English | 简体中文

Pandora ChatGPT API Integration for Langchain (customize-longchain-LLm.py)

This project saves the cost of using the GPT-4 API by integrating the Pandora project to call the ChatGPT web version as an API and using it in conjunction with Langchain's custom language model.

Features

• Call the ChatGPT web version's API
• Use custom. language model to implement ChatGPT's functionality
• Optional stop list support

Code examples

from langchain.callbacks.manager import CallbackManagerForLLMRun
from langchain.llms.base import LLM
import uuid
import requests
import json
from pydantic import BaseModel
import os
from tool import other_tools

class CustomChatGPT(LLM).
    ...

if __name__ == "__main__".
    llm = CustomChatGPT(base_url="your url") # need to build the pandora application on the server itself
    response = llm("This is a foobar thing")

How to use

1. Deploy the Pandora project to your server. 2.
2. Replace your url with your server URL. 3.
3. Instantiate the CustomChatGPT class. 4.
4. Call the _call method (also known as llm("prompt")), pass in the relevant parameters and get the returned message.

Caution

• Make sure your server has successfully deployed the Pandora project.
• When using the _call method, make sure you pass in the correct parameters, such as prompt, stop, etc.

Pandora ChatGPT API Integration for Langchain - additional content (pandora-usage.py)

This project not only provides integration with Pandora, but also introduces basic usage of Pandora.

Pandora Basic Usage

The following code provides some basic Pandora usage, including

• List available models
• List sessions in a paginated manner
• Deleting all sessions
• Get the details of a given session by its session ID
• Deleting a specified session by session ID
• Set the specified session title by session ID
• Automatically generate a title for a specified new session
• Have ChatGPT regenerate replies
• Allow ChatGPT to continue the previous recovery
• Ask ChatGPT a question and wait for a reply

import requests
import json
import uuid

# base_url = "https://your-api-domain" # Replace with your API domain name

# List the models available for the account
def list_models().
    ...

# List the conversations in a paginated manner
def list_conversations(offset=1, limit=20).
    ...

# Delete all conversations
def delete_all_conversations().
    ...

# Get the details of a given session by its session ID
def get_conversation(conversation_id).
    ...

# Delete the specified conversation by its session ID
def delete_conversation(conversation_id).
    ...

# Set the specified session title by the session ID
def set_conversation_title(conversation_id, title).
    ...

# Automatically generate the title of the specified new session
def generate_conversation_title(conversation_id, model, message_id).
    ...

# Make ChatGPT regenerate replies
def regenerate_chatgpt(prompt, model, message_id, parent_message_id, conversation_id, stream=True).
    ...

# Let ChatGPT continue with the previous recovery
def continue_chatgpt(model, parent_message_id, conversation_id, stream=True).
    ...

# Ask ChatGPT a question and wait for a reply
def ask_chatgpt(prompt, model, message_id, parent_message_id, conversation_id=None, stream=False).
    ...

def ask().
    ...

def main().
    ...

if __name__ == "__main__".
    main()

Usage

1. Replace base_url with the domain name of your API.
2. Call the corresponding function to perform the corresponding operation.

Caution

• Please make sure your API domain name is configured correctly.
• When calling the function, please make sure to pass the correct parameters.

Contributors

Feel free to contribute to this project! Please submit an issue or pull request on GitHub.

Translated with www.DeepL.com/Translator (free version)