Training engine
Can I fine-tune this LLM?
Running a model and trainingone are very different memory problems. Fine-tuning has to hold gradients and optimizer state — often several times the weights themselves — so a model you can run might be far out of reach to fully fine-tune. Pick a model and your GPU; we'll show what full fine-tuning, LoRA, and QLoRA each need, and which fit.
Pick your GPU above to see which fine-tuning methods fit.
Just want to run it? Back to the inference calculator →