hiyouga/LlamaFactory — fine-tune 100+ LLMs in one framework
Single codebase handles fine-tuning across Llama, Qwen, Deepseek, Gemma, and dozens more models with minimal setup. Supports LoRA, QLoRA, full tuning, and vision models—no need to rewrite for each architecture. Built for practitioners who want reproducible, efficient training without framework switching.
#agent#ai#deepseek#fine-tuning