BigHugger
GH Repository · RUC-NLPIR

Search-o1

🔍 Search-o1: Agentic Search-Enhanced Large Reasoning Models [EMNLP 2025]

stars
1,247
30-day movement
-1-0/day
Related entries
60
Connections
1
Pythonreasoningragqwqo1r1livecodegpqapythonamcmathaimo

Search-o1 is a Python toolchain from RUC NLPIR that adds agentic search enhancement to large reasoning models, published at EMNLP 2025. It combines retrieval-augmented generation with reasoning models like QwQ and R1-style models.

You want to pair retrieval with step-by-step reasoning models instead of a plain LLM pipeline.

Use it to

  • Enhance reasoning models with agentic search
  • Run retrieval-augmented generation over reasoning traces
  • Evaluate on math and QA benchmarks like GPQA, AMC, AIMO
  • Test code reasoning with LiveCode tasks

For Researchers and engineers working on RAG with reasoning models

Role
rag
Language
Python
Licence
MIT
Forks
110
Open issues
2
Last push
2025-11-17
topicsragreasoningsearchpythonbenchmarksllm