weixin_45649658的博客

原创 MIT Missing Semester 1

Virtual MachineHypervisors VirtualBox(open-source) Virt-manager(open-source, manages KVM virtual machines and LXC containers) VMWare(commercial, available from IS&Tfor MIT students) ContainersVirtual Machines are relatively heavy-weight;...

转载 Self-study Plan for becoming a quantitative trader

Self-Study Plan for Becoming a Quantitative Trader - Part I | QuantStartNice BeginBBeginner's Guide to Quantitative Trading | QuantStartQuantitative trading is an extremely sophisticated area of quant finance. It can take a significant amount of time

原创 股票入门-基金

你每天需要多看看这些网站，看看每天的重大新闻，看看这些新闻对个股和股市产生了哪些影响，再看看大家的解读，这些非常重要，别看小道消息，多看这些专业大型的网站。千万不要相信任何小道消息！要用多少钱投资闲钱，炒股风险很大，比你想象的大100倍，所以闲钱炒股是唯一的选择，假如你存款100万，闲钱应该是要去除养老的钱，养孩子的钱，再留一些钱，最多20万投资股票是合理的。对于一个新手来说差不多20%的存款就是极限了，等你真正学会炒股以后你可以适当的加大投入，但是我个人的体会来看根本不需要，因为真正会炒股

转载 新手入门pytorch

去看pytorch.org 的tutorial！！！I'd say that the official tutorials are a great start (Welcome to PyTorch Tutorials). There you have a lot of examples of all the things you'll probably run into when trying to design an architecture and train it: dataloaders, N

转载 THE FUNDAMENTALS OF AUTOGRAD

What Do We Need Autograd For?A machine learning model is afunction, with inputs and outputs. For this discussion, we’ll treat the inputs a as ani-dimensional vector\vec{x}x, with elementsx_{i}xi​. We can then express the model,M, as a vector-valued ...

原创 Some tools

import argparse parser = argparse.ArgumentParser() parser.add_argument('--save', type=str, default='work') parser.add_argument('--nEpoch', type=int, default=100) # parser.add_argument('--testBatchSz', type=int, default=2048) parser.add_.

转载 TORCHVISION OBJECT DETECTION FINETUNING TUTORIAL

For this tutorial, we will be finetuning a pre-trainedMask R-CNNmodel in thePenn-Fudan Database for Pedestrian Detection and Segmentation. It contains 170 images with 345 instances of pedestrians, and we will use it to illustrate how to use the new feat...

转载 Optimizing model Parameters

HyperparametersHyperparameters are adjustable parameters that let you control the model optimization process. Different hyperparameter values can impact model training and convergence rates (read moreabout hyperparameter tuning)We define the following.

转载 Build the NN & auto.grad

nn.FlattenWe initialize thenn.Flattenlayer to convert each 2D 28x28 image into a contiguous array of 784 pixel values ( the minibatch dimension (at dim=0) is maintained)nn.Sequentialnn.Sequentialis an ordered container of modules. The data is pa...

转载 Datasets & Dataloaders

PyTorch provides two data primitives:torch.utils.data.DataLoaderandtorch.utils.data.Datasetthat allow you to use pre-loaded datasets as well as your own data.Datasetstores the samples and their corresponding labels, andDataLoaderwraps an iterable a...

转载 Pytorch.org/tutorials Tensor

Initializing a tensortorch.tensortorch.from_numpyInput tensortorch.ones_like()retain the shapetorch.rand_like() retain the shapeInput the shapetorch.rand()torch.ones()torch.zeros()tensor.attributetensor.shapetensor.dtypeten.

转载 Pytorch.org QuickStart

import torch, torchvisionmodel = torchvision.models.resnet18(pretrained=True)data = torch.rand(1, 3, 64, 64)labels = torch.rand(1, 1000)prediction = model(data) # forward passloss = (prediction - labels).sum()loss.backward() # backward passloss .

原创 ucb DATA100 Note 3

answer1 = pd.unique(calls[calls['CVLEGEND'] == 'LARCENY']['OFFENSE']).tolist()python to list.tolist()pd.unique()Pandas VisualizationPandas offers basic functionality for plotting. For example, theDataFrameandSeriesclasses both have aplotme...

原创 Leetcode 60题了

2021/12/14Leetcode力扣必备算法知识和练习题｜手画图解版【持续更新ing】_哔哩哔哩_bilibiliB站的一个视频，感觉讲的挺好的双链表：普通双指针：同向对撞双指针： 有序链表快慢双指针二分查找：class Solution: def search(self, nums: List[int], target: int) -> int: if not nums: return -1 ...

原创 Missing Semester MIT (continuing)

an interesting courseThe Missing Semester of Your CS Education · the missing semester of your cs educationMy computer is Windows!!!!while the course maily teaches us how to use shell on Linux or MacOS.Gan! Wanna a brand-new computer with MacOS syste.

