Search-R1
Search-R1: An Efficient, Scalable RL Training Framework for Reasoning & Search Engine Calling interleaved LLM based on veRL
Stars: 1550
Search-R1 is a tool that trains large language models (LLMs) to reason and call a search engine using reinforcement learning. It is a reproduction of DeepSeek-R1 methods for training reasoning and searching interleaved LLMs, built upon veRL. Through rule-based outcome reward, the base LLM develops reasoning and search engine calling abilities independently. Users can train LLMs on their own datasets and search engines, with preliminary results showing improved performance in search engine calling and reasoning tasks.
README:
Search-R1 is an extension of DeepSeek-R1(-Zero) methods for training reasoning and searching (tool-call) interleaved LLMs. We built upon veRL.
Through RL (rule-based outcome reward), the 3B base LLM (both Qwen2.5-3b-base and Llama3.2-3b-base) develops reasoning and search engine calling abilities all on its own.
Paper: link; Model and data: link; Twitter thread: link; Full experiment log 1: link; Full experiment log 2: link
You can refer to this link for detailed instructions on reproducing the results from the paper.
- Installation
- Quick start
- Preliminary results
- Inference
- Use your own dataset
- Use your own search engine
- Ackowledge
- Citations
conda create -n searchr1 python=3.9
conda activate searchr1
# install torch [or you can skip this step and let vllm to install the correct version for you]
pip install torch==2.4.0 --index-url https://download.pytorch.org/whl/cu121
# install vllm
pip3 install vllm==0.6.3 # or you can install 0.5.4, 0.4.2 and 0.3.1
# verl
pip install -e .
# flash attention 2
pip3 install flash-attn --no-build-isolation
pip install wandbIf you would like to call a local retriever as the search engine, you can install the environment as follows. (We recommend using a seperate environment.)
conda create -n retriever python=3.10
conda activate retriever
# we recommend installing torch with conda for faiss-gpu
conda install pytorch==2.4.0 torchvision==0.19.0 torchaudio==2.4.0 pytorch-cuda=12.1 -c pytorch -c nvidia
pip install transformers datasets
## install the gpu version faiss to guarantee efficient RL rollout
conda install -c pytorch -c nvidia faiss-gpu=1.8.0
## API function
pip install uvicorn fastapiTrain a reasoning + search LLM on NQ dataset with e5 as the retriever and wikipedia as the corpus.
(1) Download the indexing and corpus.
save_path=/the/path/to/save
python scripts/download.py --save_path $save_path
cat $save_path/part_* > $save_path/e5_Flat.index
gzip -d $save_path/wiki-18.jsonl.gz(2) Process the NQ dataset.
python scripts/data_process/nq_search.py(3) Launch a local retrieval server.
conda activate retriever
bash retrieval_launch.sh(4) Run RL training (PPO) with Llama-3.2-3b-base.
conda activate searchr1
bash train_ppo.sh(1) The base model (llama3.2-3b-base) learns to call the search engine and obtain improved performance.
(2) The base model (Qwen2.5-7b-base) can learn to conduct multi-turn search engine calling and reasoning with RL.
(1) Launch a local retrieval server.
conda activate retriever
bash retrieval_launch.sh(2) Run inference.
conda activate searchr1
python infer.pyYou can modify the question on line 7 to something you're interested in.
For each question-answer sample, it should be a dictionary containing the desired content as below:
data = {
"data_source": data_source,
"prompt": [{
"role": "user",
"content": question,
}],
"ability": "fact-reasoning",
"reward_model": {
"style": "rule",
"ground_truth": solution
},
"extra_info": {
'split': split,
'index': idx,
}
}
You can refer to scripts/data_process/nq_search.py for a concrete data processing example.
It is recommended to make your corpus a jsonl file, where each line (a dictionary with "id" key and "contents" key) corresponds to one passage. You can refer to example/corpus.jsonl for an example.
