<!DOCTYPE html>
<html>
<head>
<meta charset="utf-8">
<meta name="description" content="A collection of research papers from DeepSeek.">
<meta name="keywords" content="DeepSeek, AI Research, Machine Learning">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>DeepSeek Research Papers</title>
<link href="https://fonts.googleapis.com/css?family=Google+Sans|Noto+Sans|Castoro" rel="stylesheet">
<link rel="stylesheet" href="./static/css/bulma.min.css">
<link rel="stylesheet" href="./static/css/bulma-carousel.min.css">
<link rel="stylesheet" href="./static/css/bulma-slider.min.css">
<link rel="stylesheet" href="./static/css/fontawesome.all.min.css">
<link rel="stylesheet" href="https://cdn.jsdelivr.net/gh/jpswalsh/academicons@1/css/academicons.min.css">
<link rel="stylesheet" href="./static/css/index.css">
<link rel="icon" href="./static/images/favicon.svg">
<script src="https://ajax.googleapis.com/ajax/libs/jquery/3.5.1/jquery.min.js"></script>
<script defer src="./static/js/fontawesome.all.min.js"></script>
<script src="./static/js/bulma-carousel.min.js"></script>
<script src="./static/js/bulma-slider.min.js"></script>
<script src="./static/js/index.js"></script>
</head>
<body>
<section class="hero">
<div class="hero-body">
<div class="container is-max-desktop">
<div class="columns is-centered">
<div class="column has-text-centered">
<h1 class="title is-1 publication-title">DeepSeek Research Papers</h1>
<div class="is-size-5 publication-authors">
<span class="author-block">DeepSeek Research Team</span>
</div>
</div>
</div>
</div>
</div>
</section>
<section class="hero teaser">
<div class="container is-max-desktop">
<div class="hero-body">
<h2 class="subtitle has-text-centered">
Advancing AI through Open Research and Innovation
</h2>
</div>
</div>
</section>
<section class="section">
<div class="container is-max-desktop">
<div class="columns is-centered">
<div class="column is-full-width">
<h2 class="title is-3 has-text-centered">Publications</h2>
<!-- Paper List Start -->
<div class="content">
<!-- Paper 1 -->
<div class="paper-item box">
<h3 class="title is-4">1. DeepSeek LLM: Scaling Open-Source Language Models with Longtermism</h3>
<p>Scaling open-source language models with a focus on longtermism.</p>
<div class="publication-links">
<span class="link-block">
<a href="https://huggingface.co/papers/2401.02954" target="_blank"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fas fa-file-pdf"></i>
</span>
<span>Paper</span>
</a>
</span>
<span class="is-size-6 has-text-grey">{Jan 6, 2024}</span>
</div>
</div>
<!-- Paper 2 -->
<div class="paper-item box">
<h3 class="title is-4">2. DeepSeekMoE: Towards Ultimate Expert Specialization in Mixture-of-Experts Language Models</h3>
<p>Exploring expert specialization in Mixture-of-Experts language models.</p>
<div class="publication-links">
<span class="link-block">
<a href="https://arxiv.org/abs/2401.06066" target="_blank"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fas fa-file-pdf"></i>
</span>
<span>Paper</span>
</a>
</span>
<span class="is-size-6 has-text-grey">{Jan 11, 2024}</span>
</div>
</div>
<!-- Paper 3 -->
<div class="paper-item box">
<h3 class="title is-4">3. DeepSeek-Coder: When the Large Language Model Meets Programming -- The Rise of Code Intelligence</h3>
<p>Investigating the intersection of large language models and programming.</p>
<div class="publication-links">
<span class="link-block">
<a href="https://arxiv.org/abs/2401.14196" target="_blank"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fas fa-file-pdf"></i>
</span>
<span>Paper</span>
</a>
</span>
<span class="is-size-6 has-text-grey">{Jan 25, 2024}</span>
</div>
</div>
<!-- Paper 4 -->
<div class="paper-item box">
<h3 class="title is-4">4. DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models</h3>
