OpenAIrequest(Userinp,AIhistory,RequestHistory)- returns the openAI Chat anwer for a single question |
 | % OpenAIrequest(Userinp,AIhistory,RequestHistory) - returns the openAI Chat anwer for a single question % (by Tim Lueth, VLFL-Lib, 2024-NOV-09 as class: ARTIFICIAL INTELLIGENCE) % % if called without an input the fnctn will establish a conversation % until "Cntr-C" is used (Status of: 2024-11-09) % % Introduced first in SolidGeometry 5.5 % % See also: jarvisChatSGLib, json2matlab, jsonplansequence, OpenAIcostAPI % % [AIansw,AIhistory,RequestHistory]=OpenAIrequest([Userinp,AIhistory,Requ % estHistory]) % === INPUT PARAMETERS === % Userinp: A charcter chain as input; default is empty % AIhistory: the struct that is used by addResponseMessage, to maintain a % chat history; default ist empty % RequestHistory: a cell list that describes the history of Userinp in % case of a conversation % === OUTPUT RESULTS ====== % AIansw: Answer to the last Userinp % AIhistory: AI history of response messages if a conversion is planned % RequestHistory: The history of Userinp if a a conversion is planned % % EXAMPLE: % OpenAIrequest('What is the core diameter of a DIN13 M4 nut') % return a text answer % OpenAIrequest % start a conversation loop % % See also: jarvisChatSGLib, json2matlab, jsonplansequence, OpenAIcostAPI % % % Copyright 2024 Tim C. Lueth |
jsonplansequence(Userinp,AIchat,AIhistory)- returns a plan step sequence from a single chat task |
 | % jsonplansequence(Userinp,AIchat,AIhistory) - returns a plan step sequence from a single chat task % (by Tim Lueth, VLFL-Lib, 2024-NOV-07 as class: ARTIFICIAL INTELLIGENCE) % % The original openAI/matlab interface is able to create sequences of % tasks, but the argument references are faulty, which is why a complex % task has to be explicitly broken down into a sequence and then % processed as a series of requests. (Status of: 2024-11-09) % % Introduced first in SolidGeometry 5.5 % % See also: jarvisChatSGLib, json2matlab, OpenAIrequest, OpenAIcostAPI % % [Userscript,AIhistory]=jsonplansequence(Userinp,AIchat,AIhistory) % === INPUT PARAMETERS === % Userinp: User input as one string % AIchat: AIchat % AIhistory: AI History % === OUTPUT RESULTS ====== % Userscript: plan sequence as single string % AIhistory: modified AI History % % EXAMPLE: % jarvisChatSGLib % what is typical diameter and height of coffe mug, create zylinder with this dimensions % % See also: jarvisChatSGLib, json2matlab, OpenAIrequest, OpenAIcostAPI % % % Copyright 2024 Tim C. Lueth |
json2matlab(SGfuncname,SGfuncargs,Variables)- convertes a json reaction of an openAI-Chat into a matlab code line |
 | % json2matlab(SGfuncname,SGfuncargs,Variables) - convertes a json reaction of an openAI-Chat into a matlab code line % (by Tim Lueth, VLFL-Lib, 2024-NOV-06 as class: ARTIFICIAL INTELLIGENCE) % % Robin Schregle used a switch structure with matlab code to implement % the SG-Lib fnctn calls. This requires the maintenance of the json code % for fnctns f..... and in addition the code for the fnctn calls in the % matlab code (see SGlib_Fnctn_Calling_OpenAI). Tim Lueth's solution uses % the JSON code of the answer directly to create the matlab code to be % executed by eval. % This fnctn is this decoding and evaluation of the matlab code, % extracted from the json answer. % % % % AIchat = openAIChat("gpt-4o",Temperature=0, % Tools=[f0,f1,f2,f3,f4,f5]); % SGlib fnctns f0..f5 as JSON % % AIhistory = messageHistory(); % % AIhistory = addUserMessage(AIhistory, Userinp); % Add a user % request % % [AItextansw, AIexecplan] = generate(AIchat, AIhistory); % generate % the chat answer % % AIhistory = addResponseMessage(AIhistory, AIexecplan); % add the % answer to the chat history % % SGlibcalls = AIexecplan.tool_calls; % find the % appropriate SGlib calls in the answer % % SGfuncname = SGlibcalls(k).fnctn.name; % can be an % array therefor use index k % % SGfuncargs = SGlibcalls(k).fnctn.arguments; % can be an % array therefor use index k % % SGresult= % % SGresultname = SGfuncname + string(length(AIhistory.Messages)); % One % possibility for naming % % Variables.