DeepSeek-R1 · GitHub Models · GitHub

DeepSeek-R1 excels at reasoning jobs using a step-by-step training procedure, such as language, scientific thinking, and coding jobs. It includes 671B overall criteria with 37B active criteria, and 128k context length.

DeepSeek-R1 builds on the development of earlier reasoning-focused models that improved performance by extending Chain-of-Thought (CoT) thinking. DeepSeek-R1 takes things further by combining support learning (RL) with fine-tuning on thoroughly picked datasets. It progressed from an earlier version, DeepSeek-R1-Zero, which relied solely on RL and showed strong reasoning abilities but had concerns like hard-to-read outputs and language inconsistencies. To attend to these constraints, DeepSeek-R1 incorporates a small quantity of cold-start information and follows a refined training pipeline that mixes reasoning-oriented RL with supervised fine-tuning on curated datasets, resulting in a model that achieves advanced efficiency on thinking standards.

Usage Recommendations

We recommend adhering to the following setups when using the DeepSeek-R1 series models, consisting of benchmarking, to attain the efficiency:

– Avoid adding a system prompt; all guidelines ought to be consisted of within the user prompt.
– For mathematical issues, it is suggested to include a regulation in your prompt such as: “Please reason step by step, and put your final answer within boxed .”.
– When examining model efficiency, it is suggested to carry out several tests and balance the results.

Additional recommendations

The model’s reasoning output (consisted of within the tags) might contain more damaging content than the design’s last action. Consider how your application will utilize or display the thinking output; you may wish to reduce the reasoning output in a production setting.