| # AOAI-Prompt-Optimization-Tool |
|
|
| The Azure OpenAI Prompt Optimization Tool is an interactive Python application powered by OpenAI's GPT-3 technology and tkinter GUI framework. This tool evaluates user-provided prompts, optimizes them, and provides enhanced responses. It features dynamic input, response presentation, and user-friendly functionalities, all while leveraging the power of AI-driven language optimization. |
|
|
| ## Prerequisites |
| An existing Azure OpenAI resource and model deployment of a chat model (e.g. gpt-35-turbo, gpt-4) |
|
|
| ## Deploy the tool |
|
|
| ### Deploy from your local machine |
|
|
| 1. Create an environment variables file `.env` with your API Key, Endpoint, and Deployed Model name. |
| - `AZURE_OPENAI_API_KEY`: your Azure OpenAI resource API key. |
| - `AZURE_OPENAI_API_ENDPOINT`: your Azure OpenAI resource endpoint. |
| - `AZURE_OPENAI_DEPLOYED_MODEL`: your Azure OpenAI deployment name. |
|
|
| 2. Run the following command to install all dependencies. |
| ``` |
| pip install -r requirements.txt |
| ``` |
|
|
| 3. Run `app.py` file. |
|
|
| ## How it works |
|
|
| The interface consists of four main elements: |
| - The prompt input text field to write your prompt. |
| - The response output window to see the omptimization recommendation. |
| - The Submit button to send the prompt for AI analysis. |
| - The Clear button to erase the prompt and start over. |
|
|
|  |
|
|
| Once you enter and submit a prompt, the following response will be generated: |
| - Grade: a grade describing how good the prompt is on a scale from 1 to 10. |
| - Reason for Grade: the analysis of the prompts highlighting missing key elements and how to improve it. |
| - Optimized Prompt: an example of an optimized prompt that will achieve the same goal in a more efficient way. |
|
|
|  |
|
|
| ## Optimization criteria |
| The tool analyzes the prompt to determine its optimization level based on several criteria that aim to make the prompt clear, effective, and well-structured for the intended purpose. Below are the criteria used to evaluate whether a prompt is optimized or not: |
| - Clarity: The prompt should be easy to understand without ambiguity. It should convey the desired task or question clearly, so that both human readers and AI models can comprehend it accurately. |
| - Conciseness: An optimized prompt is concise and to the point. Unnecessary or redundant words should be eliminated to ensure efficient communication of the task to the AI model. |
| - Relevance: The prompt should directly relate to the desired output. Irrelevant information can confuse the AI model and result in inaccurate or off-topic responses. |
| - Contextual Information: Including sufficient context in the prompt can help guide the AI model towards the desired response. Relevant details or examples can improve the quality of generated content. |
| - Specificity: A well-optimized prompt is specific about the expected format, details, or structure of the response. Clear instructions can guide the AI model in producing accurate and relevant output. |
| - Language Quality: The language used in the prompt should be grammatically correct, coherent, and free of jargon that might confuse the AI model. |
|
|
| Overall, prompt optimization involves balancing these criteria to create a prompt that maximizes the AI model's ability to produce accurate, relevant, and contextually appropriate responses while minimizing the potential for misunderstandings or errors. |
|
|
