LLM-Adapters is an easy-to-use framework that integrates various adapter methods into Large Language Models to execute parameter-efficient fine-tuning for different tasks. It supports multiple adapter families including LoRA, Bottleneck adapters, Parallel adapters, Prefix Tuning, and Adapter variants. The framework extends HuggingFace’s PEFT library and enables training on open-access LLMs like LLaMA, OPT, BLOOM, and GPT-J with minimal parameter updates.