Training Guides

End-to-end recipes for fine-tuning and pretraining LLMs on Alauda AI.

Pick a path

When you want…UseGuide
Reusable templates, repeatable runs, optional Kueue quotasKubeflow Trainer v2 + LlamaFactoryFine-Tuning with Kubeflow Trainer v2
Mix training with online inference, yield GPU back on demandKueue cohort + preemption + checkpoint resumePreemptible TrainJobs with Kueue
A curated set of TrainingRuntime images (CUDA / CANN)Trainer v2 runtime catalogTraining Runtime Images
One-shot quick start of distributed PyTorch on Trainer v2ClusterTrainingRuntime + MNISTKubeflow Trainer Quick Start
Production SFT / OSFT with automatic memory managementtraining_hubFine-tuning LLMs with Training Hub
Interactive exploration, custom scripts, VolcanoJob submissionWorkbench NotebookFine-tuning LLMs using Workbench
Full-parameter SFT / pretraining on Ascend NPUWorkbench PyTorch CANN / MindSpore CANNFine-tune and Pretrain on Ascend NPU