I am playing around with the LangChain JavaScript library and wanted to make sure I am doing this correctly.
My goals is to set the tone of the AI (who they are) and then pass in some context and let it ask a question about that context.
So this is what I do:
promptTemplate = new PromptTemplate({
template: templates.initial,
inputVariables: ["context"],
});
inputWithContext = await promptTemplate.format({
context: text,
});
const res = await model.call(inputWithContext);
Where templates.initial is:
CONTEXT: {context}
WHO YOU ARE: You are someone who gives back bad feedback.
QUERY: Given the context and who you are, give the user some feedback and end with a question.
First question:
Is that the correct way of doing this? I noticed there are some SystemMessage objects but not sure if I need those? If not how would you structure this?
