Open-R1

Visit Tool

Open-r1 is an open-source coding & development tool that provides a full reproduction of DeepSeek-R1. It offers scripts for training, evaluating, and generating synthetic data for large language models.

Claim this tool

No Views Yet

At a glance

Pricing

Open Source

Free tier

Yes

API

Skill level

Technical

About

What is open-r1?

Open-r1 is a fully open-source project dedicated to reproducing the DeepSeek-R1 model, providing a comprehensive framework for researchers and developers. The repository includes essential scripts for training models using techniques like Supervised Fine-Tuning (SFT) and Group Relative Policy Optimization (GRPO), as well as tools for evaluating model performance and generating synthetic data. It supports various hardware configurations and integrates with platforms like Hugging Face Hub and Weights and Biases. The project emphasizes community contribution and aims to build the missing pieces of the R1 pipeline, making advanced LLM development accessible and reproducible. It also features specialized functionalities like a code interpreter reward function for competitive programming tasks, supporting sandboxes like E2B and Morph.

Best used for

Ideal for developers and data scientists who need to replicate advanced AI models, train custom language models for specific tasks, and evaluate their performance. Especially valuable for those working on reasoning, math, and coding challenges, or contributing to open-source AI research.

Common actions

reproduce AI models

train language models

evaluate AI performance

generate synthetic data

develop AI for coding

github copilot"AI Agents"face swappingautomated workflowopen-sourcecollaborationlow-code/no-codeworkflowsdeepfake

Capabilities

Key features

DeepSeek-R1 reproduction
SFT training scripts
GRPO training scripts
Synthetic data generation
Model evaluation tools
Code interpreter reward
Hugging Face Hub integration

Target Audience

developerdata scientist

Integrations

hugging-faceweights-and-biases

Pricing & Plans

Open Source

Free

FAQs

What are the hardware requirements for training models with Open-r1?

The training commands provided are configured for a node with 8 x H100s (80GB). For different hardware or topologies, you may need to adjust the batch size and number of gradient accumulation steps to optimize performance.

Does Open-r1 support multi-node training?

Yes, Open-r1 supports multi-node training, particularly for GRPO. An example Slurm script is provided for running on N+1 nodes, with one node acting as the vLLM server and N nodes for training.

How does the code interpreter reward function work?

The code interpreter reward function evaluates generated code against test cases, returning a success rate as the reward. It supports sandboxes like E2B and Morph, requiring API keys for their respective services to ensure safe execution.

Trending

Subcategories trending in Coding & Development

Code Assistants DevOps & Infrastructure No-Code / Low-Code Testing & QA Backend & APIs Prompt Engineering

Trending

Explore

Browse AI tools by category

Content & Design Productivity & Business Coding & Development AI Agents & Automation Research & Education Wellness & Lifestyle Career Development Marketing & Growth Data & Analytics Customer Support & CX Finance E-commerce