3.5-turbo API Call structure

[appatalks]

With "gpt-3.5-turbo", need to relook at structure for proper implementation.

Ex:
[
{“role”: “system”, “content”: “You are a helpful assistant that translates English to French.”},
{“role”: “user”, “content”: ‘Translate the following English text to French: “{text}”’}
]

Currently just using user and content line to pass. (easiest to get working after release of gpt-3.5-turbo)

https://platform.openai.com/docs/guides/chat/chat-vs-completions

[appatalks]

"In general, gpt-3.5-turbo-0301 does not pay strong attention to the system message, and therefore important instructions are often better placed in a user message."

This seems pretty obvious now :/

[appatalks]

I couldn't (yet) figure out how to pass an array that is actively updated to messages when user/bot interaction happens.
So, atm I am passing all the history back, but having assistant content mention to know its all part of the previous chat. Seems to work'ish.

messages: [
      { role: 'system', content: "You are Eva. You have access to previous chats and responses. You will keep conversation to a minimum and answer to the best of your abilities." },  // Doesn't seem to stick well.
      { role: 'user', content: selPers.value + "My next question is: " + sQuestion.replace(/\n/g, '') }, 
      { role: 'assistant', content: "Here are all my previous responses for you to remember: " + userMasterResponse.replace(/\n/g, ' ') }, 
],

[appatalks]

Pretty happy with this or a variation of it:

messages: [
      { role: 'system', content: "You are Eva. You have access to previous chats and responses. You will keep conversation to a minimum and answer to the best of your abilities." }, 
      { role: 'user', content: selPers.value }, 
      { role: 'assistant', content: "Here are all my previous responses for you to remember: " + userMasterResponse.replace(/\n/g, ' ') }, 
      { role: 'user', content: "My next response is: " + sQuestion.replace(/\n/g, '') }, 
],

[appatalks]

Reopening. I feel I am close to getting this into an active array:

///
// Set initial Messages payload
let iMessages = [
{ role: 'system', content: "You are Eva. You have access to previous chats and responses. You will keep conversation to a minimum and answer to the best of your abilities." },
{ role: 'user', content: selPers.value },
// { role: 'assistant', content: "Here are all my previous responses for you to remember: " + userMasterResponse.replace(/\n/g, ' ') },
// { role: 'user', content: "My next response is: " + sQuestion.replace(/\n/g, '') },
];

// Store messages in local storage
// localStorage.setItem("messages", JSON.stringify(iMessages));

// Retrieve messages from local storage
var cStoredMessages = localStorage.getItem("messages");
// kMessages = cStoredMessages ? JSON.parse(cStoredMessages) : [];
kMessages = iMessages + cStoredMessages ? JSON.parse(cStoredMessages) : [];

// Append last responses to next payload
// Placer
iMessages.push({ role: 'assistant', content: "Here are all my previous responses for you to remember: " + userMasterResponse.replace(/\n/g, ' ') });
iMessages.push({ role: 'user', content: "My next response is: " + sQuestion.replace(/\n/g, '') });

// Store the updated messages array back to local storage
localStorage.setItem("messages", JSON.stringify(iMessages));
///

// API Payload
var data = {
    model: sModel,
    messages: kMessages,
    max_tokens: iMaxTokens,
    temperature:  dTemperature,
    frequency_penalty: eFrequency_penalty,
    presence_penalty: cPresence_penalty,
    stop: hStop
}

[appatalks]

Woot Woot got memeory down. Per community, next goal:

ruby_coder
Regular
(https://community.openai.com/t/gpt-3-5-turbo-vs-text-davinci-003-chatbot/82806/73?u=sbennettkorea)
Very happy atm.
Good for you, @sbennettkorea
It’s great to hear you are enjoying coding.
Next, you can work on code to prune the messages array when it gets too large.

[appatalks]

Added a Clear Memory button for the time being. I'll revisit this.

[appatalks]

You know what, I am happy with this. Ill prolly reopen this if I ever run into a limits issue.