Mustafa Suleyman· @mustafasuleyman · X·· 3 小时前AI 评分51
AI 导读
Mustafa Suleyman 提出 scaling laws 正在叠加效率增益，并称在 held-out 代码上，其团队用少 12 倍的训练算力就达到了 DeepSeek-V4-Pro-Base 的预训练分数。
正文
Scaling laws, scaling... with compounding efficiency!
On held-out code, we matched DeepSeek-V4-Pro-Base's pre-trained score using 12x less training compute.
来源：Mustafa Suleyman · x.com