- 文章信息
- 作者: kaiwu
- 点击数:1034
From Wikipedia, the free encyclopedia
Generative Pre-trained Transformer 3 (GPT-3) is an autoregressive language model that uses deep learning to produce human-like text.
It is the third-generation language prediction model in the GPT-n series (and the successor to GPT-2) created by OpenAI, a San Francisco-based artificial intelligence research laboratory.[2] GPT-3's full version has a capacity of 175 billion machine learning parameters. GPT-3, which was introduced in May 2020, and was in beta testing as of July 2020,[3] is part of a trend in natural language processing (NLP) systems of pre-trained language representations.
This is a real example, see https://www.ambrogiorobot.com/en. Disclosure: LF owns one.
See https://en.wikipedia.org/wiki/ELIZA. A classic book still worth reading on the ELIZA effect and AI in general is (Weizenbaum 1976). In 2014 some people claimed, mistakenly, that a chatbot had passed the test. Its name is “Eugene Goostman”, and you can check it by yourself, by playing with it here: http://eugenegoostman.elasticbeanstalk.com/. When it was tested, I was one of the judges, and what I noticed was that it was some humans who failed to pass the test, asking the sort of questions that I have called here “irreversible”, such as (real examples, these were asked by a BBC journalist) “do you believe in God?” and “do you like ice-cream”. Even a simple machine tossing coins would “pass” that kind of test.
See for example the Winograd Schema Challenge (Levesque et al. 2012).
For an excellent, technical and critical analysis, see McAteer (https://matthewmcateer.me/blog/messing-with-gpt-3/ ." data-track="click" data-track-action="reference anchor" data-track-label="link" data-test="citation-ref" aria-label="Reference 2020">2020). About the “completely unrealistic expectations about what large-scale language models such as GPT-3 can do” see Yann LeCun (Vice President, Chief AI Scientist at Facebook App) here: https://www.facebook.com/yann.lecun/posts/10157253205637143.
The following note was written by the journalists, not the software: “[…] GPT-3 produced eight different outputs, or essays. Each was unique, interesting and advanced a different argument. The Guardian could have just run one of the essays in its entirety. However, we chose instead to pick the best parts of each, in order to capture the different styles and registers of the AI. Editing GPT-3’s op-ed was no different to editing a human op-ed. We cut lines and paragraphs, and rearranged the order of them in some places. Overall, it took less time to edit than many human op-eds.” (GPT-3 2020).
For some philosophical examples concerning GPT-3, see http://dailynous.com/2020/07/30/philosophers-gpt-3/.
For a more extended, and sometimes quite entertaining, analysis see (Lacker https://lacker.io/ai/2020/07/06/giving-gpt-3-a-turing-test.html ." data-track="click" data-track-action="reference anchor" data-track-label="link" data-test="citation-ref" aria-label="Reference 2020">2020).
For an interesting analysis see (Elkins and Chun 2020).
- 文章信息
- 作者: kaiwu
- 点击数:3242
软件名称:Scimago Graphica
官方网站:www.graphica.app
开发机构:SCImago Lab (www.scimagolab.com)
软件大小:264兆
更新日期:2021-9-30
A new way to explore, visually communicate and make sense of data. Scimago Graphica is completely free, but if you need even more features, we are working on a pro version.
Scimago Graphica作为Scimago Journal & Country Rank(SJR期刊排名)的数据可视化工具(www.scimagojr.com/viztools.php),完全免费,支持windows,mac和linux系统,目前提供了30种图形和示例数据,支持交互功能,很多方面已经很接近商业软件tableau了(www.tableau.com)。
- 文章信息
- 作者: kaiwu
- 点击数:603
https://gs.statcounter.com/browser-market-share/desktop/worldwide/#monthly-200901-202301
The market share of web browsers for desktop.
全球桌面电脑网页浏览器软件的市场份额,数据来自statcounter.com
1.Google chrome
https://www.google.com/chrome/
https://pc.qq.com/detail/1/detail_2661.html
2.Microsoftware IE,Edge_Legacy,Edge
https://www.microsoft.com/edge
3.Firefox
https://www.mozilla.org/en-US/firefox/new/
http://www.firefox.com.cn/
4.Apple Safari
https://www.apple.com.cn/safari/
5.Opera
https://www.opera.com/browsers/opera
- 文章信息
- 作者: kaiwu
- 点击数:654
R Notebook
1.参考资料
1.1 Modern optimization with R
Cortez, P. (2021). Modern optimization with R.Springer.
