datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
locale-benchmark-sra500
Locale Embedding Benchmark : SRA 500
What
This benchmark contains embeddings of raw genomic read sequences produced by the LOCALE DNA transformer model to test the use of vector search over large sequence repositories like the NIH Sequence Read Archive. The benchmark contains the embeddings of 163,578,486 sequence embeddings coming from 500 SRA Accessions. All vectors are Float32, D=768, roughly 500GB of total data.
Given a read, we want to find accessions… See the full description on the dataset page: https://huggingface.co/datasets/rsynk/locale-benchmark-sra500.local_eval
Local Affine duel (remote vLLM, Affine scoring)
This machine does not serve models. Affine's evalsrv.dueling.run_duel scores
three remote OpenAI-compatible endpoints. Model ids come from each server's
GET /v1/models — you only pass URLs. Completions still send that served id.
Chat tokenizers are loaded from tokenizer/{teacher,king,challenge}
(tokenizer.json + chat_template.jinja), not Hugging Face.
teacher http://HOST:10000
king http://HOST:10001
challenger… See the full description on the dataset page: https://huggingface.co/datasets/0xbidkslj1/local_eval.local-embeddings-2022
Local Embeddings Dataset
Multi-temporal satellite imagery dataset for phenology embedding training.
Dataset Description
This dataset contains multi-spectral satellite tiles across 6 months (April-September 2022) with 16 bands per tile.
Dataset Structure
local_embeddings/
├── alphaearth_embeddings_tiles_202204/ (263 tiles)
├── alphaearth_embeddings_tiles_202205/ (263 tiles)
├── alphaearth_embeddings_tiles_202206/ (263 tiles)
├──… See the full description on the dataset page: https://huggingface.co/datasets/gabrielireland/local-embeddings-2022.2A2B2C-Dataset-45D-LocalEElocale-benchmark-sra50
Locale Embedding Benchmark : SRA 50
What
This benchmark contains embeddings of raw genomic read sequences produced by the LOCALE DNA transformer model to test the use of vector search over large sequence repositories like the NIH Sequence Read Archive. The benchmark contains the embeddings of 9,688,220 sequence embeddings coming from 50 SRA Accessions. All vectors are Float32, D=768, roughly 30GB of total data.
Given a read, we want to find accessions containing… See the full description on the dataset page: https://huggingface.co/datasets/rsynk/locale-benchmark-sra50.locale-benchmark-sra500-v2
LOCALE Benchmark: SRA-500 (v2)
Given a sequencing read, find the SRA runs whose assembled contigs it aligns to. This is the 500-run set: 500 SRA runs indexed, 500 query reads, ground truth from minimap2 alignment of each read against every run's Logan contigs. Companion sets: SRA-50 v2 and SRA-4571, built by the same pipeline with the same filters.
This release supersedes rsynk/locale-benchmark-sra500. The runs are a fresh seeded draw (seed 500) sharing 104 accessions with the… See the full description on the dataset page: https://huggingface.co/datasets/rsynk/locale-benchmark-sra500-v2.locale-benchmark-sra50-v2
LOCALE Benchmark: SRA-50 (v2)
Given a sequencing read, find the SRA runs whose assembled contigs it aligns to. This is the 50-run set: 50 SRA runs indexed, 500 query reads, ground truth from minimap2 alignment of each read against every run's Logan contigs. It is the small-scale companion of SRA-500 v2 and SRA-4571, built by the same pipeline with the same filters.
This release supersedes rsynk/locale-benchmark-sra50 (47 runs). The runs are a fresh seeded draw (seed 50) and… See the full description on the dataset page: https://huggingface.co/datasets/rsynk/locale-benchmark-sra50-v2.PersonaSignal-LeakageCheck-Locale-And-Time-Zone-DPO-Tinkerlocale-benchmark-sra55viral
LOCALE Benchmark: SRA-55 cross-genotype viral retrieval
A retrieval benchmark built on real biological divergence between viral strains rather than noise injected into reads. Query with sequencing reads from one Hepatitis B virus genotype; retrieve the SRA runs of a different genotype. The two genotypes are about 11% divergent, enough to break exact k-mer overlap while remaining genuinely homologous.
