[ouraf]

the "son" model might just be the future of LLMs! Please release "son" and the weights used to train him it on github (with a permissive license, if possible)

[AndrewKemendo]

Zero shot no less!