LLaVA community / model projects
About LLaVA community / model projects
LLaVA is a research project on visual instruction tuning and language-and-vision assistants. The original project connects a vision encoder with a language model and releases supporting models, code and data. It is a multi-institution research effort rather than a single model-serving company.
Focus
- Visual instruction tuning
- Language-and-vision models
- Research code and datasets
Projects and interfaces
LLaVA
multimodal research project
Research models combine a visual encoder and a language model for visual-language tasks.
Scope and limitations
- Community versions and later derivatives may use different base models and training conditions.
Official references
- LLaVA official research project. Checked .
Models
29 named catalog records. These are name and distribution records, not a count of distinct model architectures.
- bakllava:7b
- bakllava:latest
- bakllava:latest-128k
- bakllava:latest-16k
- bakllava:latest-32k
- bakllava:latest-64k
- bakllava:latest-8k
- llava-llama3:latest
- llava-llama3:latest-128k
- llava-llama3:latest-16k
- llava-llama3:latest-32k
- llava-llama3:latest-64k
- llava-llama3:latest-8k
- llava-phi3:3.8b
- llava-phi3:latest
- llava-phi3:latest-128k
- llava-phi3:latest-16k
- llava-phi3:latest-32k
- llava-phi3:latest-64k
- llava-phi3:latest-8k
- llava:13b
- llava:13b-128k
- llava:13b-16k
- llava:13b-32k
- llava:13b-64k
- llava:13b-8k
- llava:latest
- llava:latest-128k
- llava:latest-64k
Model families
- bakllava
- llava
- llava-llama3
- llava-phi3