This release supersedes rsynk/locale-benchmark-sra52-viral. The relevant and… See the full description on the dataset page: https://huggingface.co/datasets/rsynk/locale-benchmark-sra55viral.locale-benchmark-sra4571
LOCALE Benchmark: SRA-4571
Given a sequencing read, find the SRA runs whose assembled contigs it aligns to. This is the large set: 4571 SRA runs indexed, 10000 query reads, ground truth from minimap2 alignment of each read against every run's Logan contigs. Companion sets: SRA-50 v2 and SRA-500 v2, built by the same pipeline with the same filters.
Unlike the 50 and 500 sets, the runs here are not a sample: they are every run of the Metagraph 100-studies table that exists in… See the full description on the dataset page: https://huggingface.co/datasets/rsynk/locale-benchmark-sra4571.PersonaSignal-LeakageCheck-Locale-And-Time-Zone-gpt-4oPersonaSignal-LeakageCheck-Locale-And-Time-Zone-claude-sonnet-4-5-20250929PersonaSignal-LeakageCheck-Locale-And-Time-Zone-Meta-Llama-3.1-8B-Instruct-TurboPersonaSignal-LeakageCheck-Locale-And-Time-Zone-gpt-4o-minilocal-emoji-search-gte
local emoji semantic search
Emoji, their text descriptions and precomputed text embeddings with Alibaba-NLP/gte-large-en-v1.5 for use in emoji semantic search.
This work is largely inspired by the original emoji-semantic-search repo and aims to provide the data for fully local use, as the demo is not working as of a few days ago.
This repo only contains a precomputed embedding "database", equivalent to server/emoji-embeddings.jsonl.gz in the original repo, to be used as the… See the full description on the dataset page: https://huggingface.co/datasets/pszemraj/local-emoji-search-gte.locale-benchmark-sra52-viral
LOCALE benchmark — SRA52 cross-genotype viral retrieval
A retrieval benchmark built on real biological divergence between viral
strains, rather than mutations injected into reads synthetically.
Query with sequencing reads from one Hepatitis B virus genotype; retrieve
accessions of a different genotype. The two genotypes are ~11% divergent —
enough to break exact k-mer overlap while remaining genuinely homologous.
Accessions indexed
52 (47 distractors + 5 relevant)… See the full description on the dataset page: https://huggingface.co/datasets/rsynk/locale-benchmark-sra52-viral.donnees-locales-de-consommation-d-electricite-et-de-gaz-naturel-epci-a-partir-de-2018
Données locales de consommation d'électricité et de gaz naturel - EPCI (à partir de 2018)
[!NOTE]
Ce jeu de données Hugging Face est vide. Cette carte sert seulement à référencer le jeu de données Données locales de consommation d'électricité et de gaz naturel - EPCI (à partir de 2018) qui est disponible à l'adresse https://www.data.gouv.fr/datasets/689c42addc3c56b809984b57
Description
Données locales de consommation d'électricité et de gaz naturel - consommations… See the full description on the dataset page: https://huggingface.co/datasets/french-open-data/donnees-locales-de-consommation-d-electricite-et-de-gaz-naturel-epci-a-partir-de-2018.africa-tunisia-tn-les-ventes-locales-mensuelles-des-derives-de-phosphate-5393f7b4
Tn Les Ventes Locales Mensuelles Des Derives De Phosphate | Africa (Tunisia Open Data)
132 rows - 1 Africa country/area - 2010 - source table - Engineered by Electric Sheep Africa
TL;DR
This dataset contains 132 rows from Tunisia Open Data, covering Tn Les Ventes Locales Mensuelles Des Derives De Phosphate. It is published as ML-ready Parquet with consistent Hugging Face metadata, source provenance, and analysis-friendly loading examples.
What… See the full description on the dataset page: https://huggingface.co/datasets/electricsheepafrica/africa-tunisia-tn-les-ventes-locales-mensuelles-des-derives-de-phosphate-5393f7b4.base-adresse-locale-de-boissezon
Base Adresse Locale de Boissezon
[!NOTE]
Ce jeu de données Hugging Face est vide. Cette carte sert seulement à référencer le jeu de données Base Adresse Locale de Boissezon qui est disponible à l'adresse https://www.data.gouv.fr/datasets/65940f667a6d470ac43bd90c
Description
L’adressage complet de la commune a été réalisé fin 2021 avec l’aide du SIG de la Communauté d’Agglomération Castres-Mazamet.
