<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[JamesShuker的博客]]></title><description><![CDATA[]]></description><link>https://blog.youkuaiyun.com/JamesShuker</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; JamesShuker]]></copyright><item><title><![CDATA[STK导入自己的模型并定义关节【Getting CAD Models into STK with SpaceClaim】]]></title><link>https://blog.youkuaiyun.com/JamesShuker/article/details/131947814</link><guid>https://blog.youkuaiyun.com/JamesShuker/article/details/131947814</guid><author>JamesShuker</author><pubDate>Wed, 26 Jul 2023 21:27:12 +0800</pubDate><description><![CDATA[ansys官方的课，使用SpaceClaim为STK建模卫星模型：https://grabcad.com/library/120685#!Operator’s Toolbox 传感器显示窗口的STK插件ACT_STK_SpaceClaimAddIn SpaceClaim导入插件。]]></description><category></category></item><item><title><![CDATA[【linux+cuda11.4安装Pytorch-GPU】找不到包PackagesNotFoundError | 错下载为CPU版本Pytorch]]></title><link>https://blog.youkuaiyun.com/JamesShuker/article/details/128323721</link><guid>https://blog.youkuaiyun.com/JamesShuker/article/details/128323721</guid><author>JamesShuker</author><pubDate>Thu, 15 Dec 2022 00:29:25 +0800</pubDate><description><![CDATA[linux+cuda11.4安装pytorch-gpu
找不到PackagesNotFoundError | 错下载为CPU版本Pytorch]]></description><category></category></item><item><title><![CDATA[pytorch-cuda安装报错]]></title><link>https://blog.youkuaiyun.com/JamesShuker/article/details/128257519</link><guid>https://blog.youkuaiyun.com/JamesShuker/article/details/128257519</guid><author>JamesShuker</author><pubDate>Fri, 09 Dec 2022 18:44:43 +0800</pubDate><description><![CDATA[Found conflicts! Looking for incompatible packages.
This can take several minutes.  Press CTRL-C to abort.]]></description><category></category></item><item><title><![CDATA[服务器上Tensorflow、Keras、Cuda版本不匹配]]></title><link>https://blog.youkuaiyun.com/JamesShuker/article/details/125923204</link><guid>https://blog.youkuaiyun.com/JamesShuker/article/details/125923204</guid><author>JamesShuker</author><pubDate>Fri, 22 Jul 2022 09:34:05 +0800</pubDate><description><![CDATA[Tensorflow、Keras、Cuda不匹配]]></description><category></category></item><item><title><![CDATA[深度学习Review【四】编解码]]></title><link>https://blog.youkuaiyun.com/JamesShuker/article/details/119789294</link><guid>https://blog.youkuaiyun.com/JamesShuker/article/details/119789294</guid><author>JamesShuker</author><pubDate>Wed, 25 Aug 2021 22:06:12 +0800</pubDate><description><![CDATA[课程地址

动手学深度学习笔记【三】一、编码-解码二、Seq2seq
一、编码-解码
编码：处理输出，把输入编程成中甲你表达形式（特征）
解码：生成输出，把特征解码成输出
from torch import nn

class Encoder(nn.Module):
    """编码器-解码器结构的基本编码器接口。"""
    def __init__(self, **kwargs):
        super(Encoder, self).__init__(**kwargs)

    def f.]]></description><category></category></item><item><title><![CDATA[深度学习Review【三】序列、RNN、LSTM（GRU）、DRNN]]></title><link>https://blog.youkuaiyun.com/JamesShuker/article/details/119760939</link><guid>https://blog.youkuaiyun.com/JamesShuker/article/details/119760939</guid><author>JamesShuker</author><pubDate>Wed, 18 Aug 2021 22:13:22 +0800</pubDate><description><![CDATA[课程地址

动手学深度学习笔记【三】一、序列模型马尔科夫假设潜变量二、RNN 循环神经网络
一、序列模型
序列数据：实际中很多数据是随着时间变化而变化。
根据条件概率一直x1到xT的概率可以算出x的概率。

