AI界的“调参玄学”：一个被遗忘的旋钮
[LG] Optimal Embedding Learning Rate in LLMs: The Effect of Vocabulary Size
[UC Berkeley & Microsoft Research]
https://arxiv.org/abs/2506.15025
在小宇宙查看该单集文稿
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.