The "id" key corresponds to the passage id, while the "contents" key corresponds to the passage content. For example:
{"id": "0", "contents": "Evan Morris Evan L. Morris (January 26, 1977 \u2013 July 9, 2015) was a lobbyist for Genentech and its parent corporation Roche in Washington."}
...
{"id": "100", "contents": "Three years later, when the United States Exploring Expedition to little-known portions of the globe was organised under Charles Wilkes, Hale was recommended, while yet an undergraduate."}
...
Index your corpora (optional). If you would like to use a local retriever as the search engine, you can index your own corpus by:
bash search_r1/search/build_index.sh
You can change retriever_name and retriever_model to your interested off-the-shelf retriever.
The main philosophy is to launch a local or remote search engine server separately from the main RL training pipeline.
The LLM can call the search engine by calling the search API (e.g., "http://127.0.0.1:8000/retrieve").
You can refer to search_r1/search/retriever_server.py for an example of launching a local retriever server.
- Support google search / bing search / brave search API and others.
- Support LoRA tuning.
- Support supervised finetuning.
- Support off-the-shelf rerankers.
The concept of Search-R1 is inspired by Deepseek-R1 and TinyZero. Its implementation is built upon veRL and RAGEN. We sincerely appreciate the efforts of these teams for their contributions to open-source research and development. We thank Jinsung Yoon and Sercan Arik for insightful discussions.
@article{jin2025search,
title={Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning},
author={Jin, Bowen and Zeng, Hansi and Yue, Zhenrui and Wang, Dong and Zamani, Hamed and Han, Jiawei},
journal={arXiv preprint arXiv:2503.09516},
year={2025}
}For Tasks:
Click tags to check more tools for each tasksFor Jobs:
Alternative AI tools for Search-R1
Similar Open Source Tools
Search-R1
Search-R1 is a tool that trains large language models (LLMs) to reason and call a search engine using reinforcement learning. It is a reproduction of DeepSeek-R1 methods for training reasoning and searching interleaved LLMs, built upon veRL. Through rule-based outcome reward, the base LLM develops reasoning and search engine calling abilities independently. Users can train LLMs on their own datasets and search engines, with preliminary results showing improved performance in search engine calling and reasoning tasks.
oasis
OASIS is a scalable, open-source social media simulator that integrates large language models with rule-based agents to realistically mimic the behavior of up to one million users on platforms like Twitter and Reddit. It facilitates the study of complex social phenomena such as information spread, group polarization, and herd behavior, offering a versatile tool for exploring diverse social dynamics and user interactions in digital environments. With features like scalability, dynamic environments, diverse action spaces, and integrated recommendation systems, OASIS provides a comprehensive platform for simulating social media interactions at a large scale.
VMind
VMind is an open-source solution for intelligent visualization, providing an intelligent chart component based on LLM by VisActor. It allows users to create chart narrative works with natural language interaction, edit charts through dialogue, and export narratives as videos or GIFs. The tool is easy to use, scalable, supports various chart types, and offers one-click export functionality. Users can customize chart styles, specify themes, and aggregate data using LLM models. VMind aims to enhance efficiency in creating data visualization works through dialogue-based editing and natural language interaction.
embodied-agents
Embodied Agents is a toolkit for integrating large multi-modal models into existing robot stacks with just a few lines of code. It provides consistency, reliability, scalability, and is configurable to any observation and action space. The toolkit is designed to reduce complexities involved in setting up inference endpoints, converting between different model formats, and collecting/storing datasets. It aims to facilitate data collection and sharing among roboticists by providing Python-first abstractions that are modular, extensible, and applicable to a wide range of tasks. The toolkit supports asynchronous and remote thread-safe agent execution for maximal responsiveness and scalability, and is compatible with various APIs like HuggingFace Spaces, Datasets, Gymnasium Spaces, Ollama, and OpenAI. It also offers automatic dataset recording and optional uploads to the HuggingFace hub.
raptor
RAPTOR introduces a novel approach to retrieval-augmented language models by constructing a recursive tree structure from documents. This allows for more efficient and context-aware information retrieval across large texts, addressing common limitations in traditional language models. Users can add documents to the tree, answer questions based on indexed documents, save and load the tree, and extend RAPTOR with custom summarization, question-answering, and embedding models. The tool is designed to be flexible and customizable for various NLP tasks.
