petergriffinjin/search-r1

★ 4,692PythonAudience · researcherComplexity · 5/5Setup · hard

Mindmap

mindmap
  root((search-r1))
    What it does
      Trains models to search
      Reinforcement learning
      Tool-calling AI
    Retrieval Backends
      Sparse keyword search
      Dense vector search
      Online search engines
    Supported Base Models
      Llama variants
      Qwen variants
    Use Cases
      Search-augmented reasoning
      QA with live retrieval
      Custom retrieval research

mindmap root((search-r1)) What it does Trains models to search Reinforcement learning Tool-calling AI Retrieval Backends Sparse keyword search Dense vector search Online search engines Supported Base Models Llama variants Qwen variants Use Cases Search-augmented reasoning QA with live retrieval Custom retrieval research

Click or tap to explore — scroll the page freely

Things people build with this

USE CASE 1

Train a language model to automatically issue search queries during reasoning and incorporate results before answering.

USE CASE 2

Run inference on a trained search-augmented model and watch it query a local Wikipedia index in real time.

USE CASE 3

Plug in a custom retrieval system or dataset to experiment with different search backends during training.

USE CASE 4

Compare search-augmented reasoning against baseline models using the datasets and training scripts provided.

Tech stack

PythonPyTorchCUDALlamaQwen

Getting it running

Difficulty · hard Time to first run · 1day+

Requires GPU hardware, large model weight downloads, a local vector index, and familiarity with Python ML environments and training workflows.

In plain English

Search-R1 is a research framework for training AI language models to search the web (or a local document database) as part of their reasoning process. The core idea is that a language model should not just answer questions from what it already knows. Instead, it should learn when to pause, issue a search query, read the results, and then continue reasoning with that new information, all in one continuous process. Search-R1 provides the training infrastructure to teach a model this behavior using reinforcement learning. Reinforcement learning here means the model is rewarded when it produces correct answers and penalized when it does not, without being given explicit step-by-step instructions on how to reason. Over many training rounds, the model figures out on its own when and how to call the search engine effectively. The framework extends the open-source work behind DeepSeek-R1 and is described by its authors as an open alternative to OpenAI's DeepResearch product. The framework supports several popular language models as starting points, including Llama and Qwen variants, and works with multiple types of search backends: a local sparse retriever (keyword matching), a local dense retriever (meaning-based similarity search with a vector index), or online search engines. Researchers can also plug in their own datasets and their own retrieval systems. The quick-start workflow shown in the README involves downloading a Wikipedia index, preparing a question-answering dataset, launching a local retrieval server, and then running a training script. After training, you can run inference to ask the trained model questions and watch it search and reason in real time. The project is primarily aimed at AI researchers who want to experiment with tool-calling and search-augmented reasoning in language models. It requires GPU hardware and familiarity with setting up Python environments, model weights, and vector indexes. Two research papers are linked in the README with detailed results comparing the trained models against baselines.

Copy-paste prompts

Prompt 1

Using Search-R1, walk me through the full setup: downloading the Wikipedia index, preparing a QA dataset, starting the retrieval server, and running the training script on a Llama model.

Prompt 2

Show me how to run inference with a trained Search-R1 model and have it search a local document index before producing an answer.

Prompt 3

How do I plug my own retrieval system into Search-R1 instead of the default sparse keyword retriever?

Prompt 4

Set up the Search-R1 dense vector retrieval server locally and test it with a sample question-answering query.

Prompt 5

Explain the reinforcement learning reward signal in Search-R1 and how the model learns when to call the search engine without being told explicitly.

Open on GitHub → Explain another repo

← petergriffinjin on gitmyhub — every repo by this author, as a profile.

Verify against the repo before relying on details.