自回归模型：用的数据，预测现在的数据
马尔科夫假设
当前数据只和t个过去的数据相关

潜变量
引入潜变量h来表示过去的信息
绿色代表不同的潜变量h，每次在改变x后计算新的潜变量h

二、RNN 循环神经网络


...]]></description><category></category></item><item><title><![CDATA[深度学习Review【二】LeNet、AlexNet、VGG、NiN、GoogLeNet、ResNet、DenseNet]]></title><link>https://blog.youkuaiyun.com/JamesShuker/article/details/119684358</link><guid>https://blog.youkuaiyun.com/JamesShuker/article/details/119684358</guid><author>JamesShuker</author><pubDate>Tue, 17 Aug 2021 16:57:23 +0800</pubDate><description><![CDATA[动手学深度学习笔记【二】一、LeNettorch实现二、AlexNet特点torch实现三、VGG结构torch实现四、NiN思路、结构torch实现
一、LeNet
手写数字识别（MNIST）

使用了Conv2d、AvgPooling、Linear
高宽减半时通道数翻倍，保证信息能匹配更多的模式（将信息分配到多个通道）
输入超过100x100时MLP不如CNN，输入少时mlp更快
torch实现
import torch
from torch import nn
from d2l import torc]]></description><category></category></item><item><title><![CDATA[深度学习 Review【一】线性回归、Softmax、感知机、卷积]]></title><link>https://blog.youkuaiyun.com/JamesShuker/article/details/119651768</link><guid>https://blog.youkuaiyun.com/JamesShuker/article/details/119651768</guid><author>JamesShuker</author><pubDate>Fri, 13 Aug 2021 19:22:33 +0800</pubDate><description><![CDATA[线性回归、Softmax、感知机、卷积]]></description><category></category></item><item><title><![CDATA[Pytorch不得不看的基础]]></title><link>https://blog.youkuaiyun.com/JamesShuker/article/details/119522861</link><guid>https://blog.youkuaiyun.com/JamesShuker/article/details/119522861</guid><author>JamesShuker</author><pubDate>Fri, 13 Aug 2021 19:21:57 +0800</pubDate><description><![CDATA[Conv
Conv1d
torch.nn.Conv1d(in_channels, out_channels, kernel_size, stride=1, padding=0, dilation=1, groups=1, bias=True, padding_mode=‘zeros’, device=None, dtype=None)

in_channels(int) – 输入信号的通道。在文本分类中，即为词向量的维度
out_channels(int) – 卷积产生的通道。有多少个out_channel]]></description><category></category></item><item><title><![CDATA[Crypto模块安装 坑]]></title><link>https://blog.youkuaiyun.com/JamesShuker/article/details/119638843</link><guid>https://blog.youkuaiyun.com/JamesShuker/article/details/119638843</guid><author>JamesShuker</author><pubDate>Thu, 12 Aug 2021 11:11:17 +0800</pubDate><description><![CDATA[项目场景：
pip了各种Crypto，pycryptodome之类的，但仍然找不到Crypto模块

问题描述：
No Module Named ‘Crypto’


原因分析：
模块的名字为小写，导入时大小写敏感

解决方案：
找到lib/python3.6/site-packages目录下的crypto文件夹，把名字改为Crypto

...]]></description><category></category></item><item><title><![CDATA[wave-u-net：端到端（end-to-end）音源分离 1806]]></title><link>https://blog.youkuaiyun.com/JamesShuker/article/details/119363481</link><guid>https://blog.youkuaiyun.com/JamesShuker/article/details/119363481</guid><author>JamesShuker</author><pubDate>Sat, 07 Aug 2021 17:26:59 +0800</pubDate><description><![CDATA[https://arxiv.org/pdf/1806.03185.pdf

ABSTRACT
忽略了相位信息，性能依赖于频谱前端的超参数（hyper-parameters for the spectral front-end）。
Wave-u-net将u-net适应于一维时域的方法，通过重复对特征图重采样来计算和组合不同时间尺度的特征。
由（enforces source additivity）输出层、上采样、上下文感知的框架，减少输出。
INTRODUCTION
现有的方法：将信号（复值谱图）进行傅里叶.]]></description><category></category></item><item><title><![CDATA[进入环境后pip list没有下载好的环境，pip不指向conda]]></title><link>https://blog.youkuaiyun.com/JamesShuker/article/details/119358029</link><guid>https://blog.youkuaiyun.com/JamesShuker/article/details/119358029</guid><author>JamesShuker</author><pubDate>Tue, 03 Aug 2021 20:54:37 +0800</pubDate><description><![CDATA[项目场景：
Mac无法使用conda中的pip，由于pip更新后总是使用本机的pip

