mms-300m-waxal-lin

Home/Models/waxal-benchmarking/mms-300m-waxal-lin

waxal-benchmarking/mms-300m-waxal-lin

kind: original
quant: bf16 (16-bit)
family: misc
modality: automatic-speech-recognition
root model: facebook/mms-300m
provider: waxal-benchmarking
license: cc-by-nc-4.0
language: lin
context: —
library: transformers
vRAM (GB): —
size on disk: —
downloads: 431
likes: 0
created: —
last modified: 2026-06-05T22:41:03+00:00 (2026)
huggingface: https://huggingface.co/waxal-benchmarking/mms-300m-waxal-lin
tags: transformerssafetensorswav2vec2automatic-speech-recognitionafrican-languageswaxalwaxalnetlindataset:waxal-benchmarking/waxalarxiv:2606.02375base_model:facebook/mms-300mbase_model:finetune:facebook/mms-300mlicense:cc-by-nc-4.0endpoints_compatibleregion:us

benchmark	category	score	context	conf	evidence
WER	speech-asr	100.0		low	`\| 5.5307 \| 1.2258 \| 500 \| 2.8357 \| 1.0 \| 1.0 \|`
CER	speech-asr	100.0		low	`\| 5.5307 \| 1.2258 \| 500 \| 2.8357 \| 1.0 \| 1.0 \|`
WER	speech-asr	42.6	Test WER	medium	`\| Test WER \| 42.6% \|`
WER	speech-asr	42.6	WER	medium	`\| WER \| 42.6% \|`
CER	speech-asr	18.9	Test CER	medium	`\| Test CER \| 18.9% \|`
CER	speech-asr	18.9	CER	medium	`\| CER \| 18.9% \|`

none scraped.