# DeepSeek-R1 | DeepSeek | Knowledge Base | Instagit GitHub Stars: 91.9k Repository: https://github.com/deepseek-ai/DeepSeek-R1 --- ## Articles ### [")[1].strip()](/deepseek-ai/DeepSeek-R1/deepseek-r1-licensing-and-usage-terms) Understand DeepSeek-R1 licensing and usage terms for the DeepSeek-R1 repository. Learn how to effectively use and deploy this powerful AI model. - Tags: getting-started - Published: 2026-02-27 ### [Can DeepSeek-R1 Models Be Further Trained or Adapted? A Complete Guide to Fine-Tuning and Customization](/deepseek-ai/DeepSeek-R1/can-deepseek-r1-models-be-further-trained-or-adapted) Learn how to fine-tune and adapt DeepSeek-R1 models. This guide covers reinforcement learning, supervised fine-tuning, LoRA, and distillation for customized AI. - Tags: how-to-guide - Published: 2026-02-27 ### [Limitations of DeepSeek-R1 Models: Critical Constraints for Production Deployment](/deepseek-ai/DeepSeek-R1/what-are-the-limitations-of-deepseek-r1-models) Explore DeepSeek-R1 model limitations: strict temperature needs, system prompt issues & formatting challenges. Understand production deployment constraints. - Tags: deep-dive - Published: 2026-02-27 ### [Examples of Using DeepSeek-R1 for Complex Reasoning Tasks: Implementation Guide](/deepseek-ai/DeepSeek-R1/examples-of-using-deepseek-r1-for-complex-reasoning-tasks) Explore DeepSeek-R1 examples for complex reasoning tasks. Learn implementation using its 671B MoE architecture with reinforcement learning for advanced problem solving. Optimize your configurations. - Tags: how-to-guide - Published: 2026-02-27 ### [`# How to Optimize Inference Speed for DeepSeek-R1: A Complete Guide to High-Performance Deployment](/deepseek-ai/DeepSeek-R1/how-to-optimize-inference-speed-for-deepseek-r1) Boost DeepSeek-R1 inference speed with our complete guide. Learn practical techniques for high-performance deployment and unlock the full potential of this powerful AI model. - Tags: performance - Published: 2026-02-27 ### [DeepSeek-R1 vs Other Reasoning Models: Performance Comparison and Benchmark Analysis](/deepseek-ai/DeepSeek-R1/deepseek-r1-vs-other-reasoning-models-performance-comparison) Compare DeepSeekR1 to other reasoning models. Discover 97.3% MATH-500 accuracy, exceeding OpenAI's models—fully open-source for your projects. Analyze performance now. - Tags: performance - Published: 2026-02-27 ### [How to Adjust Temperature for DeepSeek-R1 on Specific Tasks: A Complete Guide](/deepseek-ai/DeepSeek-R1/how-to-adjust-temperature-for-deepseek-r1-on-specific-tasks) Learn how to adjust temperature for DeepSeek-R1 on specific tasks. Discover the optimal default setting of 0.6 for balanced creativity and coherence, plus recommended ranges. - Tags: how-to-guide - Published: 2026-02-27 ### [How to Run DeepSeek-R1 Models Locally Using SGLang: A Complete Setup Guide](/deepseek-ai/DeepSeek-R1/how-to-run-deepseek-r1-models-locally-using-sglang) Learn to run DeepSeek-R1 models locally with SGLang. This guide shows you how to set up the asynchronous inference server for high-throughput text generation via a local HTTP API. - Tags: how-to-guide - Published: 2026-02-27 ### [How to Run DeepSeek-R1 Models Locally Using vLLM: Complete Setup Guide](/deepseek-ai/DeepSeek-R1/how-to-run-deepseek-r1-models-locally-using-vllm) Easily run DeepSeek-R1 models locally with vLLM. Follow this guide to install vLLM and launch the DeepSeek-R1 server for efficient local inference. Get started now. - Tags: how-to-guide - Published: 2026-02-27 ### [What Is the Context Length of DeepSeek-R1 Models? 128K Token Architecture Explained](/deepseek-ai/DeepSeek-R1/what-is-the-context-length-of-deepseek-r1-models) Discover the impressive 128K context length of DeepSeek-R1 models. Process long prompts code and conversations without truncation thanks to its advanced architecture. - Tags: deep-dive - Published: 2026-02-27 ### [How Many Activated Parameters Does DeepSeek-R1 Use in MoE?](/deepseek-ai/DeepSeek-R1/how-many-activated-parameters-does-deepseek-r1-use-in-moe) Discover how many activated parameters DeepSeek-R1 utilizes in its Mixture-of-Experts (MoE) architecture. Learn about its efficient inference capabilities. - Tags: deep-dive - Published: 2026-02-27 ### [DeepSeek-R1 Architecture: Understanding the MoE (Mixture of Experts) Setup](/deepseek-ai/DeepSeek-R1/deepseek-r1-architecture-understanding-the-moe-mixture-of-experts-setup) Explore the DeepSeek-R1 architecture's MoE setup. Learn how it uses a sparse subset of expert networks to achieve full-scale performance with fractional compute costs. - Tags: architecture - Published: 2026-02-27 ### [What Base Model Is DeepSeek-R1 Based On?](/deepseek-ai/DeepSeek-R1/what-base-model-is-deepseek-r1-based-on) Discover which base model powers DeepSeek-R1. Learn how DeepSeek-R1 and DeepSeek-R1 Zero are fine-tuned from the DeepSeek-V3-Base model for optimal performance. - Tags: deep-dive - Published: 2026-02-27 ### [Where to Find Distilled Versions of DeepSeek-R1 Models: Complete Download Guide](/deepseek-ai/DeepSeek-R1/where-to-find-distilled-versions-of-deepseek-r1-models) Find distilled DeepSeek-R1 models easily. Download all six openly hosted versions from Hugging Face with our complete guide. Access model details now. - Tags: getting-started - Published: 2026-02-27 ### [DeepSeek-R1-Zero Training Without Supervised Fine-Tuning: A Pure RL Approach](/deepseek-ai/DeepSeek-R1/how-does-deepseek-r1-zero-training-differ-without-supervised-fine-tuning) Discover DeepSeek-R1-Zero training, a pure RL approach that bypasses supervised fine-tuning. Learn how this method directly applies large-scale reinforcement learning to the base model for optimal LLM initialization. - Tags: deep-dive - Published: 2026-02-27 ### [How to Run DeepSeek-R1-Zero Locally: A Complete Setup Guide](/deepseek-ai/DeepSeek-R1/how-to-run-deepseek-r1-zero-model-locally) Learn how to run DeepSeek R1 Zero locally with this complete setup guide. Follow simple steps to install the inference runtime and serve the model using vLLM or SGLang for efficient local execution. - Tags: how-to-guide - Published: 2026-02-27