Skip to content

Latest commit

 

History

History
71 lines (55 loc) · 2.67 KB

README.md

File metadata and controls

71 lines (55 loc) · 2.67 KB

OpenAI(API)-Ollama(Local)-Workflow

This Python project showcases an AI integration using both OpenAI’s API and local Ollama models to perform tasks such as crafting detailed prompts, enhancing outlines, and incorporating specific knowledge. It loops through local models and uses API calls to enhance the structure and content of a message. The script can be used to generate detailed outlines, improve textual data, and generate well-structured responses using several AI models.

Features

  • Local use of Ollama models (qwen2.5:1.5b and llama3.2) for creating prompts and outlines.
  • Integration with OpenAI GPT-4o and GPT-4o-mini for improving outlines and generating responses.
  • History of AI-generated content, including timestamps, in a local file.

Requirements

  • Python 3.7+
  • OpenAI Python SDK
  • Downloaded ollama models on PC
  • LangChain integration with Ollama models

Installation

Download Ollama models if you do not have one

  1. Go and download ollama from here
  2. Browse models of Ollama from here
  3. To install qwen2.5:1.5b and llama3.2 open your command prompt and enter the following command to strat/to download:
ollama run llama3.2
  1. Now, activate a virtual environment (if not already activated):
    • On Windows/Linux: python -m venv env
    • On macOS: source ./env/bin/activate

Note:

  • Ensure you have the necessary permissions to install packages and activate virtual environments.
  • The script will loop through different stages of outline generation, improvement, and response creation, saving all outputs in a text file with timestamps within this folder.

Install Python Dependencies

You can install the required Python packages using pip:

pip install langchain_ollama
pip install openai

Set your OpenAI API key

Replace "YOUR_API_KEY" with your actual OpenAI API key in the code:

client = OpenAI(api_key = "YOUR_API_KEY")

The code will interact with the following models:

  • Qwen 2.5:1.5b (for prompt and improving outline generation)
  • Llama 3.2 (for adding information to the outline)
  • GPT-4o-mini (for improving outlines)
  • GPT-4o (for final response generation)

Saving output:

The script saves the generated response to a file History.txt in the specified folder. You can modify the file path by changing FOLDER_ADDRESS in the script.

with open(file = r"FOLDER_ADDRESS\History.txt", mode = "a") as file:
    present_time = datetime.now()
    file.write(f"Time {present_time}\n")
    file.write(gpt4o_response)

Enjoy changing and playng with the parameters and instructions