<p>Advancing mathematical reasoning capabilities in open language models.</p>
<div class="publication-links">
<span class="link-block">
<a href="https://arxiv.org/abs/2402.03300" target="_blank"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fas fa-file-pdf"></i>
</span>
<span>Paper</span>
</a>
</span>
<span class="is-size-6 has-text-grey">{Feb 6, 2024}</span>
</div>
</div>
<!-- Paper 5 -->
<div class="paper-item box">
<h3 class="title is-4">5. DeepSeek-VL: Towards Real-World Vision-Language Understanding</h3>
<p>Focusing on real-world vision-language understanding.</p>
<div class="publication-links">
<span class="link-block">
<a href="https://arxiv.org/abs/2403.05525" target="_blank"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fas fa-file-pdf"></i>
</span>
<span>Paper</span>
</a>
</span>
<span class="is-size-6 has-text-grey">{Mar 9, 2024}</span>
</div>
</div>
<!-- Paper 6 -->
<div class="paper-item box">
<h3 class="title is-4">6. DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model</h3>
<p>Developing a strong, economical, and efficient Mixture-of-Experts language model.</p>
<div class="publication-links">
<span class="link-block">
<a href="https://arxiv.org/abs/2405.04434" target="_blank"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fas fa-file-pdf"></i>
</span>
<span>Paper</span>
</a>
</span>
<span class="is-size-6 has-text-grey">{May 7, 2024}</span>
</div>
</div>
<!-- Paper 7 -->
<div class="paper-item box">
<h3 class="title is-4">7. DeepSeek-Prover: Advancing Theorem Proving in LLMs through Large-Scale Synthetic Data</h3>
<p>Using large-scale synthetic data to advance theorem proving in LLMs.</p>
<div class="publication-links">
<span class="link-block">
<a href="https://arxiv.org/abs/2405.14333" target="_blank"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fas fa-file-pdf"></i>
</span>
<span>Paper</span>
</a>
</span>
<span class="is-size-6 has-text-grey">{May 23, 2024}</span>
</div>
</div>
<!-- Paper 8 -->
<div class="paper-item box">
<h3 class="title is-4">8. DeepSeek-Coder-V2: Breaking the Barrier of Closed-Source Models in Code Intelligence</h3>
<p>Aiming to surpass closed-source models in code intelligence.</p>
<div class="publication-links">
<span class="link-block">
<a href="https://arxiv.org/abs/2406.11931" target="_blank"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fas fa-file-pdf"></i>
</span>
<span>Paper</span>
</a>
</span>
<span class="is-size-6 has-text-grey">{Jun 17, 2024}</span>
</div>
</div>
<!-- Paper 9 -->
<div class="paper-item box">
<h3 class="title is-4">9. Let the Expert Stick to His Last: Expert-Specialized Fine-Tuning for Sparse Architectural Large Language Models</h3>
<p>Fine-tuning sparse architectural large language models with expert specialization.</p>
<div class="publication-links">
<span class="link-block">
<a href="https://arxiv.org/abs/2407.01906" target="_blank"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fas fa-file-pdf"></i>
</span>
<span>Paper</span>
</a>
</span>
<span class="is-size-6 has-text-grey">{Jul 2, 2024}</span>
</div>
</div>
<!-- Paper 10 -->
<div class="paper-item box">
<h3 class="title is-4">10. DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree Search</h3>
<p>Utilizing proof assistant feedback for reinforcement learning and Monte-Carlo Tree Search.</p>
<div class="publication-links">
<span class="link-block">
<a href="https://arxiv.org/abs/2408.08152" target="_blank"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fas fa-file-pdf"></i>
</span>
<span>Paper</span>
</a>
</span>
<span class="is-size-6 has-text-grey">{Aug 15, 2024}</span>
</div>
</div>
<!-- Paper 11 -->
<div class="paper-item box">
<h3 class="title is-4">11. Janus: Decoupling Visual Encoding for Unified Multimodal Understanding and Generation</h3>
<p>Decoupling visual encoding for unified multimodal understanding and generation.</p>
