Agents & Tool Use · 2025
Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning
Trains LLMs by RL to interleave step-by-step reasoning with live search-engine calls using retrieved-token masking, giving a training mechanism for agentic retrieval instead of prompted RAG.
Editorial record
Plain-language summary
Trains LLMs by RL to interleave step-by-step reasoning with live search-engine calls using retrieved-token masking, giving a training mechanism for agentic retrieval instead of prompted RAG.
Knowledge graph
Relationships
Descendants
Conceptual ancestorEvidence: Strongly supported
ReAct: Synergizing Reasoning and Acting in LMs
Builds on the agents lineage in the archive
freshness sweep 2026
Source record
Provenance
- Record ID
- P-601
- Record created
- 2026-07-13
- Last reviewed
- 2026-07-14
- Record version
- 2
- https://arxiv.org/abs/2503.09516
- arXiv:2503.09516
Citation caveat: Citation metadata is approximate and marked unverified in the source dataset.