-
2026视觉语言模型四大技术方案拆解:落地差异与选型避坑指南
视觉语言模型(VLM)的核心迭代逻辑,是如何高效将视觉特征与预训练语言模型融合,当前行业所有主流VLM均可归为四大技术路径。不同方案在训练成本、…
-
读懂「254A, Notes 3: The operator norm of a random mat」:先把定义、反例和适用范围钉死
Now that we have developed the basic probabilistic tools that we will ne…
-
把「254A, Notes 3a: Eigenvalues and sums of Hermitia」写成可执行的阅读路径
Let be a Hermitian matrix. By the spectral theorem for Hermitian matrice…
-
「“Density Hales-Jewett and Moser numbers” – neari」笔记:问题从哪来、证明卡在哪
After a hiatus of several months, I’ve made an effort to advance the wri…
-
「254A, Notes 3b: Brownian motion and Dyson Browni」:数学直觉如何落到可检验的步骤
One theme in this course will be the central nature played by the gaussi…
-
小样本深度学习数据增强全攻略(2026落地版)含降噪训练与质量量化技巧
小样本场景下深度学习模型泛化能力差、易过拟合、标签噪声干扰严重是行业普遍痛点,核心解决思路为通过合规数据扩充提升样本量,依托量化标准把控数据质量…
-
把「Some notes on group extensions」写成可执行的阅读路径
In mathematics, one frequently starts with some space and wishes to exte…
-
「Linear approximate groups」:数学直觉如何落到可检验的步骤
Emmanuel Breuillard, Ben Green, and I have just uploaded to the arXiv ou…
-
「Ultralimit analysis, and quantitative algebraic」笔记:问题从哪来、证明卡在哪
I have blogged a number of times in the past about the relationship betw…
-
「254A, Notes 4: The semi-circular law」:数学直觉如何落到可检验的步骤
We can now turn attention to one of the centerpiece universality results…