PaperQA2

PaperQA2 is a Python research assistant for researchers and developers who need answers grounded in scientific literature. It searches a local document collection and connects its answers to evidence through in-text citations.

• Questions and comparisons: Ask questions about papers, summarize findings, or look for contradictions across documents.
• Iterative research: A language agent can refine its searches and answers, gathering additional evidence as it works through a question.
• Evidence selection: The tool ranks relevant passages and summarizes them in the context of your question before composing an answer. It also uses document metadata to inform retrieval.
• Document search: Search a local collection of PDFs and text files with full-text search, or draw on Microsoft Office documents and source code.
• Paper metadata: Automatic lookups retrieve citation and journal quality information from external providers, including Semantic Scholar and Crossref.

The package runs locally and is open source under Apache 2.0. It provides a command-line interface and a Python library. LiteLLM support allows a choice of model providers, including locally hosted models, and Sentence Transformers can supply local embeddings.

The default OpenAI models and embeddings use cloud APIs and require an API key. Local model support doesn't make the external paper metadata services local; those lookups need an internet connection.