facebook/mms-1b-all

Kamba(kam)ASRwav2vec 2.0 (1B params)Encoder (wav2vec 2.0)

View Source

Model Info

Language

Kamba (kam)

Architecture

Encoder (wav2vec 2.0)

Institution

Pratap, Vineel et al. (Meta AI)

Last Updated

3 Jun 2026

Intended Use Cases

ASR (1107+ languages)

License & Access

SPDX identifierCC-BY-NC-4.0

Open with restrictions — attribution required; some commercial or derivative use limited.

Provenance

Source URL

https://huggingface.co/facebook/mms-1b-all

Fetch Timestamp

3 Jun 2026

Submitter

Pratap, Vineel et al. (Meta AI)

Full Record

23 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: kam
Model Name: facebook/mms-1b-all
Where is it hosted: HuggingFace
Link: https://huggingface.co/facebook/mms-1b-all
Year: 2023
Creator/Author(s): Pratap, Vineel et al. (Meta AI)
Paper Reference: Pratap et al. (2023) — Scaling Speech Technology to 1,000+ Languages. arXiv:2305.13516
NLP Task: ASR
Architecture: Encoder (wav2vec 2.0)
Base Model: wav2vec 2.0 (1B params)
Fine-tuned?: No
License: CC-BY-NC-4.0
Open Access: Open
Has Model Card: Yes
SOTA Tasks: ASR (1107+ languages)
Verified By: Not stated
Date Indexed: 2026-06-03
Comment: Kamba (kam) covered in Meta MMS 1B ASR. Also used in FLEURS and XTREME-S speech benchmarks for evaluation.

Readiness (LRS) (6/7)

model id: facebook-mms-1b-all-kam
Model Type: ASRderivedHugging Face card: pipeline_tag
Training Data Sources: MMS multilingual corpus (491K hrs, 1406 languages)
Benchmark Used: MMS ASR Benchmark (Biafada adapter)derivedregistry benchmark records for this model
Benchmark Score: ASR WER for Biafada
Benchmark Date: 2023
Status: Not stated