model train the fit
sklearn模块中的preprocessing、model_selection、feature_selection
sklearn模块中的preprocessing、model_selection、feature_selection，包括分类数据处理、模型选择、参数选择、特征选择。 ......
java.sql.SQLSyntaxErrorException: You have an error in your SQL syntax; check the manual that corresponds to your MySQL server version for the right syntax to use near ''',b_5='17',b_6='' wher
java.sql.SQLSyntaxErrorException: You have an error in your SQL syntax; check the manual that corresponds to your MySQL server version for the right s ......
Unlock the Power of High-Performance Networking with the IPQ9554
Unlock the Power of High-Performance Networking with the IPQ9554 In today's world, reliable and high-speed internet connectivity is essential for ever ......
【快应用】表单组件的model属性实际运用
【关键词】 表单组件、model 【问题背景】 在使用表单组件时，往往需要同时绑定值和 change 事件动态更新数据，当表单交互较多的场景下会有大量与业务无关的代码。 快应用从1100版本开始引入 model 指令，使用 model 指令可以简化代码逻辑，在框架内部对值进行绑定与更新，实现双向绑定 ......
Vicuna-13B, an open-source chatbot trained by fine-tuning LLaMA
一、项目背景 We introduce Vicuna-13B, an open-source chatbot trained by fine-tuning LLaMA on user-shared conversations collected from ShareGPT. Preliminary ......
论文解读（VAT）《Virtual Adversarial Training: A Regularization Method for Supervised and Semi-Supervised Learning》
论文信息 论文标题：Virtual Adversarial Training: A Regularization Method for Supervised and Semi-Supervised Learning论文作者：Takeru Miyato, S. Maeda, Masanori Koya ......
Ray Tracing: The Next Week
2. Motion Blur 在前面，反走样是通过像素内取多个路径实现的，此外多条路径的选择也跟后面的漫反射、模糊反射、散焦模糊等一系列随机过程有关，如果继续暴力解法，也可以实现运动模糊。 在真实的相机中，照片的形成是对一段时间内光线的记录，与快门有关，因此，为了模拟摄影，加入运动模糊。 首先运动模 ......
论文解读《Do We Need Zero Training Loss After Achieving Zero Training Error?》
论文信息 论文标题：Do We Need Zero Training Loss After Achieving Zero Training Error?论文作者：Takashi Ishida, I. Yamane, Tomoya Sakai, Gang Niu, M. Sugiyama论文来源：20 ......
论文解读（PGD）《Towards deep learning models resistant to adversarial attacks》
论文信息 论文标题：Towards deep learning models resistant to adversarial attacks论文作者：Aleksander Madry, Aleksandar Makelov, Ludwig Schmidt, Dimitris Tsipras, Ad ......
猛读论文13 |【CVPR 2022 UDA】Unleashing Potential of Unsupervised Pre-Training with Intra-Identity Regularization for Person Re-Identification
动机 解决（1）对比学习管道中的增强通常会扭曲人物图像中的判别线索（2）细粒度的局部特征人物图像尚未得到充分探索。 思路 方法 ......
The 1st Universal Cup. Stage 12: Ōokayama
G 容斥完之后发现要求一个m次多项式的n次方，并且得到$n\times m$项。 原本很sb地直接套了个多项式LnExp上去（即使知道大概率过不了），然后狂TLE。。。 其实但凡从常数的角度分析，Exp的常数有14倍，已经比$log(m)$大了，所以不如写快速幂，然后写着就会发现卷积的长度总和其实是 ......
Leader of the Opinion Leaders
题目 Weibo is known as the Chinese version of Twitter. One user on Weibo may have many followers, and may follow many other users as well. According to ......
The Second Run of Quicksort
代码 #include<iostream> #include<vector> #include<cstdio> using namespace std; const int maxn = 100010; vector<int> sequence; int maxL, minR, cnt, k, n, ......
D. Solve The Maze
D. Solve The Maze 大意:建墙让所有好人可以到达坐标$(n,m)$，任何一个坏人都不能到达坐标$(n,m)$ 分析： 把坏人直接关起来，在坏人的四面建墙， 统计好人的人数 从坐标$(n,m)$去遍历，整个地图，看能不能遇到所有好人 3可以通过dsu， flood-fill等去做吧 细 ......