cover_modern_optimization_with_R.jpg
1.2 Modeling and Solving Linear Programming with R
1.3 Linear Programming with R: Exploring the “lpSolve” R package
Roberto Salazar,Nov 17, 2019
Linear Programming with R: Exploring the “lpSolve” R package
1.4 lpSolveAPI Package Users Guide by Kjell Konis
2. prepraration
2.1 import libraries
# Import lpSolve package
library(lpSolve)
library(XLConnect)
2.2 the problem: 班次安排问题
## 2.3 连接Excel文件
mybook<-loadWorkbook("D:/kedu/teaching_datasets/Excel_model/sm_solver.xls")
rsheets<-getSheets(mybook)
rsheets
[1] "table-chair" "beef" "transport" "staff"
3. linear programming
3.1 Set coefficients of the objective function
设定目标函数,因为是求和,所以矩阵是1,1,1,1,1,1 matrix在这里非常重要 研究一下如何生成重复的相同元素
#f.obj <- c(4, 2)
f.obj <-as.matrix(rep(1,each=7))
3.2 Set matrix corresponding to coefficients of constraints by rows
Do not consider the non-negative constraint; it is automatically assumed
就是员工工作的可用矩阵 注意:这里面通常包含标题,默认header=TRUE,那么startRow=7
#f.con <- matrix(c(5, 15,20, 5), nrow = 2, byrow = TRUE)
f.con <-as.matrix(readWorksheet(mybook, sheet = "staff", startRow = 8, endRow = 14,
startCol = 17, endCol = 23,header=FALSE))
3.3 Set unequality signs
设置不等式符号 可以每行设置不同的符号
f.dir <- as.matrix(rep(">=",each=7))
3.4 Set right hand side coefficients
约束条件的范围
#f.rhs <- c(50,40)
f.rhs <-as.matrix(readWorksheet(mybook, sheet = "staff", startRow = 8, endRow =14,
startCol = 27, endCol = 27,header=FALSE))
3.5 设定变量取整数 f.intvec,例如f.intvec <- c(1,2)表示x1,x2取整数
如果自变量全都是整数,那么 all.int = TRUE就可以了
#f.intvec <- c(1,2)
3.6 Final value (z)
计算结果
#report_lp<-lp("min", f.obj, f.con, f.dir, f.rhs,int.vec = f.intvec)
# 部分变量取整数:int.vec = f.intvec
report_lp<-lp("min", f.obj, f.con, f.dir, f.rhs, all.int = TRUE)
report_lp
Success: the objective function is 9
report_lp$objval
[1] 9
# output the final value
writeWorksheet(mybook,report_lp$objval,sheet = "staff", startRow =16,startCol = 25,header = FALSE)
saveWorkbook(mybook)
3.7 Variables final values
report_solution<-report_lp$solution
report_solution
[1] 0 1 0 0 1 3 4
## 保存变量取值
## 结果是一个矩阵(1列),所以为了在excel变为1行,需要转置,t()
writeWorksheet(mybook,t(report_solution),sheet = "staff", startRow =16,startCol = 17,header = FALSE)
saveWorkbook(mybook)
3.8 Sensitivities
敏感度分析
report_lp<-lp("min", f.obj, f.con, f.dir, f.rhs, all.int = TRUE,compute.sens=TRUE)
report_lp$sens.coef.from
[1] 1e+00 -1e+30 0e+00 0e+00 0e+00 0e+00 0e+00
report_lp$sens.coef.to
[1] 1.000000e+30 1.000000e+30 1.333333e+00 1.000000e+00 1.333333e+00
[6] 1.333333e+00 1.333333e+00
3.9 Dual Values (first dual of the constraints and then dual of the variables)
Duals of the constraints and variables are mixed
report_lp$duals
[1] 0 1 0 0 0 0 0 0 1 0 0 0 0 0
3.10 Duals lower and upper limits
report_lp$duals.from
[1] -1e+30 -1e+30 7e+00 6e+00 6e+00 5e+00 4e+00 -1e+30 -1e+30 -1e+30
[11] -1e+30 -1e+30 -1e+30 -1e+30
report_lp$duals.to
[1] 1.000000e+30 1.000000e+30 8.333333e+00 8.666667e+00 8.666667e+00
[6] 9.000000e+00 9.333333e+00 1.250000e+00 2.500000e-01 1.000000e+30
[11] 1.000000e+30 1.000000e+30 1.000000e+30 1.000000e+30