Skip to content

llava-llama-3-8b-v1_1

An XTuner-tuned LLaVA model for image and text tasks

Visit official site

What llava-llama-3-8b-v1_1 is

The model is described as an XTuner-optimised LLaVA model designed to handle image and text together. It was fine-tuned with ShareGPT4V-PT and InternVL-SFT.

What you can do with it

  • Understand and generate text related to an image
  • Work with downstream deployment toolkits
  • Run the model at FP16 precision

Application scenarios

  • Image captions are generated for search work
  • Social media analysis combines images and text
  • A chatbot back end gains richer interaction

Pros & cons

✓ What we like

  • Compatible with several deployment and evaluation toolkits
  • The feature list cites 8.03B parameters and figures such as 72.3% and 66.4%

! What to watch out for

  • Inputs need both an image and related text prepared
  • Loading the model is done from Hugging Face

FAQ

What is it built on?

The material names Meta-Llama-3-8B-Instruct and CLIP-ViT-Large-patch14-336.

Does it support FP16?

The feature list states the model supports FP16 precision.

Last reviewed: 2026-11-15

More AI image generation tools

View all →

How we review