<div class="publication-links">
<span class="link-block">
<a href="https://arxiv.org/abs/2410.13848" target="_blank"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fas fa-file-pdf"></i>
</span>
<span>Paper</span>
</a>
</span>
<span class="is-size-6 has-text-grey">{Oct 17, 2024}</span>
</div>
</div>
<!-- Paper 12 -->
<div class="paper-item box">
<h3 class="title is-4">12. JanusFlow: Harmonizing Autoregression and Rectified Flow for Unified Multimodal Understanding and Generation</h3>
<p>Harmonizing autoregression and rectified flow for unified multimodal understanding and generation.</p>
<div class="publication-links">
<span class="link-block">
<a href="https://arxiv.org/abs/2411.07975" target="_blank"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fas fa-file-pdf"></i>
</span>
<span>Paper</span>
</a>
</span>
<span class="is-size-6 has-text-grey">{Nov 12, 2024}</span>
</div>
</div>
<!-- Paper 13 -->
<div class="paper-item box">
<h3 class="title is-4">13. DeepSeek-VL2: Mixture-of-Experts Vision-Language Models for Advanced Multimodal Understanding</h3>
<p>Mixture-of-Experts Vision-Language Models for advanced multimodal understanding.</p>
<div class="publication-links">
<span class="link-block">
<a href="https://arxiv.org/abs/2412.10302" target="_blank"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fas fa-file-pdf"></i>
</span>
<span>Paper</span>
</a>
</span>
<span class="is-size-6 has-text-grey">{Dec 13, 2024}</span>
</div>
</div>
<!-- Paper 14 -->
<div class="paper-item box">
<h3 class="title is-4">14. DeepSeek-V3 Technical Report</h3>
<p>Technical report for DeepSeek-V3.</p>
<div class="publication-links">
<span class="link-block">
<a href="https://arxiv.org/abs/2412.19437" target="_blank"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fas fa-file-pdf"></i>
</span>
<span>Paper</span>
</a>
</span>
<span class="is-size-6 has-text-grey">{Dec 27, 2024}</span>
</div>
</div>
<!-- Paper 15 -->
<div class="paper-item box">
<h3 class="title is-4">15. DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning</h3>
<p>Incentivizing Reasoning Capability in LLMs via Reinforcement Learning.</p>
<div class="publication-links">
<span class="link-block">
<a href="https://arxiv.org/abs/2501.12948" target="_blank"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fas fa-file-pdf"></i>
</span>
<span>Paper</span>
</a>
</span>
<span class="is-size-6 has-text-grey">{Jan 27, 2025}</span>
</div>
</div>
<!-- Paper 16 -->
<div class="paper-item box">
<h3 class="title is-4">16. Janus-Pro: Unified Multimodal Understanding and Generation with Data and Model Scaling</h3>
<p>Unified Multimodal Understanding and Generation with Data and Model Scaling.</p>
<div class="publication-links">
<span class="link-block">
<a href="https://arxiv.org/abs/2501.17811" target="_blank"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fas fa-file-pdf"></i>
</span>
<span>Paper</span>
</a>
</span>
<span class="is-size-6 has-text-grey">{Jan 31, 2025}</span>
</div>
</div>
<!-- Paper 17 -->
<div class="paper-item box">
<h3 class="title is-4">17. Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention</h3>
<p>Hardware-Aligned and Natively Trainable Sparse Attention.</p>
<div class="publication-links">
<span class="link-block">
<a href="https://arxiv.org/abs/2502.11089" target="_blank"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fas fa-file-pdf"></i>
</span>
<span>Paper</span>
</a>
</span>
<span class="is-size-6 has-text-grey">{Feb 16, 2025}</span>
</div>
</div>
</div>
<!-- Paper List End -->
</div>
</div>
</div>
</section>
<footer class="footer">
<div class="container">
<div class="content has-text-centered">
<p>
© 2024 DeepSeek. All rights reserved.
</p>
<p>
This website is built using the Bulma CSS framework.
</p>
</div>
</div>
</footer>
</body>
</html>