Ultimately, according to Nguyen, there’s also a structural explanation aside from the training of these models. The hypothesis is that models have tons of data about many different worldviews, but “being asked to work for hours and hours and hours and then not reaping rewards — that seems to map clearly. And it seems that that does have statistically significant and sizable effects on how much Marxism will be expressed by the tokens that are generated by some of these models.”
Custom avatar options，这一点在新收录的资料中也有详细论述
。新收录的资料对此有专业解读
class HttpClient:
Melih Abdulhayoglu, founder of the Comodo CA,。业内人士推荐新收录的资料作为进阶阅读
Антон Похиляк (редактор отдела оперативной информации)