these thresholds in future versions of the corpus.
For the purposes of this restriction, a foundation model is understood as an artificial-intelligence system trained on large volumes of data, of general purpose, and adaptable to a wide variety of downstream tasks through fine-tuning, adaptation, or programming interfaces. Commercial use is understood as use directed at the economic exploitation of the resulting model or of services derived from it, whether directly or indirectly, regardless of whether the model is distributed openly or in a closed manner.
This restriction does not apply to academic research, development by public institutions or non-profit civil-society organizations, or the development of specialized applications or models built on top of pre-existing foundation models. Under no circumstances do these exceptions authorize the use of the data by a large enterprise to train foundation models intended for commercial use.
Rationale: reciprocity
This condition responds to a principle of reciprocity. LatamGPT-Corpus was built with the voluntary contribution of public, academic, and civil-society institutions across Latin America and the Caribbean, through a collective effort aimed at strengthening the region's technological capabilities. Its release seeks to empower that community, not to subsidize the development of commercial models whose benefits do not return to the ecosystem that made the corpus possible.
CENIA acknowledges that this condition means that LatamGPT-Corpus, considered as a compilation, does not qualify as open source under the Open Source Initiative's definition nor as open data under the Open Definition, since both exclude restrictions based on field of use or type of user. This is a deliberate decision, adopted after weighing the trade-off between unrestricted openness and the sustainability of the regional effort. CENIA will review this condition in future versions of the corpus in light of accumulated experience.
Scope and enforcement mechanisms
This restriction operates through two complementary channels:
a. Repository access terms. Access to the repositories where CENIA makes LatamGPT-Corpus available is subject to prior acceptance of these access terms, which the user expressly accepts at the time of requesting the download. For this repository, access additionally requires express authorization from CENIA following review of the request. Obtaining the data constitutes acceptance of these conditions and creates a contractual relationship between the user and CENIA, governed by Chilean law and submitted to the jurisdiction of the ordinary courts of justice seated in the commune of Santiago, Chile.
b. Condition on the compilation. Independently of the above, this restriction applies to LatamGPT-Corpus as an integrated compilation curated and distributed by CENIA, comprising its selection, organization, processing, enrichment, metadata, and structure — elements that constitute the project's own intellectual contribution. This condition does not modify the original licenses applicable to third-party data incorporated into the corpus, nor does it limit their use when obtained independently from their original sources. Anyone who redistributes LatamGPT-Corpus as a compilation, in whole or in substantial part, must incorporate these same terms as a condition of the redistribution and record them in the corresponding documentation or metadata.
Consequences of non-compliance
Non-compliance with these conditions constitutes a violation of the access terms and of LatamGPT-Corpus's governance and licensing conditions, and may result in revocation of the granted authorization, suspension of access, and the exercise of any legal actions applicable under relevant law.
Authorization requests
Requests are submitted through the access form on this page. Organizations that require additional permission under this section, or that wish to discuss a use case not covered by the form, may write to latam-gpt@cenia.cl. CENIA will evaluate each case considering, among other factors, the initiative's contribution to the Latin American and Caribbean AI ecosystem.
Note on traceability mechanisms: released datasets into which this mechanism has been incorporated contain technical traceability signals, such as statistical watermarks, designed to assess whether they were used to train an artificial-intelligence model.
Note on compliance: each dataset preserves its original license in the metadata. End users assume responsibility for complying with those licenses, while the project maintains a permanent channel to identify and remove any data that lacks a compatible license (contact:
latam-gpt@cenia.cl).
Dataset structure
The corpus is distributed in 150 Parquet shards of approximately 500 MB each:
train-00000-of-00150.parquet
...
train-00149-of-00150.parquet
Row order is globally shuffled with a fixed seed, so documents from the same source are not contiguous.
Once your request has been approved, authenticate your machine:
hf auth login
The dataset can then be loaded with 🤗 Datasets:
from datasets import load_dataset
dataset = load_dataset(
"latam-gpt/LatamGPT-Corpus-1.0-research",
split="train",
streaming=True,
)
example = next(iter(dataset))
print(example)
Streaming is recommended for initial exploration because the release is approximately 72 GB. To download and materialize all shards locally, omit streaming=True and ensure that sufficient disk space is available.
Data fields
| Field | Type | Nulls | Description |
|---|---|---|---|
id |
string | 0% | Unique document hash |
text |
string | 0% | Document text |
language |
string | 0% | Detected language (portuguese, spanish, english) |
text_length |
int64 | 0% | Length of text in characters |
license |
string | 0% | License associated with the original source |
country |
string | 83.43% | One or more geographic labels separated by commas; the annotation space covers 20 countries and territories |
topic |
string | 51.63% | One or more topic labels separated by commas; the taxonomy covers ten main categories |
dataset_name |
string | 0% | Dataset identifier (58 distinct datasets) |
Country and topic annotations are sparser in this partition than in the open release, reflecting the composition of its sources.
Data splits
| Split | Examples | Shards | Size | Characters of text |
|---|---|---|---|---|
| train | 34,963,201 | 150 | ~72 GB | 188,615,334,050 |
Language distribution
| Language | Documents | Share |
|---|---|---|
| Portuguese | 26,782,248 | 76.60% |
| Spanish | 8,005,015 | 22.90% |
| English | 175,938 | 0.50% |
Complete-corpus distributions
Scope of the figures
The charts in this section describe the complete LatamGPT-Corpus-1.0 pretraining corpus, containing approximately 296.5 billion tokens, and not exclusively this non-commercial release.
