DeepSeek's first-generation reasoning designs, attaining efficiency equivalent to OpenAI-o1 across math, code, and
thinking tasks.
Models
DeepSeek-R1
Distilled models
DeepSeek team has shown that the
thinking patterns of larger models can be
distilled into smaller models, leading to better
efficiency compared to the
thinking patterns found through RL on small models.
Below are the designs developed through
fine-tuning against
numerous dense designs commonly
utilized in the research neighborhood utilizing reasoning information
produced by DeepSeek-R1. The
assessment results show that the distilled smaller sized thick
designs perform exceptionally well on criteria.
DeepSeek-R1-Distill-Qwen-1.5 B
DeepSeek-R1-Distill-Qwen-7B
DeepSeek-R1-Distill-Llama-8B
DeepSeek-R1-Distill-Qwen-14B
DeepSeek-R1-Distill-Qwen-32B
DeepSeek-R1-Distill-Llama-70B
License
The
design weights are accredited under the MIT License. DeepSeek-R1 series assistance
business use, permit any adjustments and
derivative works,
consisting of, but not
restricted to,
distillation for
training other LLMs.
Author