Unsloth
Open-source toolkit for fine-tuning and running large language models faster and with less memory.
About
Unsloth is designed for developers and researchers who want to fine-tune models on consumer-grade hardware without the cost of cloud GPUs. Its custom kernels accelerate LoRA and FP8 training while reducing VRAM usage significantly compared to standard approaches. The platform supports 500+ models across text, vision, audio, and embedding tasks, and includes a no-code training interface, automatic dataset creation from PDFs or CSVs, and export to GGUF or Safetensors formats for use with llama.cpp, vLLM, or Ollama. The core library is free and open-source; Pro and Enterprise tiers with multi-GPU and multi-node support are available on request.