update head img

shawnau · Feb 18, 2023 · 753217d · 753217d
1 parent 0435dd9
commit 753217d
Show file tree

Hide file tree

Showing 16 changed files with 36 additions and 31 deletions.
diff --git a/content/posts/2017-07-26-new-start.md b/content/posts/2017-07-26-new-start.md
@@ -3,12 +3,12 @@ title: 工坊装修记录
 date: 2017-07-27 01:31:19
 ---
 
-![](https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/8855672_p0.jpg)
-
 很久没有动blog了. 
 
 <!--more-->
 
+![](https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/8855672_p0.jpg)
+
 修整了一下, 希望能做得有趣一点, 别老更新冷冰冰的技术文, 也写点别的, 狗窝也得有个窝样是不是(x
 
 1. 更换主题(感谢原作者Litten, 伊莉雅真快) (done)

diff --git a/content/posts/2017-08-08-poisson.md b/content/posts/2017-08-08-poisson.md
@@ -3,10 +3,8 @@ title: Poisson Distribution Summary
 date: 2017-08-08 15:47:56
 tags: 
  - stat
-
-
-
 ---
+
 泊松分布是随机过程中的一个重要分布
 
 <!--more-->

diff --git a/content/posts/2017-08-18-core-python-programming-1.md b/content/posts/2017-08-18-core-python-programming-1.md
@@ -5,11 +5,13 @@ tags:
  - python
 
 ---
-![](https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/080_S.jpg)
+
 这本书翻译稀烂, 然而我还是忍着看完了第一部分
 
 <!--more-->
 
+![](https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/080_S.jpg)
+
 ## Python基础
 ### 基本风格指南
 1. 跨行代码可以使用反斜杠`\`

diff --git a/content/posts/2017-08-20-python-functional-programming.md b/content/posts/2017-08-20-python-functional-programming.md
@@ -5,10 +5,13 @@ tags:
  - python
 
 ---
-![](https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/gamersky_028small_056_201785187DEA.jpg)
+
 不学好函数式编程, 总觉得人生有遗憾
+
 <!--more-->
 
+![](https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/gamersky_028small_056_201785187DEA.jpg)
+
 ## 什么是函数
 1.  python函数可以返回一个值或对象, 但返回一个容器对象的时候看起来像是返回了多个值: `return 'abc', 'def'`实际上是返回了一个有两个元素的元组而已
 2. 函数有标准调用和关键字调用两种形式. 有关可变长度参数调用参考后文

diff --git a/content/posts/2017-09-04-gradient-boosting.md b/content/posts/2017-09-04-gradient-boosting.md
@@ -3,16 +3,14 @@ title: 梯度提升(Gradient Boosting)
 date: 2017-09-04 22:05:06
 tags:
  - gbdt
-
-
-
 ---
-![](https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/50782598_p0.png)
 
 自己写的老物了, 整理一下发出来, 可能会修改
 
 <!--more-->
 
+![](https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/50782598_p0.png)
+
 ### 1. 任意损失函数的Boosting
 损失函数的一般表示是:
 $$ L(y\_i, f(x\_i)) $$

diff --git a/content/posts/2017-09-21-numpy.md b/content/posts/2017-09-21-numpy.md
@@ -7,11 +7,11 @@ tags:
 
 ---
 
-![](https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/56167711_p0.jpg)
 用了这么久numpy居然没有总结一下
-<!--more-->
 