(SGresultname) = SGResult; % create a % list of results for referencing % % % (Status of: 2024-11-09) % % Introduced first in SolidGeometry 5.5 % % See also: jarvisChatSGLib, jsonplansequence, OpenAIrequest, % OpenAIcostAPI % % [SGansw,cmdstr]=json2matlab(SGfuncname,SGfuncargs,Variables) % === INPUT PARAMETERS === % SGfuncname: Name of the current SG fnctn extracted from the AIexecplan % SGfuncargs: Struct containing the parameternames as fields and expected % values extracted from the AIexecplan % Variables: History of already used SGresults as cell list of % SGfuncnames with the call number % === OUTPUT RESULTS ====== % SGansw: evaluated result from the matlab code extracted from the json % answer. % cmdstr: the command string executed within json2matlab to create res % % See also: jarvisChatSGLib, jsonplansequence, OpenAIrequest, % OpenAIcostAPI % % % Copyright 2024 Tim C. Lueth |
OpenAIcostAPI- opens the web page of company openAI to show the users expenses |
 | % OpenAIcostAPI - opens the web page of company openAI to show the users expenses % (by Robin Schregle & Tim Lueth, VLFL-Lib, 2024-NOV-04 as class: % ARTIFICIAL INTELLIGENCE) % % 1) Download from Gitgub (see link below) the Large Language Model (LMM) % fnctns of Matlab and store the files in a folder in your Matlab code % directory using the % folder name [pwd filesep 'llms-with-matlab'] % 2) add this path to your main directory using % addpath([pwd filesep 'llms-with-matlab']) % 3) create an API account at openAI % you have to start an account store your credit card info (see link % below) % 4) create an environment ascii file with name "OPENAI_API_KEY.env" that % contains a single line % OPENAI_API_KEY=" % 5) load this environmentfile by % loadenv("OPENAI_API_KEY.env") % 6) set the model to % modelName = "gpt-4o"; % (Status of: 2024-11-09) % % Introduced first in SolidGeometry 5.5 % % See also: jarvisChatSGLib, json2matlab, jsonplansequence, OpenAIrequest % % OpenAIcostAPI % % Additional Hyperlinks: % Matlab's LLM at Github, OpenAI API-Key Generation % % EXAMPLE: % OpenAIcostAPI % % See also: jarvisChatSGLib, json2matlab, jsonplansequence, OpenAIrequest % % % Copyright 2024 Tim C. Lueth |
jarvisChatSGLib- this fnct/script is a ChatGPT based control of the SGLib and defines the development start of SGLib 6.0 |
 | % jarvisChatSGLib - this fnct/script is a ChatGPT based control of the SGLib and defines the development start of SGLib 6.0 % (by Robin Schregle & Tim Lueth, VLFL-Lib, 2024-NOV-04 as class: % ARTIFICIAL INTELLIGENCE) % % At the end of October 2024, Tim Lueth encouraged his team to work with % the Matlab/ChatGPT API interface, and Felix Pancheri and Robin Schregle % started to use it. Robin Schregle implemented a first version called % “SGlib_Fnctn_Calling_OpenAI”. This fnctn “jarvisChatSGLib” is an % improved version of “SGlib_Fnctn_Calling_OpenAI” to achieve similar % results to the already existing SGLib fnctn “jarvis” from January 2022, % which was programmed by Tim Lueth on the recommendation of Franz % Irlinger.The original fnctn "jarvis" was based on the use von Apple's % Siri and the SGLib's parser "siri2jarvis" % 1) Download from Gitgub (see link below) the Large Language Model (LMM) % fnctns of Matlab and store the files in a folder in your Matlab code % directory using the % folder name [pwd filesep 'llms-with-matlab'] % 2) add this path to your main directory using % addpath([pwd filesep 'llms-with-matlab']) % 3) create an API account at openAI % you have to start an account store your credit card info (see link % below) % 4) create an environment ascii file with name "OPENAI_API_KEY.env" that % contains a single line % OPENAI_API_KEY=" % 5) load this environmentfile by % loadenv("OPENAI_API_KEY.env") % 6) set the model to % modelName = "gpt-4o"; % (Status of: 2024-11-09) % % Introduced first in SolidGeometry 5.5 % % See also: chat, jarvis, SGlib_Fnctn_Calling_OpenAI, OpenAIcostAPI % % jarvisChatSGLib % % Additional Hyperlinks: % Matlab's LLM at Github, OpenAI API-Key Generation % % EXAMPLE: % jarvisChatSGLib % % See also: chat, jarvis, SGlib_Fnctn_Calling_OpenAI, OpenAIcostAPI % % % Copyright 2024 Tim C. Lueth |