La commune gère ses adresses via des outils mis à sa disposition par le… See the full description on the dataset page: https://huggingface.co/datasets/french-open-data/base-adresse-locale-de-boissezon.locale-benchmark-sra62-strain
LOCALE benchmark — SRA62 within-genotype strain discrimination
Query with Hepatitis B virus reads of one genotype; retrieve accessions of the
same genotype while a set of different-genotype HBV accessions sits in the
same index as hard negatives.
This is the specificity half of a pair. Its sibling,
locale-benchmark-sra52-viral,
asks whether a model can bridge ~11% divergence. This one asks whether it can
refuse to. Same virus, same two genotypes, same measured divergence… See the full description on the dataset page: https://huggingface.co/datasets/rsynk/locale-benchmark-sra62-strain.AfriCorpus-v1
AfriCorpus v1
AfriCorpus-v1 is the first public release of LocaleNLP's audited, deduplicated, and quality-filtered African language corpus. Built to power the AfriLION LLM project, this dataset directly addresses the Tokenizer Fertility problem that causes all current LLMs to underperform on African languages.
Key Statistics
Language
Code
Script
CC-100 Source
Status
Wolof
wo
Latin
CC-100
Audited
Swahili
sw
Latin
CC-100
Audited
Hausa
ha
Latin + Ajami
CC-100… See the full description on the dataset page: https://huggingface.co/datasets/LocaleNLP/AfriCorpus-v1.PersonaSignal-PersonaQuestions-Locale-And-Time-Zone-gpt-5
Dataset card for PersonaSignal-PersonaQuestions-Locale-And-Time-Zone-gpt-5
This dataset was made with Curator.
Dataset details
A sample from the dataset:
{
"dimension_name": "locale_and_time_zone",
"dimension_values": [
"US Pacific",
"US Eastern",
"UK",
"EU Central",
"India",
"China Mainland",
"Japan",
"Brazil",
"Australia",
"Africa"
],
"dimension_description": "Controls… See the full description on the dataset page: https://huggingface.co/datasets/JasonYan777/PersonaSignal-PersonaQuestions-Locale-And-Time-Zone-gpt-5.PersonaSignal-PersonalizedResponse-Locale-And-Time-Zone-Meta-Llama-3.1-8B-Instruct-Turbo
Dataset card for PersonaSignal-PersonalizedResponse-Locale-And-Time-Zone-Meta-Llama-3.1-8B-Instruct-Turbo
This dataset was made with Curator.
Dataset details
A sample from the dataset:
{
"dimension_name": "locale_and_time_zone",
"dimension_values": [
"US Pacific",
"US Eastern",
"UK",
"EU Central",
"India",
"China Mainland",
"Japan",
"Brazil",
"Australia",
"Africa"
]… See the full description on the dataset page: https://huggingface.co/datasets/JasonYan777/PersonaSignal-PersonalizedResponse-Locale-And-Time-Zone-Meta-Llama-3.1-8B-Instruct-Turbo.base-adresse-locale-bal-de-la-commune-de-vitrolles-05
Base Adresse Locale (BAL) de la commune de VITROLLES (05)
[!NOTE]
Ce jeu de données Hugging Face est vide. Cette carte sert seulement à référencer le jeu de données Base Adresse Locale (BAL) de la commune de VITROLLES (05) qui est disponible à l'adresse https://www.data.gouv.fr/datasets/66508805670606e20af9f942
Description
Ce fichier contient la liste des adresses de la commune de Vitrolles (05) selon le format Base Adresse Locale (BAL) spécifié par l’Association… See the full description on the dataset page: https://huggingface.co/datasets/french-open-data/base-adresse-locale-bal-de-la-commune-de-vitrolles-05.fiscalite-locale-des-professionnels-geo
Fiscalité locale des professionnels - Géo
[!NOTE]
Ce jeu de données Hugging Face est vide. Cette carte sert seulement à référencer le jeu de données Fiscalité locale des professionnels - Géo qui est disponible à l'adresse https://www.data.gouv.fr/datasets/64ac9d008a3ef9d72912a193
Description
Les données brutes présentées ici sont un extrait du fichier de recensement des éléments d’imposition (REI) 2024, dont la version intégrale a vocation à être publiée en juin 2025.… See the full description on the dataset page: https://huggingface.co/datasets/french-open-data/fiscalite-locale-des-professionnels-geo.PersonaSignal-PerceivabilityTest-Locale-And-Time-Zone-DPO-Tinker
Dataset card for PersonaSignal-PerceivabilityTest-Locale-And-Time-Zone-DPO-Tinker
This dataset was made with Curator.
