TrainShellTrain models. Run agents.
Fine-tuning, post-training, and agent workflows in one native workspace.
macOSWindowsLinuxiOSAndroidPrivate beta
Vaults ⌄
Training2
New training
Run name
Instruction tuning
Training type
Text fine-tuning
Parameter mode
LoRA
Base model
Qwen 3 4B
Dataset
Upload your dataset⌄ Training settingsInstruction dataset · 1,200 examples
Context length
2,048
LoRA rank
8
Epochs
3
Learning rate
0.0001
Fine-tune models on your data.
Prepare text or vision datasets, choose a model, and track training runs.
- LoRA and full-parameter training
- Training metrics
- Run reports