mangopy/SearchLM

Official code for NeurIPS2025 "Iterative Self-Incentivization Empowers Large Language Models as Agentic Searchers"

42
/ 100
Emerging

Implements an expectation-maximization framework where LLMs iteratively refine search strategies through self-incentivization—using entropy-based trajectory weighting to identify high-value reasoning paths without external rewards. Integrates ColBERT for dense retrieval over Wikipedia corpora and supports distributed training via DeepSpeed, enabling end-to-end optimization of both search decisions and answer generation across multi-turn agentic workflows.

225 stars.

No Package No Dependents
Maintenance 10 / 25
Adoption 10 / 25
Maturity 16 / 25
Community 6 / 25

How are scores calculated?

Stars

225

Forks

4

Language

Python

License

Apache-2.0

Last pushed

Jan 14, 2026

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/rag/mangopy/SearchLM"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.