## Overview

This repository introduces the DeepSeek-R1 series of large language models developed by DeepSeek AI. It includes the first-generation reasoning models, DeepSeek-R1-Zero and DeepSeek-R1, along with several smaller models distilled from DeepSeek-R1. The primary focus of these models is enhanced reasoning capabilities, achieved through innovative reinforcement learning (RL) techniques and knowledge distillation.

The repository provides:

- Model weights and download links (via Hugging Face).
- Detailed performance benchmarks comparing DeepSeek-R1 models against other state-of-the-art models.
- Instructions and recommendations for running the models locally.
- The associated research paper (`DeepSeek_R1.pdf`).
- Licensing information (MIT License).