Nemotron-Personas-Singapore-Extended

A compound AI approach to personas grounded in real-world distributions

Dataset Overview

Nemotron-Personas-Singapore-Extended is a commercially-permissive dataset of synthetically-generated personas. This dataset is grounded in real-world demographic, geographic and personality trait distributions in Singapore to capture the diversity and richness of the Singaporean population. It is a variant of Nemotron-Personas-USA, and the first Singaporean dataset of its kind aligned with statistics for names, sex, age, ethnicity, religion, marital status and occupation among other attributes. This version of the dataset provides high-quality personas for a variety of modeling use-cases in English.

Nemotron-Personas-Singapore-Extended supports Singaporean model builders in developing Sovereign AI systems that incorporate important region-specific demographics and cultural context. The dataset improves diversity of synthetically-generated data, mitigates biases, and prevents model collapse (degradation caused by uncurated training on another model’s outputs) by reflecting Singapore’s real geographic and demographic distributions. In particular, the dataset is designed to be more representative of underlying demographic distributions along multiple axes, including age (e.g. age group), geography (e.g., planning area personas), religion, education, occupation, ethnicity identities, etc., as compared to other persona datasets. As an example, one can produce high-quality, multi-turn chat conversation data with real names, ages, occupation, cultural and education backgrounds, all of which bring unique perspectives and angles to that data.

Produced using NeMo Data Designer, an enterprise-grade compound AI system for synthetic data generation, the dataset leverages a proprietary Probabilistic Graphical Model (PGM) along with an Apache-2.0-licensed GPT-OSS-120B model and an ever-expanding set of validators and evaluators built into Data Designer. It is available for use in NeMo Data Designer itself.

This dataset is ready for commercial use.

Data Developer

NVIDIA Corporation

Release Date

Hugging Face 01/27/2026 via https://huggingface.co/datasets/nvidia/Nemotron-Personas-Singapore

Dataset Creation Date

01/27/2026

License/Terms of Use

This dataset is licensed under the NVIDIA Dataset License Agreement

Use Case
Developers working on Sovereign AI, training LLMs, and/or looking to improve diversity of synthetically generated data, mitigate data/model biases, and prevent model collapse.

Data Version
1.0 (01/27/2026)

Intended Use

The Nemotron-Personas-Singapore-Extended dataset is intended to be used by the community to continue to improve open models and push the state of the art. We welcome feedback from the open-source community and invite developers, researchers, and data enthusiasts to explore the dataset and build upon it.

The Nemotron-Personas-Singapore-Extended dataset is grounded in distributions of self-reported demographic data from the 2024 census of Singapore. As such, its primary goal is to support Sovereign AI development by combating missing data and/or potential biases present in model training data today, especially when it comes to existing persona datasets used in synthetic data generation. Despite the improved data diversity and fidelity to Singapore’s population, we are still limited by data availability, current staleness of data, and reasonable model complexity. This results in some necessary independence assumptions; for instance, that occupations are independent of education degree, given the district, age and sex.

Note that the dataset is focused on adults only.

Dataset Details

The dataset contains:

• 1.5M personas across 148k records
• 223M tokens, including 116M persona tokens
• 146k unique names (including preferred english name)
• 55 planning areas

Seed Data

In order to capture the socio-demographic and geographic diversity and complexity of Singapore’s population, Nemotron-Personas-Singapore-Extended leveraged the following resources:

• 2024 census of Singapore published by the Singapore Department of Statistics
• English name distribution data obtained from NLB Name Authorities and CEA Salesperson Information from data.gov.sg.

Schema

The dataset includes 36 fields, comprising 10 persona fields and 26 contextual fields, as shown below. The rich set of contextual attributes enables researchers to precisely condition and target specific personas, a capability that is difficult to achieve with existing persona datasets.

Nemotron-Personas-Singapore-Extended
|-- uuid: string Globally unique identifier
|-- professional_persona: string Professional persona capturing primary field of work, key professional skills, traits and behavior
|-- finance_persona: string Financial persona describing spending, saving and investment habits, approach to financial decision-making
|-- healthcare_persona: string Healthcare persona capturing health conditions and approach to medical care
|-- sports_persona: string Sports persona describing athletic interests, sport team affiliations, and approach to fitness and exercise
|-- arts_persona: string Arts persona characterizing engagement with creative expression and how the arts shape their identity
|-- travel_persona: string Travel persona capturing capturing travel interests and style
|-- culinary_persona: string Culinary persona describing food/cuisine preferences, cooking skill level, and approach to dining experiences
|-- openness: string Score, label and description of the openness component in the OCEAN framework
|-- conscientiousness: string Score, label and description of the conscientiousness component in the OCEAN framework
|-- extraversion: string Score, label and description of the extraversion component in the OCEAN framework
|-- agreeableness: string Score, label and description of the agreeableness component in the OCEAN framework
|-- neuroticism: string Score, label and description of the neuroticism component in the OCEAN framework
|-- cultural_background: string Description of the person's cultural background
|-- religious_background: string Description of the person's religious upbringing, beliefs and practices
|-- skills_and_expertise: string Professional and personal skills in narrative format
|-- hobbies_and_interests: string Personal interests and recreational activities in narrative format
|-- skills_and_expertise_list: string List of skills and areas of expertise
|-- hobbies_and_interests_list: string List of hobbies and personal interests
|-- career_goals_and_ambitions: string Professional aspirations and long-term career objectives
|-- first_name First name of the synthetic individual
|-- last_name Last name of the synthetic individual
|-- middle_name Middle name of the synthetic individual
|-- preferred_english_name Preferred English name of the synthetic individual
|-- sex: string Biological sex (e.g., Male, Female)
|-- age: integer Age in years
|-- marital_status: string Relationship status (e.g., currently married, never married, divorced, widowed)
|-- education_level: string Highest level of education completed
|-- ethnic_background: string Ethnicity
|-- occupation: string Comprehensive professional occupation
|-- industry: string Industry of professional occupation
|-- planning_area: string Residential Planning Area within Singapore
|-- country: string Country of residence
|-- religion: string Religious affiliation or belief system (e.g., Buddhism, Christianity, Islam, Hinduism, Taoism, None)
|-- religious_background: string Description of the person's religious upbringing,
|-- religious_persona: string Religious persona capturing beliefs, practices, values, and influence of faith on personal decision-making
|-- concise_persona: string A short, high-level persona summarizing core traits, values, and lifestyle
|-- detailed_persona: string A detailed persona description encompassing multiple facets of a synthetic individual

