Description
Fine-tuning Qwen or Llama on your data means different scripts and parameters for each model and never enough VRAM. LlamaFactory is a framework for unified, efficient fine-tuning of 100+ LLMs and VLMs.
It supports LoRA, QLoRA, full tuning and RLHF with a no-code web UI, used by Amazon, NVIDIA and Aliyun.
100+ models:Qwen, Llama and DeepSeek.
Efficient:LoRA and QLoRA.
Alignment:RLHF and DPO.
Web UI:No code.
It supports LoRA, QLoRA, full tuning and RLHF with a no-code web UI, used by Amazon, NVIDIA and Aliyun.
Features
100+ models:Qwen, Llama and DeepSeek.
Efficient:LoRA and QLoRA.
Alignment:RLHF and DPO.
Web UI:No code.
