Thank you so much for converting out team's Skywork R1V3 model to the FP8 version!
#1
by
xuchensong
- opened
Thank you so much for converting out team's Skywork R1V3 model to the FP8 version and sharing it on Hugging Face! This is really helpful for both the community and our team I truly appreciate your contribution and support! 🙏
Best,
Xuchen
For sure! Working on an MR to the vLLM project for
- better vLLM config support for skywork quantized models
- mlp1 layer support for vLLM quantization
- reasoning parsing support