learning summary fourth the

概述增强式学习(Reinforcement Learning)

概述增强式学习(Reinforcement Learning) Supervised Learning(自监督学习):告诉机器输入和输出,用有标注的训练资料训练出的Network Reinforcement Learning(增强式学习):给机器一个输入,我们不知道最佳输出是什么(适用于标注困难或者 ......
Reinforcement Learning

【大联盟】20230701 传送(b) QOJ1878 【No Rest for the Wicked】

## 题目描述 [here](https://qoj.ac/problem/1878)。 ## 题解 考虑一条路径上只有 $a$ 的前缀 $\max$ 才是有用的,不妨考虑按照前缀 $\max$ 来划分。可以发现,这些连续段直接存在单向边连接。 现在,我们考虑如何求出这些连续段。一个点 $i$ 可以 ......
大联盟 20230701 Wicked 1878 Rest

[ABC310G] Takahashi And Pass-The-Ball Game

Problem StatementThere are $N$ Takahashi. The $i$-th Takahashi has an integer $A_i$ and $B_i$ balls. An integer $x$ between $1$ and $K$, inclusive, wi ......
Pass-The-Ball Takahashi 310G Pass Ball

ERROR 1709 (HY000): Index column size too large. The maximum column size is 767 bytes.

MySQL版本5.6.35 在一个长度为512字符的字段上创建unique key报错 CREATE DATABASE dpcs_metadata DEFAULT CHARACTER SET utf8; select * from information_schema.SCHEMATA; + + + ......
column size maximum ERROR Index

LLM is all you need for the backend

一、LLM给软件开发范式带来了什么改变? 人们一直在说Github Copilot将取代程序员。我们认为这是错误的。我们已经有了类似GPT-4这种强大的LLM模型,却还要把自己限制在编写传统代码上吗?不!所有代码都有bug! 代码不是对业务逻辑进行编码的理想方式,代码必须经过审查,并且它按照程序员的 ......
backend need LLM all for

what is the retrievers in LangChain?

see: Retrievers | 🦜️🔗 Langchain https://blog.langchain.dev/retrieval/ 按文档说的Retriever就是一个接口,可以把和query相关的文档都查询出来,而vectorDB是否成retriever的关键组件之一。 ......
retrievers LangChain what the is

Measuring and Narrowing the Compositionality Gap in Language Models

[TOC] > [Press O., Zhang M., Min S., Schmidt L., Smith N. A. and Lewis M. Measuring and narrowing the compositionality gap in language models. arXiv p ......

Kingbase8: was aborted: ERROR: sorry, too many global temporary relations in backend already. Call getNextException to see other errors in the batch.

==额....今天踩了一个坑,找个半天找到的解决方法== 设置一个会话内最多允许的全局临时表数据和索引的最大文件数量,默认值为32,最小值为1,最大值为1024。这个文件数量并不是实际磁盘看到的文件数目,而是一些依附于这个表的扩展关系文件,比如索引、大对象等。一般情况下,一个全局临时表只会占用参数中 ......

C# 使用EPPlus 操作excel The given key '8' was not present in the dictionary.

使用EPPlus 删除excel中某一个sheet中的几列的时候,出现了The given key '8' was not present in the dictionary.的报错; 最开始的写法,是从前往后删除,出现错误//ExcelWorksheet sheet = package.Workb ......
dictionary present EPPlus excel given

安装 MySQL for Windows 时报错 The configuration for MySQL Server 8.0.34 has failed. You can find more information about the failures in the 'Log' tab. 解决方法

今天在安装 MySQL for Windows 时报错 ```txt The configuration for MySQL Server 8.0.34 has failed. You can find more information about the failures in the 'Log' ......
MySQL configuration information for the

解决./composer.json" does not match the expected JSON schema

如果执行composer install的时候报错 "./composer.json" does not match the expected JSON schema: - name : Does not match the regex pattern ^[a-z0-9]([_.-]?[a-z0-9 ......
composer expected schema match JSON

CF1083F The Fair Nut and Amusing Xor

#### 简要题意: 给你两个序列 $a,b$,一次操作可以将 $a$ 的某一个长度为 $k$ 的子区间全部异或上任意值,$f(a,b)$ 为使得 $a$ 和 $b$ 相同的最少的操作数量。 支持单点修改 $a,b$,并在开头和每次修改后输出 $f(a,b)$ 的值。 $n,k,q\le 2\tim ......
Amusing 1083F 1083 Fair The

CF1361E James and the Chase

#### Description 给定一个有 $n$ 个点 $m$ 条边的**有向强连通图**。称一个点是**好的**当且仅当它到其他点都有且只有一条**简单路径**。如果好的点至少有 $20\%$ 则输出所有好的点,否则输出 `-1`。 $\sum n\leq 10^5,\sum m\leq 2\ ......
1361E James Chase 1361 and

CF1152F2 Neko Rules the Catniverse (Large Version) 题解

发现挨位考虑填哪个不太现实,考虑值域。 令 $dp_{i,j,st}$ 表示考虑到 $i$,此时序列长度为 $j$,$i-m$ 到 $i-1$ 填空状态为 $st$ 的方案数,考虑选/不选数即可: $dp_{i,j,st}\times (\text{popcount}(st)+1)\to dp_{i ......
题解 Catniverse Version 1152F Large

arcgis注册postgresql失败。postgres be sure the databse client software is installed and configured correct

解决: 将PostgreSQL\9.1\bin目录下的Libpq.dll,Libeay32.dll, Libintl-8.dll, Ssleay32.dll复制一份放在ArcGIS Server安装目录的Server\bin***意:这里拷贝的文件需要都是64位的。 重启arcgis server服 ......

NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running.

###问题提示: ` NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and runn ......

【863】Calculate records based on the same value

Suppose we have a dataframe, it has a column of "country". It lists different names of country's names, and for one country maybe it has multiple reco ......
Calculate records based value same

Linux系统Apache报错httpd: Could not reliably determine the server's fully qualified domain name, using localhost.localdomain. Set the 'ServerName' directive globally to suppress this message

解决办法: 在配置文件中加一句ServerName localhost:端口号 # echo "ServerName localhost:8068" >> /etc/httpd/conf/httpd.conf 重启Apache即可解决。 ......

machine learning-2023-07-19

questions【链接】 │ │──math │ │──线性回归 │ │──逻辑回归 │ └──梯度下降 │ │──python │ │──numpy(科学计算库) │ │──pandas(数据分析处理库) │ │──matplotlib(数据可视化库) │ └──scikit-learn(机器学 ......
learning machine 2023 07 19

HTTP/2 stream 1 was not closed cleanly before end of the underlying stream解决

通过git clone 文件时报错 HTTP/2 stream 1 was not closed cleanly before end of the underlying stream 解决: git config --global http.versioin HTTP/1.1 重新git clon ......
stream underlying cleanly closed before

Learn about some useful truck diagnostic scanner tools

Have you ever experienced the frustration of unexpected breakdowns with your truck? Or maybe you’re tired of paying expensive diagnostic fees at your ......
diagnostic scanner useful Learn about

升级EF7连接SQL server出错SqlException: A connection was successfully established with the server, but then an error occurred during the login process. (provider: SSL Provider, error: 0 - 证书链是由不受信任的颁发机构颁发的。)

今天把项目里的Microsoft.EntityFrameworkCore.SqlServer和Microsoft.EntityFrameworkCore.Tools从6.0.6升级到了最新的7.0.9。一运行程序出错了。 ![img](https://img2023.cnblogs.com/blog ......

Learning hard C#学习笔记——读书笔记 07

## 1.值类型和引用类型 ### 1.1 什么是值类型和引用类型 * 值类型:包括简单类型,枚举类型,结构体类型等,值类型通常被分配在线程的堆栈上,变量保存的内容就是实例数据本身 * 引用类型:引用类型实例则被分配在托管堆上,变量保存的是实例数据的内存地址,引用类型主要包括类类型、接口类型、委托类 ......
笔记 Learning hard 07

Linux系统Apache添加监听端口后无法启动服务并报错:Job for httpd.service failed because the control process exited with error code. See "systemctl status httpd.service" and "journalctl -xeu httpd.service" for details.

导言:这是SE Linux安全机制导致的。 解决方法: 1.查看当前httpd端口 # semanage port -l|grep http 2.将对应端口加入SE Linux,以8068为例 # semanage port -a -t http_port_t -p tcp 8068 3.再次查看 ......
service quot httpd 端口 journalctl

大语言模型的预训练4:指示学习Instruction Learning详解以及和Prompt Learning,In-content Learning区别

# 大语言模型的预训练[4]:指示学习Instruction Learning:Entailment-oriented、PLM oriented、human-oriented详解以及和Prompt Learning,In-content Learning区别 # 1.指示学习的定义 Instruct ......

大语言模型的预训练[5]:语境学习、上下文学习In-Context Learning:精调LLM、Prompt设计和打分函数设计以及ICL底层机制等原理详解

大语言模型的预训练[5]:语境学习、上下文学习In-Context Learning:精调LLM、Prompt设计和打分函数设计以及ICL底层机制等原理详解 ......
上下文 语境 底层 In-Context 函数

LOADING Redis is loading the dataset in memory

MISCONF Redis is configured to save RDB snapshots, but is currently not able to persist on disk. Commands that may modify the data set are disabled. P ......
LOADING dataset loading memory Redis

Learning hard C#学习笔记——读书笔记 06

本文介绍了 C# 中委托的定义、使用、为什么引入委托以及委托的本质。同时,还介绍了委托链的使用,即将多个委托链接在一起,提高程序的可扩展性。 ......
笔记 Learning hard 06

[LeetCode] 2597. The Number of Beautiful Subsets

You are given an array nums of positive integers and a positive integer k. A subset of nums is beautiful if it does not contain two integers with an a ......
Beautiful LeetCode Subsets Number 2597

《DeepChain: Auditable and Privacy-Preserving Deep Learning with Blockchain-based Incentive》论文笔记

本文的研究背景: 在各种机器学习任务中,深度学习可以实现比传统机器学习算法更高的精度。最近,保护隐私的深度学习引起了信息安全界的极大关注,其中训练数据和训练模型都不会被暴露。联合学习是一种流行的学习机制,其中多方将局部梯度上传到服务器,服务器使用收集的梯度更新模型参数。然而,在联合学习中存在许多被忽 ......