GitLab

DeepSeek open-sourced DeepSeek-R1, an LLM fine-tuned with reinforcement learning (RL) to enhance thinking ability. DeepSeek-R1 attains outcomes on par with OpenAI's o1 model on a number of standards, including MATH-500 and SWE-bench.

DeepSeek-R1 is based on DeepSeek-V3, a mix of specialists (MoE) model recently open-sourced by DeepSeek. This base model is fine-tuned using Group Relative Policy Optimization (GRPO), a reasoning-oriented variation of RL. The research group also performed understanding distillation from DeepSeek-R1 to open-source Qwen and Llama designs and launched a number of variations of each; these designs outshine bigger designs, including GPT-4, on math and coding criteria.

[DeepSeek-R1 is] the initial step toward improving language model reasoning capabilities using pure reinforcement learning (RL). Our goal is to check out the potential of LLMs to establish thinking abilities with no monitored data, focusing on their self-evolution through a pure RL process...DeepSeek-R1 ... master a large range of tasks, consisting of innovative writing, basic concern answering, modifying, summarization, and more. Additionally, DeepSeek-R1 shows impressive performance on jobs needing long-context understanding, significantly outperforming DeepSeek-V3 on long-context criteria.

To develop the design, DeepSeek began with DeepSeek-V3 as a base. They initially attempted fine-tuning it only with RL, and with no supervised fine-tuning (SFT), producing a model called DeepSeek-R1-Zero, which they have likewise launched. This design displays strong reasoning efficiency, however" powerful reasoning habits, it faces numerous issues. For example, DeepSeek-R1-Zero struggles with challenges like bad readability and language mixing."

To address this, the group utilized a short stage of SFT to prevent the "cold start" issue of RL. They collected a number of thousand examples of chain-of-thought reasoning to use in SFT of DeepSeek-V3 before running RL. After the RL process converged, they then collected more SFT data utilizing rejection sampling, resulting in a dataset of 800k samples. This dataset was utilized for further fine-tuning and to the distilled designs from Llama and Qwen.

DeepSeek examined their design on a range of thinking, math, and coding criteria and compared it to other designs, consisting of Claude-3.5- Sonnet, GPT-4o, and o1. DeepSeek-R1 outshined all of them on several of the criteria, consisting of AIME 2024 and MATH-500.

DeepSeek-R1 Performance. Image Source: DeepSeek-R1 Technical Report

Within a few days of its release, the LMArena announced that DeepSeek-R1 was ranked # 3 total in the arena and # 1 in coding and mathematics. It was likewise tied for # 1 with o1 in "Hard Prompt with Style Control" category.

Django framework co-creator Simon Willison blogged about his experiments with one of the DeepSeek distilled Llama designs on his blog:

Each response starts with a ... pseudo-XML tag containing the chain of idea used to assist create the action. [Given the prompt] "a joke about a pelican and a walrus who run a tea space together" ... It then thought for 20 paragraphs before outputting the joke! ... [T] he joke is awful. But the procedure of arriving was such a fascinating insight into how these brand-new models work.

Andrew Ng's newsletter The Batch blogged about DeepSeek-R1:

DeepSeek is quickly emerging as a strong home builder of open designs. Not only are these designs excellent entertainers, however their license allows usage of their outputs for distillation, potentially pushing forward the cutting-edge for language designs (and multimodal designs) of all sizes.

The DeepSeek-R1 models are available on HuggingFace.

About the Author

Anthony Alford

Rate this Article

This material remains in the AI, ML & Data Engineering topic

Related Topics:

- AI, ML & Data Engineering

• Generative AI
• Large language models
  
  - Related Editorial
  
  Related Sponsored Content
  
  - [eBook] Getting Going with Azure Kubernetes Service
  
  Related Sponsor
  
  Free services for AI apps. Are you prepared to try out advanced technologies? You can begin constructing smart apps with totally free Azure app, hb9lc.org information, and AI services to reduce upfront expenses. Learn More.
  
  How could we improve? Take the InfoQ reader study
  
  Each year, we look for feedback from our readers to assist us improve InfoQ. Would you mind spending 2 minutes to share your feedback in our brief survey? Your feedback will straight assist us continually evolve how we support you. The InfoQ Team Take the study
  
  Related Content
  
  The InfoQ Newsletter
  
  A round-up of last week's material on InfoQ sent every Tuesday. Join a neighborhood of over 250,000 senior wavedream.wiki designers.