Beambutbetter/Deepseek-V2-Lite-16B-NVFP4

Model Card for Model ID

This modelcard aims to be a base template for new models. It has been generated using this raw template.

Model Details

Model Description

DeepSeek-V2-Lite Quantized to NVFP4 Using TensorRT-Model-Optimizer. The model has 15.6B parameters in total, in Nvidia float precision 4 bit. Require Blackwell GPU, NVFP4 supported inference engine(if use with inference engine)

• Developed by: Krisakorn Chanthasang
• Model type: LLM TextGeneration
• Language(s) (NLP): Chinese English Thai
• License: Apache 2.0
• Quantized from model: DeepSeek-V2-Lite

Model Information

Read https://huggingface.co/deepseek-ai/DeepSeek-V2-Lite for Basic model information

How to Get Started with the Model

GPU requirement: Blackwell Series

Hardware

Quantization Hardware: Nvidia RTX PRO 6000 Blackwell Workstation