jgptmem
v0.0.1
Published
The JGPTMem class is designed to handle chat interactions utilizing the OpenAI GPT-3.5 Turbo model, with a specific focus on managing a short-term memory. This memory is intended to continuously condense, update, and store interactions between the user an
Downloads
1
Readme
JGPTMem Class
The JGPTMem
class is designed to facilitate conversations with OpenAI's GPT-3.5-turbo model, while managing a short-term memory to provide a more context-aware interaction.
Requirements
- The
openai
package. - Environment variables with the OpenAI API key and organization ID.
- Node.js version compatible with the
openai
package.
Constructor: new JGPTMem(apiKey, orgId)
Create a new JGPTMem
instance.
apiKey
(String): The OpenAI API key.orgId
(String): The organization ID.
Method: async talk(userInput)
Receives user input, constructs a prompt with short-term memory instructions, and returns the response from OpenAI's GPT-3.5-turbo model.
userInput
(String): The user's question or command.
Returns
- (String): The response from OpenAI's model, formatted as
{"condensedMemory":condensedSummary, "response": "your answer"}
.
Throws
- Error: If there's any error in processing the prompt or fetching the response.
Example Usage
const jgptMem = new JGPTMem(process.env.OPENAI_API_KEY, process.env.OPENAI_ORG_ID);
const userInput = "Tell me about the weather today";
jgptMem.talk(userInput).then(result => console.log(result));
This code snippet creates an instance of JGPTMem
, provides a user question, and prints the response.
Note
- Properly configure environment variables (
OPENAI_API_KEY
,OPENAI_ORG_ID
) in your.env
file. - The
talk
method returns a Promise. Handle it appropriately in your code. - Error handling is crucial to manage unexpected responses or issues during execution.
- The class emphasizes the importance of continuously updating and condensing short-term memory without significant loss of data.
JGPTMem
provides a sophisticated way to utilize OpenAI's GPT-3.5-turbo model for interactive conversations, considering previous interactions and maintaining a condensed short-term memory for more contextual responses.