Full (non-compressed) model files
Thank you for making it available to the community!
I was wondering if you would share the full, non-compressed version of the abliterated model?
[UPD - just read another thread to contact you on support@huihui.ai for those files]
On the opposite side. Is it possible to get a Q2_K version? I've tried quanitizing and it demands for under Q4 to use a matrix and I'm too new to understand where or how to do this. And 112GB is a bit too much, while 85Gb is within the means to use. (though Q1 I'd probably try too if it was an option)
The ablation effect of Q2_K is not very good. If needed, I can upload it.
If you don't think it's worth it then you don't have to (or whatever version you think is best for Q2)
I prefer i1 Q6_K, or Q4_K_M models; Recently upgraded to 128Gb, 100Gb eats a bit more than i'd like resource-wise.