Library / SDK
Fine-tuning & Training
Unsloth
1Faster 2memory-efficient 3LLM 4fine-tuning 5made 6easy
the six Ws · specification
W1
Who
Maintained by Unsloth AI (unslothai).
W2
What
A library that accelerates LLM fine-tuning and cuts memory use via optimized kernels and QLoRA/LoRA.
W3
Where
A Python library on GPUs, including single consumer GPUs and free notebooks.
W4
When
Reach for it when fine-tuning open models faster on limited VRAM.
W5
Why
Custom Triton kernels cut training time and memory while keeping accuracy.
W6
With
An NVIDIA GPU (CUDA), PyTorch, and base model weights.
W7
Watch
Built by Unsloth AI; open-source core (Apache-2.0) with a paid pro offering (open-core); runs locally on your own GPU.