Fine-Tuning Studio

Prepare datasets, choose an adapter strategy, and queue model fine-tuning jobs.

Dataset first

Use an uploaded dataset ID or paste JSONL rows for a quick preview job.

LoRA, short epochs, validation split, and low learning rate are preselected.

The page works even when the managed GPU training backend is not connected.

Prepare datasets, choose an adapter strategy, and queue model fine-tuning jobs. Review the result before reusing it in another workflow.

When you need to prepare datasets, choose an adapter strategy, and queue model fine-tuning jobs.
Preparing a first draft or analysis with fine-tuning studio before reviewing it yourself.