原创 ucb DATA100 Note 2

Data Cleaning & Visualization The process of transforming raw data to facilitate subsequent analysis Data cleaning often addresses issues structure / formatting missing or corrupted values unit conversion encoding t

原创 ucb DATA100 Note 1

Danger Zone! 有点搞笑怎么回事Exhilarating content！pandas 包里df.describe()value_counts()这课nb！！！ucb nb！！！！！！！！！！！！！！！！！！！！！ucb cs nb！！！！！！！！！！！！...

原创 cs学习地图

一个仅供参考的CS学习规划 - CS自学指南看到上面这个链接，来学一波必学工具：MIT missing semesterMIT-Missing-Semesterweb开发两周速成版​​​​​​MIT web development courseUCB DATA100 数据科学UCB Data100: Principles and Techniques of Data Science继续我的cs182！...

原创 Deep Learning cs182 Assignment 1

希望能写出编程作业！配置一下虚拟环境C:\Users\Yasmin\Downloads\hw1>cd .env/ScriptsC:\Users\Yasmin\Downloads\hw1\.env\Scripts>activate(.env) C:\Users\Yasmin\Downloads\hw1\.env\Scripts>deactivate碰到无法running build_extbuilding 'im2col_cython' extensioner

原创 Numpy Python

immutable: int, float, tuple, stringmutable: list, dict, class, set, numpy.ndarray>>> a = 1>>> b = a>>> b = 3>>> a1int值不会改变>>> a = {2,3}>>> b = a>>> b.add(4)>>> a...

原创 UCB cs61a Python中级进阶

*有过Python基础，但是到赋值、类、拷贝（深浅） 就搞不清啦！*通过UCB CS61A 来查缺补漏啦*长期更新Name Lookup Rules值得注意的是，到了class，object全局和局部变量就会变，之后应该会有。一些专有名词，学英文啦DoctestsSeems to be an automatic debugger...'''这个例子展示如何在源码中嵌入doctest用例。'>>>' 开头的行就是doctest测试...

原创 一半了Leetcode

原创 Leetcode 200 让我永远记住你

给你一个由'1'（陆地）和 '0'（水）组成的的二维网格，请你计算网格中岛屿的数量。岛屿总是被水包围，并且每座岛屿只能由水平方向和/或竖直方向上相邻的陆地连接形成。此外，你可以假设该网格的四条边均被水包围。1. 注意是char2.注意用DFS（回溯）class Solution: def numIslands(self, grid: List[List[str]]) -> int: m = len(grid) n = len(grid[0.

原创 朝着100题冲刺 leetcode

原创 Berkely cs182 环境根本装不来

死在了最开始。。。1.环境根本装不来只要有一步出错，人就没了2.打算买个Apple的电脑get_datasets.sh好像bash的文件和windows就是不匹配！好痛苦。。。。lec5 讲了backpropagation 和 CNNlec6 要讲 AlexNet GoogLeNet......好快。。。。计算机的课就是这样快吧主要还是要实践不code光听课有啥用但是我环境装不来咋办！！！！痛苦。...

原创 小笨霖英语口语2 抱怨

Our team sucks.The traffic here really sucks.Everything sucks.Remark:I don't know I can hear that word all over the place but I still can't get used to it.. in my opinion, that's worse than saying "fuck" or "shit" but that are all bad words anyway...

原创 小笨霖英语口语1 好棒

It's cool!Cool!It is neat!Neat! = very cool!It is righteous!少用，very very cool！It's good.Good for you. Good to hear.That's great.very goodThat's wonderful.very very good.That's incredible.She is pretty.She is be...

原创 Berkeley cs182 Lec5 Backpropagation

cs182和cs282应该是一起上课的which means 研究生和本科生读一门课但是，我们还是要慨叹一下人家本科生的教学质量。。。老师一周上三天，每次1h30min，助教section每周详细的reading material。。。详细的assignment，final project指导国内的教学和助教学学人家吧！！lec1-4基本上都是ML的东西，optimizer也是熟悉的东西，基本上都用adam完事了lec5之后就是深度学习了发现这课讲完了NLP

原创 Leetcode 70 爬楼梯 回溯

题目：假设你正在爬楼梯。需要n阶你才能到达楼顶。每次你可以爬 1 或 2 个台阶。你有多少种不同的方法可以爬到楼顶呢？注意：给定n是一个正整数。给爬楼梯写了个回溯其实列举所有方法，回溯肯定是可以的？其实回溯 = 暴力枚举关键是，爬楼梯这里只要count，不要详细的stepsclass Solution: def climbStairs(self, n: int) -> int: count = 0 ...

原创 Leetcode 碎碎念

1. 第一步看清题目意思2. 第二部看清输入（特别是输入，int还是char！）输出的数据类型！3. 从easy，medium开始做题现在在做Leetcode Hot100！有空就做两题！加油！

原创 Berkeley cs182 感受

第一次写csdn。Berkeley的cs182我感觉很cs，很intuitive，数学的推导不多。我不建议新手学，反而像我这样的半吊子，能悟到很多，包括老师讲到的和没讲到的。感觉实习都不要实习了，刷点cs的网课，很香。...