OSWorld
OSWorld is a benchmarking tool designed to evaluate multimodal agents for open-ended tasks in real computer environments. It provides a platform for running experiments, setting up virtual machines, and interacting with the environment using Python scripts. Users can install the tool on their desktop or server, manage dependencies with Conda, and run benchmark tasks. The tool supports actions like executing commands, checking for specific results, and evaluating agent performance. OSWorld aims to facilitate research in AI by providing a standardized environment for testing and comparing different agent baselines.
single-file-agents
Single File Agents (SFA) is a collection of powerful single-file agents built on top of uv, a modern Python package installer and resolver. These agents aim to perform specific tasks efficiently, demonstrating precise prompt engineering and GenAI patterns. The repository contains agents built across major GenAI providers like Gemini, OpenAI, and Anthropic. Each agent is self-contained, minimal, and built on modern Python for fast and reliable dependency management. Users can run these scripts from their server or directly from a gist. The agents are patternful, emphasizing the importance of setting up effective prompts, tools, and processes for reusability.
marqo
Marqo is more than a vector database, it's an end-to-end vector search engine for both text and images. Vector generation, storage and retrieval are handled out of the box through a single API. No need to bring your own embeddings.
promptwright
Promptwright is a Python library designed for generating large synthetic datasets using a local LLM and various LLM service providers. It offers flexible interfaces for generating prompt-led synthetic datasets. The library supports multiple providers, configurable instructions and prompts, YAML configuration for tasks, command line interface for running tasks, push to Hugging Face Hub for dataset upload, and system message control. Users can define generation tasks using YAML configuration or Python code. Promptwright integrates with LiteLLM to interface with LLM providers and supports automatic dataset upload to Hugging Face Hub.
hydraai
Generate React components on-the-fly at runtime using AI. Register your components, and let Hydra choose when to show them in your App. Hydra development is still early, and patterns for different types of components and apps are still being developed. Join the discord to chat with the developers. Expects to be used in a NextJS project. Components that have function props do not work.
llm-colosseum
llm-colosseum is a tool designed to evaluate Language Model Models (LLMs) in real-time by making them fight each other in Street Fighter III. The tool assesses LLMs based on speed, strategic thinking, adaptability, out-of-the-box thinking, and resilience. It provides a benchmark for LLMs to understand their environment and take context-based actions. Users can analyze the performance of different LLMs through ELO rankings and win rate matrices. The tool allows users to run experiments, test different LLM models, and customize prompts for LLM interactions. It offers installation instructions, test mode options, logging configurations, and the ability to run the tool with local models. Users can also contribute their own LLM models for evaluation and ranking.
raglite
RAGLite is a Python toolkit for Retrieval-Augmented Generation (RAG) with PostgreSQL or SQLite. It offers configurable options for choosing LLM providers, database types, and rerankers. The toolkit is fast and permissive, utilizing lightweight dependencies and hardware acceleration. RAGLite provides features like PDF to Markdown conversion, multi-vector chunk embedding, optimal semantic chunking, hybrid search capabilities, adaptive retrieval, and improved output quality. It is extensible with a built-in Model Context Protocol server, customizable ChatGPT-like frontend, document conversion to Markdown, and evaluation tools. Users can configure RAGLite for various tasks like configuring, inserting documents, running RAG pipelines, computing query adapters, evaluating performance, running MCP servers, and serving frontends.