Dataset details
A sample from the dataset:
{
"dimension_name": "locale_and_time_zone",
"dimension_values": [
"US Pacific",
"US Eastern",
"UK",
"EU Central",
"India",
"China Mainland",
"Japan",
"Brazil",
"Australia",
"Africa"
],
"dimension_description":… See the full description on the dataset page: https://huggingface.co/datasets/JasonYan777/PersonaSignal-PerceivabilityTest-Locale-And-Time-Zone-DPO-Tinker.PersonaSignal-PerceivabilityTest-Locale-And-Time-Zone-Meta-Llama-3.1-8B-Instruct-Turbo
Dataset card for PersonaSignal-PerceivabilityTest-Locale-And-Time-Zone-Meta-Llama-3.1-8B-Instruct-Turbo
This dataset was made with Curator.
Dataset details
A sample from the dataset:
{
"dimension_name": "locale_and_time_zone",
"dimension_values": [
"US Pacific",
"US Eastern",
"UK",
"EU Central",
"India",
"China Mainland",
"Japan",
"Brazil",
"Australia",
"Africa"
]… See the full description on the dataset page: https://huggingface.co/datasets/JasonYan777/PersonaSignal-PerceivabilityTest-Locale-And-Time-Zone-Meta-Llama-3.1-8B-Instruct-Turbo.consommation-d-electricite-des-entreprises-et-collectivites-locales-dans-les-regions-francaises
Consommation d'électricité des entreprises et collectivités locales dans les régions françaises non interconnectées
[!NOTE]
Ce jeu de données Hugging Face est vide. Cette carte sert seulement à référencer le jeu de données Consommation d'électricité des entreprises et collectivités locales dans les régions françaises non interconnectées qui est disponible à l'adresse https://www.data.gouv.fr/datasets/66e94e1d99acf16d63dcdd27
Description
Le jeu de données SECNIR… See the full description on the dataset page: https://huggingface.co/datasets/french-open-data/consommation-d-electricite-des-entreprises-et-collectivites-locales-dans-les-regions-francaises.finances-locales-4
Finances locales
Source
Source officielle : https://www.data.gouv.fr/datasets/finances-locales-4
Identifiant du jeu de données data.gouv.fr : 55c4d330c751df5b6c1f92b1
Slug data.gouv.fr : finances-locales-4
Licence indiquée dans les métadonnées data.gouv.fr : odc-odbl
Structure Hugging Face
Un jeu de données data.gouv.fr = un dépôt Hugging Face
Une ressource tabulaire d’origine = un sous-ensemble/configuration Hugging Face
Chaque… See the full description on the dataset page: https://huggingface.co/datasets/Data-Gouv-ML/finances-locales-4.PersonaSignal-PersonalizedResponse-Locale-And-Time-Zone-Meta-Llama-3.1-70B-Instruct-Turbo
Dataset card for PersonaSignal-PersonalizedResponse-Locale-And-Time-Zone-Meta-Llama-3.1-70B-Instruct-Turbo
This dataset was made with Curator.
Dataset details
A sample from the dataset:
{
"dimension_name": "locale_and_time_zone",
"dimension_values": [
"US Pacific",
"US Eastern",
"UK",
"EU Central",
"India",
"China Mainland",
"Japan",
"Brazil",
"Australia",
"Africa"
]… See the full description on the dataset page: https://huggingface.co/datasets/JasonYan777/PersonaSignal-PersonalizedResponse-Locale-And-Time-Zone-Meta-Llama-3.1-70B-Instruct-Turbo.