Fixed tokenizer
Hey, @beezu . Thanks for making this quant of my model. I fixed the tokenizer issues in the original and I am uploading the fixed files now. They should be done in about an hour if you want to do something with the fixed version later.
I see the new version up now, I'll get some MLX quantizations going!
Edit: just realized I misread. I'll get these v0.2 quants fixed shortly.
Hey @sophosympatheia , I've got the updated quantizations all pushed now. I was waiting to respond until I tested further, because I was having some issues with v0.2 still, but it turns out that was due to my own misconfigured settings in SillyTavern. v0.2 MLX quantizations are behaving properly.
Sorry to have made all this extra work for you. I appreciate your grace in redoing the quants after I fixed the tokenizer issue. Thank you!
It was no problem at all! I enjoy the models you make. I'm just glad I can give back to the community in my own small way.