Datasets:

gsdf

/

EasyNegative

like 1.12k

Easynegative is rather effective, so I was hoping to get some information as to how to train a good negative embedding. I'd like to create some that are specific to my own merges for my own use.