Feature Request: Unsloth Dynamic 2.0 Quantization Support

#1
by rikunarita - opened

The GGUF conversion space is already super useful πŸ”₯
I was thinking β€” would it be possible to also support Unsloth Dynamic 2.0 quantization in the future?
It could give better quality at similar sizes and would be very attractive for people running models on low-VRAM / CPU setups.
I can help with testing and preparing example settings if you're interested.

Good idea . Will explore this 2.0

Sign up or log in to comment