LLaMA-Adapter

[Bachstelze]

There is a new adapter called LLaMA-Adapter, a lightweight adaption method for fine-tuning instruction-following LLaMA models fire, using 52K data provided by Stanford Alpaca.

Open source status

• The model implementation is available in the github repo
• The model weights are partially available: Variants of LLaMa are available, e.g. gpt4all GPTQ-for-LLaMa. The weights LLaMA-Adapter aren't available.
• Authors of LLaMA-Adapter are @ZrrSkywalker @csuhan @lupantech