VibeVoice-ASR-4-bit

4-bit quantization of the VibeVoice-ASR model making it possible to run on 16gb and even 12gb VRAM GPUs.

Usage example

  1. Follow VibeVoice-ASR installation instructions in Microsoft's VibeVoice repo
  2. pip install bitsandbytes
  3. python ./demo/vibevoice_asr_gradio_demo.py --model_path ./VibeVoice-ASR-4bit
Downloads last month
654
Safetensors
Model size
9B params
Tensor type
BF16
F32
U8
Inference Providers NEW
This model isn't deployed by any Inference Provider. 馃檵 Ask for provider support

Model tree for scerz/VibeVoice-ASR-4bit

Quantized
(2)
this model