What is this model about?
#1
by GeoMaciolek - opened
This has exceptionally good scores for low KL divergence and number of refusals after the changes. Is this "just" a decensored model, or has it been finetuned or otherwise trained, as implied by the model name?