Ask your Docs

This Django model enhances your document interaction by integrating OpenAI's text embedding and GPT-3.5-turbo chat completion models. It simplifies querying a collection of precomputed text embeddings and generating responses using GPT-3.5-turbo.

Usage

1. Initialization: Prior to using ChatModel, set your OpenAI API key in the environment variable OPENAI_API_KEY_H. The initialize method configures the necessary data structures.

   ChatModel.initialize()

2. Querying Related Strings: Discover related strings to a query and their relatedness scores with the strings_ranked_by_relatedness method.

   strings, relatednesses = ChatModel.strings_ranked_by_relatedness(query, top_n=100)

3. Generating Messages for GPT-3.5-turbo: Create a message for GPT-3.5-turbo based on related strings and a specified token budget using the query_message method.

   message = ChatModel.query_message(query, token_budget)

4. Asking Questions: Utilize the ask method to input a query and generate a response using GPT-3.5-turbo.

   response = ChatModel.ask(query, token_budget=4096 - 500)

Dependencies

• Django (for Django model usage)
• OpenAI Python library
• pandas
• tiktoken
• scipy
• scikit-learn

Note

Ensure that the required Python packages (Django, openai, pandas, tiktoken, scipy, scikit-learn) are installed before deploying ChatModel.

License

This code is distributed under the MIT License.

Process Docs View

The ProcessFilesView in this Django application streamlines document management. It facilitates file uploads, automatic processing of unprocessed files, and displays lists of uploaded and processed files. The interface ensures a user-friendly experience.

Usage

1. Upload Files:

   • Select the "Upload File" button to upload documents from your device. The list of uploaded files updates automatically.

2. Process Files:

   • Click the "Process" button to automatically process the uploaded files. Processed files are displayed in the corresponding list.

3. File Visualization:

   • Clear presentation of both uploaded and processed files for easy identification.

Considerations

• The view utilizes the openai library for text embeddings, involving calls to the OpenAI API.
• Configure your OpenAI API key in the environment before using the view.