PerplexityChat Model Wrapper for Langchain

Overview

The pplx.py module acts as a wrapper for handling perplexity chat model in Langchain projects.

Requirements

Install the required packages listed in requirements.txt:

pip install -r requirements.txt

Configuration

1. Add the following import in your main Langchain application:

   from pplx import PerplexityChat

2. Update your config.json to include any specific settings required for the pplx module.

3. In your main Langchain application, initialize the PerplexityChat class and use it as per your needs.

Usage

Here's a simple example using pplx.py in a Langchain application:

from langchain.agents import initialize_agent, AgentType, load_tools from langchain.callbacks import get_openai_callback from pplx import PerplexityChat

Instance of the PerplexityChatAI class

chat_perplexity_ai = PerplexityChat(model_name="mistral-7b-instruct", temperature=0.7, verbose=True)

Load the tools

tools = load_tools(["serpapi"], llm=chat_perplexity_ai)

Initialize the agent

agent = initialize_agent(tools, chat_perplexity_ai, agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION, verbose=True, handle_parsing_errors=True )

Run the agent

chat_answer = agent.run('Qual a cor do cavalo branco de Napoleão?') print('Answer: ', chat_answer)

For more details, refer to the comments and documentation within the pplx.py module itself.

Troubleshooting

For any issues, refer to the logs or debug the pplx.py module directly.