Improved prompting to extract questions and answers

Author: HarrySu123

conversion2025/README.md

@@ -1,32 +1,22 @@
-This is a folder where we'll generate a 'wizard' to automatically process input documents ready for in2lambda to reacte a JSON (or to directly create a JSON)
-
-We'll work on this branch 'hackathon'
-
-
 # README
 
 ## Overview
-This Jupyter Notebook (`sandbox.ipynb`) is designed for processing scientific documents, extracting mathematical expressions, and formatting them in Markdown. It leverages Azure OpenAI's LLM capabilities for text transformation.
-
-## Features
-- Loads PDFs and extracts text using `UnstructuredPDFLoader` and `PyMuPDF`.
-- Converts mathematical expressions into properly formatted Markdown.
-- Uses `langchain` and `AzureChatOpenAI` for text processing.
-- Supports structured output parsing using `pydantic`.
+This Jupyter Notebook (`file_name.ipynb`) is designed for processing scientific documents, extracting mathematical expressions, and formatting them in Markdown. It leverages Mathpix and OpenAI's LLM capabilities for text transformation.
 
 ## Requirements
 Ensure you have the following installed:
 - Python 3.8+
 - `pip install -r requirements.txt`
-- `langchain`, `langchain_openai`, `pydantic`, `dotenv`, `PyMuPDF`, `PIL`
 
 ## Setup
-1. Create a `.env` file in the root directory and add your Azure OpenAI API keys:
+1. Create a `.env` file in the root directory and add your OpenAI API keys:
    ```env
-   AZURE_OPENAI_API_KEY=<your-api-key>
-   AZURE_OPENAI_ENDPOINT=<your-endpoint>
+   OPENAI_API_KEY=<your-openai-api-key>
+   OPENAI_MODEL=<your-openai-model>
+   MATHPIX_API_KEY=<your-mathpix-key>
+   MATHPIX_APP_ID=<your-mathpix-id>
    ```
-4. Open `sandbox.ipynb` and execute the cells to process your documents.
+4. Open `file_name.ipynb` and execute the cells to process your documents.
 
 ## Notes
 - Ensure your API key and endpoint are correct, as they are required for LLM functionality.