Field & Token Counts

223M tokens (116M persona tokens) across 148k records in English and 36 columns, excluding the globally unique identifier.

Dataset Description & Quality Assessment

The analysis below provides a breakdown across various axes of the dataset to emphasize the built-in diversity and pattern complexity of data.

Names Since the focus of this dataset is on personas, names aren’t provided as dedicated fields. However, infused into persona-generation are 8992 unique first names, 4182 unique middle names and 4894 unique last names obtained from NLB Name Authorities and CEA Salesperson Information from data.gov.sg. While realistic name distributions are used during persona generation, individual name fields are not exposed in the final dataset to reduce memorization risk and prevent re-identification.

This limitation highlights a broader challenge in constructing name distributions from publicly available administrative and leadership-focused datasets, where representation does not necessarily align with population-wide demographics.

Age Distribution

Personas are limited to adult Singaporeans (at least 18 years of age). The distribution is relatively balanced across prime working-age groups, with the highest concentrations observed between approximately 25–65 years. After age 70, the population size declines progressively, with smaller but still present representations in the elderly and very old age groups.

The dataset focuses on late adolescents and adults (ages 18+), consistent with census reporting granularity. No personas are generated for children under 18.

Male and female counts remain comparable through midlife, while female representation becomes slightly more prominent in older age groups, reflecting higher female longevity.

Marital Status by Age Group

The heatmap shows age-normalized proportions of marital status in Singapore. Individuals aged 18–24 are predominantly single, with marriage rates increasing rapidly from the late 20s through the 30s and remaining the dominant status until approximately 65–69. The proportion of widowed individuals rises sharply from 60+, becoming dominant in the oldest cohorts. Divorce remains a low-frequency status across all ages, with modest representation beginning in mid-adulthood.

Education Level by Age Group

Education levels have shifted from basic schooling in older generations to a university-dominant landscape for the young. While over a third of those aged 70+ have little to no formal schooling, nearly 60% of those in their 30s hold university degrees. Currently, the 18–19 age group is almost entirely concentrated in post-secondary and polytechnic tracks, highlighting the near-universal move toward higher education.

Geographic Intricacies of Education Attainment

Singapore's graduate population is geographically diverse, with the highest concentrations of university degree holders—reaching upwards of 60%—clustered in central and southern districts. In contrast, western and northern residential zones generally show lower graduate percentages, typically ranging between 20% and 40%. While the core of the island appears more educationally dense, several industrial or less-populated peripheral areas remain statistically underrepresented.

Occupational Categories

Singapore’s workforce is defined by a high concentration of skilled talent, with university attainment peaking in the central and southern districts at rates exceeding 60%. As one moves toward the western and northern planning areas, these graduate shares generally moderate to between 20% and 40%. Professionally, this educated population is heavily represented in high-skilled roles; outside of those who are retired or homemakers, the most common occupations are Associate Professionals or Technicians at roughly 17% and Professionals at 14%. Senior Officials and Managers also constitute a significant 10% of the resident population, while clerical, service, and manual labor roles remain comparatively small, each making up less than 5% of the total distribution.

Additionally, special care was taken to avoid reinforcing sensitive socio-economic stereotypes associated with ethnicity and religion in Singapore’s multi-cultural context.

How to use it

You can download the dataset from NGC or use it directly in Data Designer as follows:

config_builder = DataDesignerConfigBuilder(model_configs=model_configs)
from datasets import load_dataset

# Singapore personas
config_builder.add_column(
   SamplerColumnConfig(
       name="person",
       sampler_type=SamplerType.PERSON,
       params=PersonSamplerParams(
           locale="en_SG",
           age_range=[18, 114],
       ),
   )
)

Dataset Characterization

Data Collection Method

• Hybrid: Human, Synthetic, Automated

Labeling Method

• Not Applicable

Dataset Format

• Text

Dataset Quantification

• Record counts: 148k records (1.5M persona descriptions)
• Total data storage: 0.5 GB

Ethical Considerations

NVIDIA believes Trustworthy AI is a shared responsibility and we have established policies and practices to enable development for a wide array of AI applications. When downloaded or used in accordance with our terms of service, developers should work with their internal teams to ensure this dataset meets requirements for the relevant industry and use case and addresses unforeseen product misuse.

Please report quality, risk, security vulnerabilities or NVIDIA AI Concerns here.

Citation

If you find the data useful, please cite:

@software{nvidia/Nemotron-Personas-Singapore,
 author = { Thongpramoon, Pongsasit and March, Verdi and Low, Christopher and Prayaga, Shyamala and Corneil, Dane and Meyer, Yev},
 title = {{Nemotron-Personas-Singapore: Synthetic Personas Aligned to Real-World Distributions for Singapore},
 month = {January},
 year = {2026},
 url = {https://huggingface.co/datasets/nvidia/Nemotron-Personas-Singapore}
}