Skip to main content

A package to download arXiv papers and interact with PDFs using Ollama LLM

Project description

Academicagent

Academicagent is a Python package that integrates downloading papers from arXiv and evaluating them using a local large model (Ollama).

English | 简体中文


Features

  • Download arXiv Papers
    Search for and download a specified number of PDF papers from arXiv based on the keywords provided by the user.

  • Large Model Q&A
    For each downloaded paper, extract the first page from the PDF and use the local large model to generate a Chinese summary along with an evaluation score for the paper.


Installation

  1. Download Ollama and Start the Local Large Model Service

    Download Ollama from: https://ollama.com/

    For example, to download the deepseek-r1:1.5b model, run:

    ollama pull deepseek-r1:1.5b
    
  2. Install paperagent

Install using pip:

pip install paperagent

Usage Example

from academicagent.agent import run_agent

run_agent(paper_keyword="object detection", total_count=1, save_path="papers", model_name="deepseek-r1:1.5b")

Input Parameters

  • paper_keyword (string): The keyword used to search for papers on arXiv.
    Example: "object detection"

  • total_count (integer): The total number of papers to download.
    Example: 5

  • save_path (string, default "papers"): The folder path where the downloaded PDFs will be saved. If not provided, it defaults to "papers".
    Example: "papers"

  • question (string, optional): The question to provide to the large model. If not specified, the default question is:
    "Please generate a Chinese summary of this paper and evaluate its value based on originality, effectiveness, and scope, on a scale of 0 to 10, then provide your score after the summary is generated."

  • model_name (string, default "deepseek-r1:1.5b"):
    The name of the local large model to be used for invoking Ollama.

Output

  • PDF Download: The PDFs of the papers are downloaded from arXiv into the specified save_path folder based on the provided keyword and count.

  • Evaluation File: The title of each paper and the large model's response are written into a Markdown file.


Version History

  • v0.1.0
    Initial release, implementing arXiv paper downloading and large model Q&A functionality.

Contact

If you have any questions or suggestions, please feel free to submit an issue or pull request.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page