metadata
license: mit
tags:
- ggml
- bark.cpp
- quantized
- 8-bit
- 16-bit
- 32-bit
base_model: suno/bark
quantized_by: Green-Sky
bark.cpp quantized models
This is the large variant https://huggingface.co/suno/bark
done using bark.cpp commit d22ad71
I only uploaded f32, f16 and q8_0 quants, because q8_0 already changes the result significantly.
Don't use the f32 version of encodec, it does not work with bark right now.