DeepSeek-R1 · GitHub Models · GitHub

DeepSeek-R1 excels at thinking jobs using a detailed training procedure, such as language, scientific thinking, and coding tasks. It features 671B overall criteria with 37B active specifications, and 128k context length.

DeepSeek-R1 builds on the progress of earlier reasoning-focused designs that enhanced performance by extending Chain-of-Thought (CoT) reasoning. DeepSeek-R1 takes things even more by integrating reinforcement learning (RL) with fine-tuning on thoroughly picked datasets. It progressed from an earlier variation, DeepSeek-R1-Zero, which relied solely on RL and revealed strong reasoning skills but had problems like hard-to-read outputs and language inconsistencies. To attend to these restrictions, DeepSeek-R1 includes a percentage of cold-start data and follows a refined training pipeline that blends reasoning-oriented RL with monitored fine-tuning on curated datasets, resulting in a design that achieves state-of-the-art performance on thinking criteria.

Usage Recommendations

We suggest to the following configurations when making use of the DeepSeek-R1 series models, including benchmarking, to accomplish the anticipated performance:

– Avoid including a system prompt; all guidelines need to be included within the user timely.
– For mathematical problems, it is suggested to include a regulation in your timely such as: “Please reason step by step, and put your final answer within boxed .”.
– When assessing design performance, it is recommended to conduct several tests and average the outcomes.

Additional suggestions

The model’s thinking output (contained within the tags) might consist of more damaging material than the design’s last reaction. Consider how your application will utilize or show the thinking output; you might desire to suppress the thinking output in a production setting.