Automatic Speech Recognition
MLX
Safetensors
moss_transcribe_diarize
speaker-diarization
ios
korean
custom_code
Instructions to use WJO/MOSS-Transcribe-Diarize-MLX-q8all-e8 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use WJO/MOSS-Transcribe-Diarize-MLX-q8all-e8 with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] huggingface-cli download --local-dir MOSS-Transcribe-Diarize-MLX-q8all-e8 WJO/MOSS-Transcribe-Diarize-MLX-q8all-e8
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
MOSS Transcribe Diarize · MLX q8all-e8
OpenMOSS-Team/MOSS-Transcribe-Diarize를 Apple Silicon과 iPhone의 MLX Swift 런타임에서 사용하기 위해 변환한 8-bit 배포본입니다.
- 선형 계층: affine 8-bit, group size 64
- 토큰 임베딩: affine 8-bit, group size 64
- 대상 런타임: Substream
MossTranscriptionKit - 기본 모델 라이선스: Apache-2.0
원본 BF16 체크포인트와 실행 형식이 다릅니다. Substream iOS 앱에서는 설정 → 기기 내 모델 → MOSS 온디바이스에서 이 저장소를 직접 다운로드하고, 구성·크기·양자화 모듈·SHA-256 manifest를 검증한 뒤 설치합니다.
- Downloads last month
- -
Model size
0.5B params
Tensor type
F16
·
U32 ·
Hardware compatibility
Log In to add your hardware
Quantized
Model tree for WJO/MOSS-Transcribe-Diarize-MLX-q8all-e8
Base model
OpenMOSS-Team/MOSS-Transcribe-Diarize