How fast are your disks? Find out the open source way, with fio
https://arstechnica.com/gadgets/2020/02/how-fast-are-your-disks-find-out-the-open-source-way-with-fio/ Storage benchmarking—much like Wi-Fi benchmarki ......
docker报错，ERROR: Got permission denied while trying to connect to the Docker daemon socket
使用docker，避免使用sudo命令 1.本地环境是Ubuntu18.04，docker version 19.03.12 以普通用户启动docker时，会报一下错误 Got permission denied while trying to connect to the Docker daemo ......
文章推荐---ChatGPT踩坑(too many signups from the same IP)
http://681314.com/A/S1A6pDeCNF 这个文章里关于注册写得不错，需要注意的地方有两点。 第一，最好在港澳台地区或类型的地方访问提供的网站，不然无法登录 第二，虚拟短信号码，最好选大国的，小国的我没收到 ......
some information about the origin of the OPT folder
In the old days, "/opt" was used by UNIX vendors like AT&T, Sun, DEC and 3rd-party vendors to hold "Option" packages; i.e. packages that you might hav ......
迁移学习（PAT）《Pairwise Adversarial Training for Unsupervised Class-imbalanced Domain Adaptation》
论文信息 论文标题：Pairwise Adversarial Training for Unsupervised Class-imbalanced Domain Adaptation论文作者：Weili Shi, Ronghang Zhu, Sheng Li论文来源：KDD 2022论文地址：dow ......
【题解】[ABC248G] GCD cost on the tree
「八云紫」无数次痛苦地询问，为什么我们还活着？ ……而「古明地恋」从不会回答。 恋恋闭上了觉之眼。 思路 容斥 + dp. $\gcd$ 相关，考虑 $\mu$ 反演或者 $\varphi$ 反演。 本质上都和容斥差不多，不如直接一步到位考虑容斥。 把权值拆成 $\gcd$ 和对应的方案数两部分，考 ......
Uva--122 Trees on the level(二叉树的层次遍历)
记录 23:27 2023-4-20 https://onlinejudge.org/external/1/122.pdf reference:《算法竞赛入门经典第二版》例题6-7 二叉树的层次遍历，这里是直接复制了作者的代码。(之前在我的数据结构学习里面手写过树、二叉树、AVL树(说是手写，其实也 ......
The Second Type of Uncertainty in Monte Carlo Tree Search
**发表时间：**2020 **文章要点：**MCTS里通常通过计算访问次数来做探索，这个被称作count-derived uncertainty。这篇文章提出了第二种uncertainty，这种uncertainty来源于子树的大小，一个直觉的想法就是，如果一个动作对应下的子树小，那就不用探索那么 ......
Layer-Dependent Importance Sampling for Training Deep and Large Graph Convolutional Networks
Zou D., Hu Z., Wang Y., Jiang S., Sun Y. and Gu Q. Layer-dependent importance sampling for training deep and large graph convolutional networks. NIPS, ......
1094 The Largest Generation
A family hierarchy is usually presented by a pedigree tree where all the nodes on the same level belong to the same generation. Your task is to find t ......
Find the sum of the series : 1.n + 2.(n – 1) + 3.(n – 2) + ….. + (n – 1).2 + n.1.
https://www.vedantu.com/question-answer/find-the-sum-of-the-series-1n-+-2n-1-+-3n-2-+-+-class-11-maths-cbse-5f34d4da5597f32583a5adca ......
Git提交时提示The file will have its original line endings in your working directory的解决方法
同事新建了一个仓库，git add的时候总是会提示 warning: LF will be replaced by CRLF in public/tinymce/README.md.The file will have its original line endings in your workin ......
Unlock the Power of High-Performance Networking with the IPQ9554
In today's world, reliable and high-speed internet connectivity is essential for everything from online gaming and streaming to remote work and learni ......
Arrange the Numbers UVA - 11481
求 1∼n 的排列 A 中，满足前 m 个数中，刚好有 K 个数使得 A[ i ]=i 的 AA 的个数。 错位排列 #include<bits/stdc++.h> using namespace std; const int mod=1e9+7; #define int long long int ......
LLaMA（Open and Efficient Foundation Language Models）学习
一、论文学习 原文链接：https://arxiv.org/pdf/2302.13971.pdf 0x1：引言 在大语料上训练得到的大语言模型(LLM)已经展示了它们的实力，在few-shot特性展现之后，有一系列的工作都关注于提高这些模型的规模。它们的努力是基于更多的参数可以带来更好的性能的假设之 ......