Country and topic annotations are multilabel: a document may contribute to more than one category. The percentages are therefore non-exclusive and are not expected to sum to 100%.
The categories and their distributions are shown below and are available in greater detail in the project dashboard.
Topic token coverage
The complete corpus is organized around ten high-level topic labels covering political, social, scientific, educational, cultural, economic, health, media, sports, and Indigenous knowledge domains.
Figure 1. Multilabel token coverage by topic in the complete LatamGPT-Corpus-1.0 pretraining corpus.
Politics and Humanities & Social Sciences have the largest label-level token coverage in the complete corpus, followed by Economy & Finance, Education, Communication & Media, and Exact Sciences. Medicine & Health, Sports & Recreation, and Indigenous Peoples have lower coverage in this corpus version.
Country token coverage
The geographic annotation system covers 20 Latin American and Caribbean countries and territories.
Figure 2. Multilabel token coverage by country in the complete LatamGPT-Corpus-1.0 pretraining corpus.
Argentina and Mexico have the largest label-level token coverage, followed by Colombia, Chile, and Brazil. The observed distribution reflects the availability and composition of the collected sources and should not be interpreted as demographic weighting, population weighting, or an intentionally uniform allocation across countries.
Interactive charts and dataset-level details are available in the LatamGPT corpus dashboard.
Licensing
This dataset aggregates documents from multiple collections released under different licenses. The license applicable to each document is recorded in the license field.
This repository gathers the sources of the LatamGPT catalog whose licenses do not permit open redistribution. The catalog groups them into two families — CC Strict (non-commercial variants) and CC Permissive (share-alike and no-derivatives) — plus sources whose terms restrict reuse to research. The document-level distribution below covers all 34,963,201 records in this repository.
| License | Family | Documents | Share | The license alone would allow commercial use |
|---|---|---|---|---|
| CC BY-NC 4.0 | CC Strict | 26,663,736 | 76.26% | ❌ no |
| CC BY-SA 4.0 | CC Permissive | 5,426,287 | 15.52% | ✅ yes, with share-alike |
| Non-Commercial / Research Only | Research | 1,843,384 | 5.27% | ❌ no |
| CC BY-SA 3.0 | CC Permissive | 868,551 | 2.48% | ✅ yes, with share-alike |
| CC BY-NC-ND 4.0 | CC Strict | 46,233 | 0.13% | ❌ no |
| CC BY-ND 3.0 | CC Permissive | 43,226 | 0.12% | ✅ yes, no derivatives |
| CC BY-NC 3.0 IGO | CC Strict | 31,969 | 0.09% | ❌ no |
| CC BY-NC-SA 4.0 | CC Strict | 28,490 | 0.08% | ❌ no |
| CC BY-NC-SA 3.0 | CC Strict | 11,261 | 0.03% | ❌ no |
| CC BY-ND 4.0 | CC Permissive | 64 | <0.01% | ✅ yes, no derivatives |
⚠️ Read the last column carefully
It states what each license would permit in isolation — not what this repository permits. 18.14% of the documents carry CC BY-SA or CC BY-ND licenses that would allow commercial use if you obtained those documents independently from their original sources.
That is not the case here. Those documents sit at this access level because they come from hybrid repositories whose components cannot be segregated automatically: licenses are managed at dataset level, and where several coexist, research use is the only use compatible with the most restrictive one present. Access to this repository therefore grants no commercial rights over any subset of it, whatever the per-document license says.
If you need commercial rights over a specific subset, obtain those documents from their original sources under their own licenses, or write to
latam-gpt@cenia.cl.
81.86% of the documents are under licenses that prohibit commercial use outright.
Users remain responsible for complying with the license associated with each document, including attribution, share-alike, and no-derivatives obligations.
Traceability mechanisms
Released datasets that incorporate this mechanism contain technical traceability signals, such as statistical watermarks, designed to assess whether they were used in training an artificial-intelligence model.
Personal and sensitive information
The corpus contains documents from public, institutional, and web sources.
During processing, automated anonymization filters and measures aimed at reducing the presence of personal and sensitive information were applied in accordance with the project's data-protection criteria.
These mechanisms should not be considered exhaustive. The corpus may contain identifiable information present in the original sources.
Users should apply any additional review and filtering measures appropriate to their use case.
Biases, risks, and limitations
The corpus reflects the characteristics, availability, and limitations of its original sources. Therefore:
- Representation across languages, countries, and topics is not uniform. This partition is predominantly Portuguese (76.60%), and country annotation is absent for 83.43% of documents.
- Quality, currency, accuracy, and licensing may vary across documents.
- Offensive, biased, outdated, inaccurate, or identifiable content may be present.
- Automated curation and filtering processes reduce risks but do not guarantee their complete removal.
Users should assess whether the dataset is appropriate for their use case and apply any additional filtering, review, and validation measures that may be required.
Citation
@misc{latamgptcorpusresearch2026,
title = {LatamGPT-Corpus-1.0-research},
author = {LatamGPT Team},
year = {2026},
howpublished = {\url{https://huggingface.co/datasets/latam-gpt/LatamGPT-Corpus-1.0-research}}
}
Acknowledgments
LatamGPT is made possible by the collaborative work of technical teams, institutions, communities, and organizations contributing to the development of open and representative artificial intelligence for Latin America and the Caribbean.
- Downloads last month
- 25