NVIDIA has released quantized version DeepSeek V3.1 FP4 on Hugging Face

Provides significant memory savings and speeds up performance when using TensorRT LLM.

At the same time, the model maintains high-quality text generation.

HuggingFace

••••••••••••••••••••••••••••••••••••••
🤖 Data Science, ML & Big Data with @DataXplore