+<!--more-->
 
+![](https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/56167711_p0.jpg)
 
 ```python
 import numpy as np

diff --git a/content/posts/2017-09-25-python-oop.md b/content/posts/2017-09-25-python-oop.md
@@ -6,11 +6,12 @@ tags:
 
 ---
 
-![](https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/64495434_p0.jpg)
 python有很好的OOP特性, 自由度也非常大
 
 <!--more-->
 
+![](https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/64495434_p0.jpg)
+
 ## 命名规则
 1. 小驼峰命名法(camel case), 适用于变量名: `varFirst`, `verSecond`
 2. 大驼峰命名法, 适用于类名: `PersonFirst`, `PersonSecond`

diff --git a/content/posts/2017-09-26-python-thread.md b/content/posts/2017-09-26-python-thread.md
@@ -6,10 +6,12 @@ tags:
 
 ---
 
-![](https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/57080648_p0.jpg)
 即使有GIL的存在使得python的多线程显得鸡肋, 但在重I/O应用中还是很实用, 并且multiprecessing是基于threading的, 有必要理解
+
 <!--more-->
 
+![](https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/57080648_p0.jpg)
+
 ## GIL
 全局解释器锁(Global Interpreter Lock, GIL)是Python虚拟机的一个特性, 保证了任意时刻只有一个线程在解释器中运行. Python虚拟机的执行方式如下:
 

diff --git a/content/posts/2017-09-27-python-extend.md b/content/posts/2017-09-27-python-extend.md
@@ -6,10 +6,12 @@ tags:
 
 ---
 
-![](https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/46228691_p0.jpg)
-左手C++, 右手Python, 性能与表达力的完美融合
+C++ 和 Python 混合编程可以兼顾性能与效率
+
 <!--more-->
 
+![](https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/46228691_p0.jpg)
+
 ## Python扩展
 [官方文档](https://docs.python.org/3/extending/extending.html)
 

diff --git a/content/posts/2017-12-26-survival-models.md b/content/posts/2017-12-26-survival-models.md
@@ -4,17 +4,14 @@ date: 2017-12-26 19:21:27
 tags:
  - stat
  - GLM
-
-
-
 ---
 
-
-![](https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/58319465_p0.jpg)
 生存模型(Survival Models)属于General Linear Model, 被广泛用于Censored Data的建模, 譬如用户流失预测. 这里介绍下最基本的生存模型以及在Censored Data上的MLE估计
 
 <!--more-->
 
+![](https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/58319465_p0.jpg)
+
 ## Survival Function
 Assume $T$ is a continuous random variable indicates the death occurrence time, we have:
 

diff --git a/content/posts/2017-12-30-BinaryTreeTraverse.md b/content/posts/2017-12-30-BinaryTreeTraverse.md
@@ -5,12 +5,13 @@ tags:
  - Algorithm
 
 ---
-![](https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/50592758_p0.jpg)
 
 茴字有四种写法, 二叉树有三种遍历手段(递归, 栈, morris遍历)
 
 <!--more-->
 
+![](https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/50592758_p0.jpg)
+
 ### 先序遍历
 1. root进栈
 2. 栈非空:

diff --git a/content/posts/2018-01-29-LSTM.md b/content/posts/2018-01-29-LSTM.md
@@ -7,11 +7,13 @@ tags:
  - NLP
 ---
 
-![](https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/RNN3.png)
+
 新年第一篇, 和永无止境的八月一样循环的神经网络(其实是存货
 
 <!--more-->
 
+![](https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/RNN3.png)
+
 ## Standard RNN
 ![](http://colah.github.io/posts/2015-08-Understanding-LSTMs/img/RNN-unrolled.png)
 

diff --git a/content/posts/2018-02-03-CRF.md b/content/posts/2018-02-03-CRF.md
@@ -8,13 +8,13 @@ tags:
  - NLP
 
 ---
-![](https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/13010631_p0_master1200.jpg)
->你的PGM也会和Steins Gate的世界线一样收束吗?
 
 可能是有史以来公式最多的文章. 敲到吐血
 
 <!--more-->
 
+{{< image src="https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/13010631_p0_master1200.jpg" caption="你的PGM也会和Steins Gate的世界线一样收束吗?" height="5" width="5">}}
+
 ## 序
 设输入是$x$, 输出是$y$, HMM刻画的是联合概率分布$P(x, y | \lambda)$, 因此是生成模型, 可以同时进行推断$P(y|x, \lambda)$和解码$P(x|y, \lambda)$. 而CRF刻画的是条件概率分布$P(y|x, \lambda)$, 是判别模型
 

diff --git a/content/posts/2018-02-03-HMM.md b/content/posts/2018-02-03-HMM.md
@@ -8,11 +8,10 @@ tags:
 
 ---
 
-![](https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/4099531_p0_master1200.jpg)
-使用HMM的赌徒, 你还好吗? hmmm....
-
 <!--more-->
 
+{{< image src="https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/4099531_p0_master1200.jpg" caption="在 Cowboy Bebop 中, Faye 是个赌徒, 她使用HMM" height="5" width="5">}}
+
 ## 0. 定义
 
 ![HMM](https://my-imgshare.oss-cn-shenzhen.aliyuncs.com/HMM.png)

diff --git a/content/posts/2023-02-14-refresh.md b/content/posts/2023-02-14-refresh.md
@@ -4,7 +4,7 @@ date: 2023-02-14T23:07:39+08:00
 draft: false
 ---
 
-经过一系列艰苦的考古, 找回了当时七牛云挂掉之后备份的阿里云图床. 这么多年来兢兢业业为我当图床还没跑路, 甚至也没花一分钱 (至今账上还有2块钱), 比隔壁的什么牛厉害多了. (参考[被坑经历]({{< ref "/posts/2018-10-08-imgshare-dead" >}})).
+经过一系列艰苦的考古, 找回了当时七牛云挂掉之后备份的阿里云图床. 这么多年来兢兢业业为我当图床还没跑路, 甚至也没花一分钱 (至今账上还有2块钱), 比隔壁的什么牛厉害多了(参考[被坑经历]({{< ref "/posts/2018-10-08-imgshare-dead" >}})).
 
 很高兴过去写的文章都回来了! 下一步就是进一步装修(比如图床全部转移到 github 上并且用 cdn 加速), 以及写新的文章啦! 上一次修整博客还远在[17年]({{< ref "/posts/2017-07-26-new-start" >}}). 但是人不能被懒惰战胜!
 

diff --git a/public b/public