OpenGVLab/LLaMA-Adapter
A parameter-efficient fine-tuning method for adapting LLaMA models to follow instructions and support multi-modal inputs.
Not currently ranked — collecting fresh signals.
star history
LLaMA-Adapter provides a lightweight adapter-based approach for fine-tuning LLaMA models to follow instructions and support multi-modal inputs. It introduces zero-init attention mechanisms to efficiently adapt pre-trained language models with minimal additional parameters. The project includes both instruction-following and visual instruction models, with V2 extending to multi-modal reasoning.
Frequently asked
- What is OpenGVLab/LLaMA-Adapter?
- A parameter-efficient fine-tuning method for adapting LLaMA models to follow instructions and support multi-modal inputs.
- Is LLaMA-Adapter open source?
- Yes — OpenGVLab/LLaMA-Adapter is open source, released under the GPL-3.0 license.
- What language is LLaMA-Adapter written in?
- OpenGVLab/LLaMA-Adapter is primarily written in Python.
- How popular is LLaMA-Adapter?
- OpenGVLab/LLaMA-Adapter has 5.9k stars on GitHub.
- Where can I find LLaMA-Adapter?
- OpenGVLab/LLaMA-Adapter is on GitHub at https://github.com/OpenGVLab/LLaMA-Adapter.