MOSS-Transcribe-preview-2B

Home/Models/OpenMOSS-Team/MOSS-Transcribe-preview-2B

OpenMOSS-Team/MOSS-Transcribe-preview-2B

kind: original
quant: bf16 (16-bit)
family: misc
modality: automatic-speech-recognition
root model: OpenMOSS-Team/MOSS-Transcribe-preview-2B
provider: OpenMOSS-Team
license: apache-2.0
language: —
context: —
library: transformers
vRAM (GB): —
size on disk: —
downloads: 2,633
likes: 44
created: —
last modified: 2026-06-26T11:49:43+00:00 (2026)
huggingface: https://huggingface.co/OpenMOSS-Team/MOSS-Transcribe-preview-2B
tags: transformerssafetensorsmosstext-generationspeechenglishqwen3audioreinforcement-learningautomatic-speech-recognitioncustom_codeendataset:openslr/librispeech_asrdataset:speechcolab/gigaspeechdataset:mozilla-foundation/common_voice_17_0dataset:facebook/voxpopulidataset:edinburghcstr/amidataset:anton-l/earnings22dataset:kensho/spgispeechlicense:apache-2.0model-indexeval-resultsregion:us

benchmark	category	score	context	conf	evidence
WER	speech-asr	8.37		low	`\| AMI \| 8.37 \|`
WER AMI	speech-asr	8.37	AMI	medium	`\| AMI \| 8.37 \|`
WER Earnings-22	speech-asr	7.84	Earnings22	medium	`\| Earnings22 \| 7.84 \|`
WER GigaSpeech	speech-asr	6.78	GigaSpeech	medium	`\| GigaSpeech \| 6.78 \|`
WER VoxPopuli	speech-asr	5.39	VoxPopuli	medium	`\| VoxPopuli \| 5.39 \|`
WER LibriSpeech	speech-asr	2.84	LibriSpeech test.other	medium	`\| LibriSpeech test.other \| 2.84 \|`
WER SPGISpeech	speech-asr	1.63	SPGISpeech	medium	`\| SPGISpeech \| 1.63 \|`
WER LibriSpeech	speech-asr	1.21	LibriSpeech test.clean	medium	`\| LibriSpeech test.clean \| 1.21 \|`

none scraped.