bunnycore

/

HyperLLama3.1-8b-Nova

like 2

Mostly follows instructions excellently, and when instructed has well above average prose. Sometimes incoherent, although that often comes with creativity. Would be interesting to see a gradient merge old mythomax style to see if it can be made a little less likely to do that. I've done a few of those, and with the right models it works for maximizing both creativity and coherency.

Anyway, props on this it's well shiny - and I'm super unimpressed with llama-3 generally (and nemo)