Skip to content

LLaVA++

Extend LLaVA vision with Phi-3 and LLaMA-3

Visit official site
Category AI models

What LLaVA++ is

The project is described as an open-source effort to extend LLaVA visual abilities by combining newer large language models. It was built by researchers at Mohamed bin Zayed University of AI.

What you can do with it

  • Use Phi-3 Mini Instruct and LLaMA-3 Instruct together
  • Start from pre-trained models or LoRA weights
  • Try an interactive chat in Google Colab

Application scenarios

  • Education tools assist language learning
  • Business systems add it to customer service
  • Research groups study language models

Pros & cons

✓ What we like

  • Performance compared across several benchmarks and datasets
  • Detailed installation and training instructions

! What to watch out for

  • Pre-training or fine-tuning is chosen per use case
  • The interactive chat runs through Google Colab

FAQ

Which models are integrated?

The feature list names Phi-3 Mini Instruct and LLaMA-3 Instruct.

Who developed it?

The material names researchers at Mohamed bin Zayed University of AI.

Last reviewed: 2026-11-15

More LLM API platform tools

View all →

How we review