问题描述：
尽管已经source activate 环境
pip -V

发现pip并不是指向conda虚拟环境中的pip
 

原因分析：
pip太多冲突，默认指向本机的pip

解决方案：
注意查看是否pip3和python3是conda的命令
直接使用
/Users/liting/opt/anaconda3/envs/facett/bin/pip install 包

使用anconda-project
https]]></description><category></category></item><item><title><![CDATA[ECDSA host key for [localhost]:6000 has changed and you have requested strict checking. Host key ver]]></title><link>https://blog.youkuaiyun.com/JamesShuker/article/details/119208458</link><guid>https://blog.youkuaiyun.com/JamesShuker/article/details/119208458</guid><author>JamesShuker</author><pubDate>Thu, 29 Jul 2021 12:23:06 +0800</pubDate><description><![CDATA[项目场景：
使用ssh连接跳板机后，将服务器映射到本地端口时，无法映射

问题描述：
@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
@    WARNING: REMOTE HOST IDENTIFICATION HAS CHANGED!     @
@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
IT IS POSSIBLE THAT SOMEONE IS DO]]></description><category></category></item><item><title><![CDATA[Mac anconda Bug：ProxyError: Conda]]></title><link>https://blog.youkuaiyun.com/JamesShuker/article/details/117822864</link><guid>https://blog.youkuaiyun.com/JamesShuker/article/details/117822864</guid><author>JamesShuker</author><pubDate>Fri, 11 Jun 2021 17:15:16 +0800</pubDate><description><![CDATA[项目场景：
使用mac终端conda下载
且有过开代理或者VPN的经历
设置过系统HTTP环境变量
问题描述：
ProxyError: Conda …



原因分析：
env | grep -i “_PROXY”
查看代理
解决方案：
依次打开
vim ~/.bash_profile
vim ~/.bashrc
vim ~/.zshrc
vim ~/.condarc

查看有无类似如下 语句
export http_proxy="http://127.0.0.1:7890"
export https_pr]]></description><category></category></item><item><title><![CDATA[SSL_ERROR_SYSCALL in connection to github.com:443]]></title><link>https://blog.youkuaiyun.com/JamesShuker/article/details/117339955</link><guid>https://blog.youkuaiyun.com/JamesShuker/article/details/117339955</guid><author>JamesShuker</author><pubDate>Thu, 27 May 2021 23:01:55 +0800</pubDate><description><![CDATA[项目场景：
Mac使用git push或hexo deploy向github发出推送
问题描述：
BUG中包含
LibreSSL SSL_connect: SSL_ERROR_SYSCALL in connection to github.com:443

 

原因分析：
由于本机挂代理，可以clone
但在push时，需要给终端挂代理

解决方案：
vim ~/.gitconfig

查看本机代理的端口替换下面的xxxx
[http]
	proxy = socks5://127.0.0.1:xxxx
[]]></description><category></category></item><item><title><![CDATA[Mac使用SecureCRT或终端直接连接跳板机访问内网]]></title><link>https://blog.youkuaiyun.com/JamesShuker/article/details/113106907</link><guid>https://blog.youkuaiyun.com/JamesShuker/article/details/113106907</guid><author>JamesShuker</author><pubDate>Mon, 25 Jan 2021 14:28:12 +0800</pubDate><description><![CDATA[文章目录一、为什么使用跳板机二、为什么使用SecureCRT三、 连接方法1、连接跳板机2.tab显示绿色的对勾就表示连接成功四、附上一些linux常用命令conda指令终端挂代理显卡命令

一、为什么使用跳板机
由于学校放假，使用跳板机可以连接布置在学校内网的服务器

二、为什么使用SecureCRT
在尝试过Item2写expect连接跳板机后毅然决然的使用了SecureCRT

SecureCRT是Mac上最好用的终端模拟器、SSH客户端（之一）
给各位读者姥爷呈上破解版下载连接
请按照包内破解方法破]]></description><category></category></item></channel></rss>