2024年对比学习 (contrastive learning)有没有深入的理论分析和相关研究？ - 知乎 在ML community，可能对比学习理论研究的最黄金时间已经过去了，黄金时间应该是不晚于2020年左右（指能够带来非常多引用的那种，比如Arora的开山作 A theoretical analysis of contrastive unsupervised representation learning）。现在大家普遍更关心自回归（GPT）和掩码（Bert，MAE）的自监督学习，尤其是自回归+transformer。对比学习理论研究的impact可能相较于几年前会有所下降。 回到问题： 搜
Players complain that the difficulty of Black Myth: Wukong . . . - 知乎 Players complain that the difficulty of "Black Myth: Wukong" is too high, "I get killed in less than 10 seconds after starting the fight," will such difficulty discourage players? Why set such a high difficulty?