MemoryLLM
MemoryLLM is a large language model designed for self-updating capabilities. It offers pretrained models with different memory capacities and features, such as chat models. The repository provides training code, evaluation scripts, and datasets for custom experiments. MemoryLLM aims to enhance knowledge retention and performance on various natural language processing tasks.
promptwright
Promptwright is a Python library designed for generating large synthetic datasets using local LLM and various LLM service providers. It offers flexible interfaces for generating prompt-led synthetic datasets. The library supports multiple providers, configurable instructions and prompts, YAML configuration, command line interface, push to Hugging Face Hub, and system message control. Users can define generation tasks using YAML configuration files or programmatically using Python code. Promptwright integrates with LiteLLM for LLM providers and supports automatic dataset upload to Hugging Face Hub. The library is not responsible for the content generated by models and advises users to review the data before using it in production environments.
llama_index
LlamaIndex is a data framework for building LLM applications. It provides tools for ingesting, structuring, and querying data, as well as integrating with LLMs and other tools. LlamaIndex is designed to be easy to use for both beginner and advanced users, and it provides a comprehensive set of features for building LLM applications.
For similar tasks
Search-R1
Search-R1 is a tool that trains large language models (LLMs) to reason and call a search engine using reinforcement learning. It is a reproduction of DeepSeek-R1 methods for training reasoning and searching interleaved LLMs, built upon veRL. Through rule-based outcome reward, the base LLM develops reasoning and search engine calling abilities independently. Users can train LLMs on their own datasets and search engines, with preliminary results showing improved performance in search engine calling and reasoning tasks.
ReasonFlux
ReasonFlux is a revolutionary template-augmented reasoning paradigm that empowers a 32B model to outperform other models in reasoning tasks. The repository provides official resources for the paper 'ReasonFlux: Hierarchical LLM Reasoning via Scaling Thought Templates', including the latest released model ReasonFlux-F1-32B. It includes updates, dataset links, model zoo, getting started guide, training instructions, evaluation details, inference examples, performance comparisons, reasoning examples, preliminary work references, and citation information.
ray
Ray is a unified framework for scaling AI and Python applications. It consists of a core distributed runtime and a set of AI libraries for simplifying ML compute, including Data, Train, Tune, RLlib, and Serve. Ray runs on any machine, cluster, cloud provider, and Kubernetes, and features a growing ecosystem of community integrations. With Ray, you can seamlessly scale the same code from a laptop to a cluster, making it easy to meet the compute-intensive demands of modern ML workloads.
aiscript
AiScript is a lightweight scripting language that runs on JavaScript. It supports arrays, objects, and functions as first-class citizens, and is easy to write without the need for semicolons or commas. AiScript runs in a secure sandbox environment, preventing infinite loops from freezing the host. It also allows for easy provision of variables and functions from the host.
dstack
Dstack is an open-source orchestration engine for running AI workloads in any cloud. It supports a wide range of cloud providers (such as AWS, GCP, Azure, Lambda, TensorDock, Vast.ai, CUDO, RunPod, etc.) as well as on-premises infrastructure. With Dstack, you can easily set up and manage dev environments, tasks, services, and pools for your AI workloads.
mobius
Mobius is an AI infra platform including realtime computing and training. It is built on Ray, a distributed computing framework, and provides a number of features that make it well-suited for online machine learning tasks. These features include: * **Cross Language**: Mobius can run in multiple languages (only Python and Java are supported currently) with high efficiency. You can implement your operator in different languages and run them in one job. * **Single Node Failover**: Mobius has a special failover mechanism that only needs to rollback the failed node itself, in most cases, to recover the job. This is a huge benefit if your job is sensitive about failure recovery time. * **AutoScaling**: Mobius can generate a new graph with different configurations in runtime without stopping the job. * **Fusion Training**: Mobius can combine TensorFlow/Pytorch and streaming, then building an e2e online machine learning pipeline. Mobius is still under development, but it has already been used to power a number of real-world applications, including: * A real-time recommendation system for a major e-commerce company * A fraud detection system for a large financial institution * A personalized news feed for a major news organization If you are interested in using Mobius for your own online machine learning projects, you can find more information in the documentation.
