MODELHugging Face

4/5

SERENGETI

Beembe (Kibembe)(beq)LM (Pretrained Encoder)XLM-RXLMRobertaForMaskedLM

Model Info

Language

Beembe (Kibembe) (beq)

Architecture

XLMRobertaForMaskedLM

Institution

Adebara, Elmadany, Abdul-Mageed, Inciarte (UBC-NLP)

Last Updated

14 May 2026

Intended Use Cases

Multiple NLU tasks across 20 datasets for African langs

License & Access

SPDX identifierProprietary (UBC license)

Restricted — review the source terms before use.

Provenance

Source URL

https://huggingface.co/UBC-NLP/serengeti

Fetch Timestamp

14 May 2026

Submitter

Adebara, Elmadany, Abdul-Mageed, Inciarte (UBC-NLP)

Full Record

24 of 25 fields stated

Every column the registry tracks. Values marked derivedwere filled by a rule or read from the source's text rather than stated in its metadata; the rule or the quoted sentence is shown under the value.

Registry columns (17/18)

iso code: beq
Model Name: SERENGETI
Where is it hosted: HuggingFace/GitHub
Link: https://huggingface.co/UBC-NLP/serengeti
Year: 2023
Creator/Author(s): Adebara, Elmadany, Abdul-Mageed, Inciarte (UBC-NLP)
Paper Reference: SERENGETI: Massively Multilingual Language Models for Africa (Findings of ACL 2023)
NLP Task: LM (Pretrained Encoder)
Architecture: XLMRobertaForMaskedLMderivedHugging Face config: architectures / model_type
Base Model: XLM-R
Fine-tuned?: Yes
License: Proprietary (UBC license)
Open Access: Open
Has Model Card: Yes
SOTA Tasks: Multiple NLU tasks across 20 datasets for African langs
Verified By: Not stated
Date Indexed: 2026-05-14
Comment: Adebara et al. 2023

Readiness (LRS) (7/7)

model id: serengeti-beq
Model Type: LLMderivedHugging Face card: pipeline_tag
Training Data Sources: 517 African languages + 10 high-resource languages
Benchmark Used: SERENGETI NLU Eval Suitederivedregistry benchmark records for this model
Benchmark Score: Multiple NLU tasks for African languages
Benchmark Date: 2023
Status: Activederivedupdated within 3 years

Readiness Score Breakdown

4/5

Data Coverage77%

Model Availability100%

Benchmarks67%

Documentation79%

Recency100%

Representativeness

Gender Balance

Regional Distribution

Dialect Coverage

Inclusivity

Relations Explorer link

Also Available In

This is the same multilingual corpus, sliced for other languages the platform indexes it under.

serengeti

fill-mask model · XLMRobertaForMaskedLM · UBC-NLP

32 languagesAfar(aar)XLMRobertaForMaskedLMfill-mask

Restricted27 Oct 2026

serengeti

fill-mask model · XLMRobertaForMaskedLM · UBC-NLP

32 languagesAfrikaans(afr)XLMRobertaForMaskedLMfill-mask

Restricted27 Oct 2026

serengeti

fill-mask model · XLMRobertaForMaskedLM · UBC-NLP

32 languagesAkan(aka)XLMRobertaForMaskedLMfill-mask

Restricted27 Oct 2026

serengeti

fill-mask model · XLMRobertaForMaskedLM · UBC-NLP

32 languagesAmharic(amh)XLMRobertaForMaskedLMfill-mask

Restricted27 Oct 2026

serengeti

fill-mask model · XLMRobertaForMaskedLM · UBC-NLP

32 languagesBambara(bam)XLMRobertaForMaskedLMfill-mask

Restricted27 Oct 2026

SERENGETI

Adebara et al. 2023

32 languagesBari(bfa)XLM-RLM (Pretrained Encoder)

Open14 May 2026

SERENGETI

Adebara et al. 2023

32 languagesMalba Birifor(bfo)XLM-RLM (Pretrained Encoder)

Open14 May 2026

SERENGETI

dug is one of 517 African languages in SERENGETI pretraining. Confirmed in supported-languages.txt on HuggingFace. Not individually benchmarked. Also: SERENG…

32 languagesDuguri(dug)XLM-RLM

Open26 May 2026

SERENGETI

ebu is one of 517 African languages in SERENGETI pretraining. Confirmed in supported-languages.txt on HuggingFace. Not individually benchmarked.

32 languagesEmbu (Kîembu)(ebu)XLM-RLM

Open26 May 2026

SERENGETI

ego is one of 517 African languages in SERENGETI pretraining. Confirmed in supported-languages.txt on HuggingFace. Not individually benchmarked.

32 languagesEggon(ego)XLM-RLM

Open26 May 2026

SERENGETI

flr (Fuliiru) confirmed in SERENGETI supported-languages list. Not individually benchmarked. Commercial use requires author permission.

32 languagesFuliiru (Kifuliiru)(flr)XLM-RLM

Open26 May 2026

serengeti

fill-mask model · XLMRobertaForMaskedLM · UBC-NLP

32 languagesFon(fon)XLMRobertaForMaskedLMfill-mask

Restricted27 Oct 2026

serengeti

fill-mask model · XLMRobertaForMaskedLM · UBC-NLP

32 languagesFula(ful)XLMRobertaForMaskedLMfill-mask

Restricted27 Oct 2026

serengeti

fill-mask model · XLMRobertaForMaskedLM · UBC-NLP

32 languagesHausa(hau)XLMRobertaForMaskedLMfill-mask

Restricted27 Oct 2026

serengeti

fill-mask model · XLMRobertaForMaskedLM · UBC-NLP

32 languagesIgbo(ibo)XLMRobertaForMaskedLMfill-mask

Restricted27 Oct 2026

serengeti

fill-mask model · XLMRobertaForMaskedLM · UBC-NLP

32 languagesGikuyu(kik)XLMRobertaForMaskedLMfill-mask

Restricted27 Oct 2026

serengeti

fill-mask model · XLMRobertaForMaskedLM · UBC-NLP

32 languagesLingala(lin)XLMRobertaForMaskedLMfill-mask

Restricted27 Oct 2026

serengeti

fill-mask model · XLMRobertaForMaskedLM · UBC-NLP

32 languagesLuganda(lug)XLMRobertaForMaskedLMfill-mask

Restricted27 Oct 2026

serengeti

fill-mask model · XLMRobertaForMaskedLM · UBC-NLP

32 languagesMalagasy(mlg)XLMRobertaForMaskedLMfill-mask

Restricted27 Oct 2026

serengeti

fill-mask model · XLMRobertaForMaskedLM · UBC-NLP

32 languagesSouthern Ndebele(nbl)XLMRobertaForMaskedLMfill-mask

Restricted27 Oct 2026

Related Models

AfroLID v1.5

Verified in supported-languages file

3 languagesBeembe (Kibembe)(beq)SERENGETIClassification (LID)

Open14 May 2026

Cheetah

Adebara et al. 2024

3 languagesBeembe (Kibembe)(beq)T5/mT5-styleNLG (Generation)

Open14 May 2026

GlotLID-M

Likely covers via Bible/Wikipedia; per-lang verification required

550 languages2 copiesBeembe (Kibembe)(beq)FastTextClassification (LID)

Open14 May 2026