Paper Summarizer is a Python library for extracting and summarizing content from research papers. It utilizes OpenAI's language models to generate comprehensive summaries and key points from PDF documents.
pip install https://github.com/organisciak/paper-summarizerThis library uses the OpenAI API. You need to specify your API key. Here are two ways:
import openai
# if key in file
openai.api_key_path = 'path/to/key'
# if key in environmental variable
openai.api_key = os.getenv("OPENAI_API_KEY")summarize() generates a long summary, which is used for subsequent shorter summaries and questions asking.
from paper_summarizer import PaperSummarizer
# Initialize the summarizer with a PDF path
summarizer = PDFPaperSummarizer("path/to/your/document.pdf", model='gpt-3.5-turbo')
# Generate a summary
summary = summarizer.summarize()Generating a summary the first time may take a few minutes. First, it chunks the full document into chunks that are up to two-thirds of the max input size of the model (for ChatGPT, or gpt-3.5-turbo, that's 4,097, for gpt-4 that's 8,192). Each chunk is outlined individually, then they're combined into a single paper summary and outline of key points.
question() can pose ad-hoc questions and requests. The md argument returns a Ipython Markdown object which renders nicely in notebooks.
summarizer.question("Rewrite this paper as a series of haikus.", md=True)- Extract text from PDF documents
- Split large documents into manageable chunks for summarization
- Cache summaries as yaml to speed up repeated queries while offering a human-readable view of summaries and outputs.
Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change. This started as quick proof-of-concept, so there's a lot to do - at the same time, I hope not to over-engineer this thing.
Peter Organisciak https://porg.dev