vasttools
This repository contains a collection of tools that can be used with vastai. The tools are free to use, modify and distribute. If you find this useful and wish to donate your welcome to send your donations to the following wallets. BTC 15qkQSYXP2BvpqJkbj2qsNFb6nd7FyVcou XMR 897VkA8sG6gh7yvrKrtvWningikPteojfSgGff3JAUs3cu7jxPDjhiAZRdcQSYPE2VGFVHAdirHqRZEpZsWyPiNK6XPQKAg RVN RSgWs9Co8nQeyPqQAAqHkHhc5ykXyoMDUp USDT(ETH ERC20) 0xa5955cf9fe7af53bcaa1d2404e2b17a1f28aac4f Paypal PayPal.Me/cryptolabsZA
co-llm
Co-LLM (Collaborative Language Models) is a tool for learning to decode collaboratively with multiple language models. It provides a method for data processing, training, and inference using a collaborative approach. The tool involves steps such as formatting/tokenization, scoring logits, initializing Z vector, deferral training, and generating results using multiple models. Co-LLM supports training with different collaboration pairs and provides baseline training scripts for various models. In inference, it uses 'vllm' services to orchestrate models and generate results through API-like services. The tool is inspired by allenai/open-instruct and aims to improve decoding performance through collaborative learning.
For similar jobs
sweep
Sweep is an AI junior developer that turns bugs and feature requests into code changes. It automatically handles developer experience improvements like adding type hints and improving test coverage.
teams-ai
The Teams AI Library is a software development kit (SDK) that helps developers create bots that can interact with Teams and Microsoft 365 applications. It is built on top of the Bot Framework SDK and simplifies the process of developing bots that interact with Teams' artificial intelligence capabilities. The SDK is available for JavaScript/TypeScript, .NET, and Python.
ai-guide
This guide is dedicated to Large Language Models (LLMs) that you can run on your home computer. It assumes your PC is a lower-end, non-gaming setup.
classifai
Supercharge WordPress Content Workflows and Engagement with Artificial Intelligence. Tap into leading cloud-based services like OpenAI, Microsoft Azure AI, Google Gemini and IBM Watson to augment your WordPress-powered websites. Publish content faster while improving SEO performance and increasing audience engagement. ClassifAI integrates Artificial Intelligence and Machine Learning technologies to lighten your workload and eliminate tedious tasks, giving you more time to create original content that matters.
chatbot-ui
Chatbot UI is an open-source AI chat app that allows users to create and deploy their own AI chatbots. It is easy to use and can be customized to fit any need. Chatbot UI is perfect for businesses, developers, and anyone who wants to create a chatbot.
BricksLLM
BricksLLM is a cloud native AI gateway written in Go. Currently, it provides native support for OpenAI, Anthropic, Azure OpenAI and vLLM. BricksLLM aims to provide enterprise level infrastructure that can power any LLM production use cases. Here are some use cases for BricksLLM: * Set LLM usage limits for users on different pricing tiers * Track LLM usage on a per user and per organization basis * Block or redact requests containing PIIs * Improve LLM reliability with failovers, retries and caching * Distribute API keys with rate limits and cost limits for internal development/production use cases * Distribute API keys with rate limits and cost limits for students
uAgents
uAgents is a Python library developed by Fetch.ai that allows for the creation of autonomous AI agents. These agents can perform various tasks on a schedule or take action on various events. uAgents are easy to create and manage, and they are connected to a fast-growing network of other uAgents. They are also secure, with cryptographically secured messages and wallets.
griptape
Griptape is a modular Python framework for building AI-powered applications that securely connect to your enterprise data and APIs. It offers developers the ability to maintain control and flexibility at every step. Griptape's core components include Structures (Agents, Pipelines, and Workflows), Tasks, Tools, Memory (Conversation Memory, Task Memory, and Meta Memory), Drivers (Prompt and Embedding Drivers, Vector Store Drivers, Image Generation Drivers, Image Query Drivers, SQL Drivers, Web Scraper Drivers, and Conversation Memory Drivers), Engines (Query Engines, Extraction Engines, Summary Engines, Image Generation Engines, and Image Query Engines), and additional components (Rulesets, Loaders, Artifacts, Chunkers, and Tokenizers). Griptape enables developers to create AI-powered applications with ease and efficiency.