# Dynamic quantization

A lightweight instruction fine-tuned language model optimized for mobile deployment

DeepSeek-R1 is the first-generation inference model launched by the DeepSeek team. Through large-scale reinforcement learning training, it does not require supervised fine-tuning (SFT) as an initial step and demonstrates excellent inference capabilities.

DeepSeek-R1 is a 1.58-bit dynamically quantized large language model optimized by Unsloth, adopting the MoE architecture and supporting English task processing.