I found this model hiding in the corner of huggingface: https://huggingface.co/Max-and-Omnis/Nemotron-3-Super-64B-A12B-Math-REAP-GGUF
Looks to be tuned specifically for math but i thought i'd give it a try since i cant run the full 12b nemotron super and it seem to hold up like a champ in agentic coding for some odd reason. been using it to code all my projects for a week now its amazing. Wouldnt dream of having 500k tokens on my potato dual TITAN RTX.
If you do happen to try it drop a cmment on your experience with it where did it break what usecase did u use it for ETC.
[link] [comments]