DeepSeek AI has developed a range of distilled models based on Meta's Llama architectures, with sizes spanning from 1.5 to 70 billion parameters, starting from the foundation of DeepSeek-R1. This distillation process involves training smaller models to replicate the behavior and reasoning of the larger 671 billion parameter DeepSeek-R1 model, effectively transferring its knowledge into more compact forms. The resulting models, including DeepSeek-R1-Distill-Llama-8B (derived from Llama-3.1-8B) and DeepSeek-R1-Distill-Llama-70B (from Llama-3.3-70B-Instruct), offer varying balances between performance and resource usage. While these distilled models may exhibit slightly reduced reasoning capabilities compared to the original 671B model, they significantly enhance inference speed and lower computational costs. For example, smaller models like the 8B version process requests more quickly and use fewer resources, making them more cost-effective for production use. This NIM efficiently deploys the distilled Llama 3.1 8B variant of DeepSeek R1 models on NVIDIA GPUs. DeepSeek R1 Distill Llama 3.1 8B is available as an NVIDIA NIM™ microservice, part of https://www.nvidia.com/en-us/data-center/products/ai-enterprise/ . NVIDIA NIM offers prebuilt containers for large language models (LLMs) that can be used to develop chatbots, content analyzers—or any application that needs to understand and generate human language. Each NIM consists of a container and a model and uses a CUDA-accelerated runtime for all NVIDIA GPUs, with special optimizations available for many configurations. NVIDIA AI Enterprise
NVIDIA AI Enterprise is an end-to-end, cloud-native software platform that accelerates data science pipelines and streamlines development and deployment of production-grade co-pilots and other generative AI applications. Easy-to-use microservices provide optimized model performance with enterprise-grade security, support, and stability to ensure a smooth transition from prototype to production for enterprises that run their businesses on AI.
