scripts

Research Collection Scripts

Automated tools for collecting and organizing RL + LLM research papers.

Scripts

arxiv_paper_collector.py

Automatically fetches recent papers from arXiv on:

• Reinforcement Learning for Code Generation
• RLHF and LLM Alignment
• Program Synthesis with RL
• LLM Reasoning with RL
• AlphaCode and related work

Usage:

cd scripts
python3 arxiv_paper_collector.py

What it does:

1. Searches arXiv with targeted queries
2. Filters papers from 2022-2025
3. Removes duplicates
4. Organizes by topic
5. Generates markdown files with paper details
6. Saves JSON database for programmatic access

Output:

• ../Modern-RL-Research/*/PAPERS.md - Organized paper lists
• papers_database.json - Complete database

Customization: Edit the SEARCH_QUERIES dictionary in the script to add/modify search terms.

Rate Limiting: The script respects arXiv's rate limits (3 seconds between requests).

Requirements

No external dependencies! Uses only Python standard library:

• urllib - For API requests
• xml.etree.ElementTree - For XML parsing
• json - For data export

Python 3.6+ required.

Tips

Updating the collection:

# Run monthly to keep papers up-to-date
python3 arxiv_paper_collector.py

Filtering results: Adjust start_year and max_results parameters in the script.

Custom searches: Add your own queries to SEARCH_QUERIES dictionary.

JSON output: Use papers_database.json for custom analysis:

import json
with open('papers_database.json') as f:
    papers = json.load(f)

Future Enhancements

Potential additions:

• Paper citation counts via Semantic Scholar API
• Automatic GitHub repo detection
• PDF download and text extraction
• Duplicate detection across versions (v1, v2, etc.)
• Email notifications for new papers
• Integration with paper management tools (Zotero, etc.)

---

Part of the Study-Reinforcement-Learning repository