Safetensors
qwen2

QwQ-32B

#8
by sm54 - opened

Hi,

Do you have any plans to make some new merges using the new QwQ-32b model? Also, a flash version of it would be great, as it outputs a huge number of tokens.

Thanks,

FuseAI org

New model merged with QwQ-32B is now ready for use!

https://huggingface.co./FuseAI/FuseO1-QwQ-DeepSeekR1-LightR1-32B

image.png

Thank you, I think using LightR1 might degrade the performance, since those models don't seem to perform very well. Any chance of a QwQ-DeepseekR1-Sky-T1 version maybe with flash? Or just QwQ-DeepseekR1 with flash?

@Wanfq Curios to know if you've tried merging merges together?

I've made my own merges now, thanks for all your work.

@SmilingWolf How is the performance looking like?

Hi! It seems you may have tagged the wrong person?

Oh my bad @SmilingWolf

@SmilingWolf How is the performance looking like?

I've not tested them yet, have almost downloaded one just now to test.

This comment has been hidden

@SmilingWolf How is the performance looking like?

I've not tested them yet, have almost downloaded one just now to test.

Let me know how it goes!!

Sign up or log in to comment