README.md

metadata

title: README
emoji: 🚀
colorFrom: pink
colorTo: red
sdk: static
pinned: true

FuriosaAI

Welcome to Furiosa AI’s official organization on Hugging Face!

Here, you can find pre-compiled models in the Furiosa AI artifact format, built from pre-trained models on the Hugging Face Hub. You can also compile additional models with different optimization configurations. For detailed instructions, please refer to our Model Preparation Workflow. In addition, visit Supported Models of Furiosa SDK in the Furiosa SDK documentation for a complete list of models supported by Furiosa AI RNGD.

Pre-compiled models

Below are the pre-compiled models built from pre-trained models available on the Hugging Face Hub. Please note that you will need Furiosa SDK 2025.2 or later to use these models.

Pre-compiled Model	Description	Base Model
furiosa-ai/bert-large-uncased-INT8	INT8 quantized, optimized for MLPerf	google-bert/bert-large-uncased
furiosa-ai/DeepSeek-R1-Distill-Llama-8B	BF16	deepseek-ai/DeepSeek-R1-Distill-Llama-8B
furiosa-ai/EXAONE-3.5-7.8B-Instruct	BF16	LGAI-EXAONE/EXAONE-3.5-7.8B-Instruct
furiosa-ai/Llama-3.1-8B-Instruct	BF16	meta-llama/Llama-3.1-8B-Instruct
furiosa-ai/Llama-3.1-8B-Instruct-FP8	FP8 quantized	meta-llama/Llama-3.1-8B-Instruct