Kubeflow Trainer
trainer.kubeflow.org
Spec sheet
- Purpose
- Kubeflow Trainer is a Kubernetes-native platform for distributed AI model training and LLM fine-tuning.Read 11 Oct 2026 · trainer.kubeflow.org
- Integrations
- The project documents integrations with Kueue, Slurm Bridge, KAI Scheduler, JobSet, and LeaderWorkerSet.Read 11 Oct 2026 · trainer.kubeflow.org
- License
- The public GitHub repository lists an Apache-2.0 license.Read 11 Oct 2026 · github.com
- Community support
- The project links users to a Kubeflow Trainer Slack channel and regular community calls.Read 11 Oct 2026 · github.com
- Data cache
- Its distributed data cache uses Apache Arrow and Apache DataFusion for zero-copy tensor streaming to GPU nodes.Read 11 Oct 2026 · trainer.kubeflow.org
- Deployment environments
- The overview says it can run in public cloud, on-premises, or hybrid environments.Read 11 Oct 2026 · trainer.kubeflow.org
- Fine-tuning
- Built-in TorchTune workflows support LoRA, QLoRA, and full fine-tuning with HuggingFace model and dataset URIs.Read 11 Oct 2026 · trainer.kubeflow.org
- Frameworks
- It provides a unified Python SDK and TrainJob API for frameworks including PyTorch, JAX, DeepSpeed, MLX, HuggingFace, Megatron, and XGBoost.Read 11 Oct 2026 · trainer.kubeflow.org
- Installation requirement
- The installation guide lists Kubernetes >= 1.31 and kubectl >= 1.31 as minimum requirements for the control plane.Read 11 Oct 2026 · trainer.kubeflow.org
- Local execution
- TrainJobs can run locally with Docker or Podman and then deploy to Kubernetes environments without code changes.Read 11 Oct 2026 · trainer.kubeflow.org
- Project status
- The GitHub repository README states that Kubeflow Trainer is currently in alpha status and its APIs may change.Read 11 Oct 2026 · github.com
- Scale
- It supports training from a single GPU to multi-node clusters, with automatic setup for DDP, FSDP, parameter servers, and gang scheduling.Read 11 Oct 2026 · trainer.kubeflow.org
- Supported users
- The documentation is organized for AI practitioners, platform administrators, and open-source contributors.Read 11 Oct 2026 · trainer.kubeflow.org
Plans and prices
Kubeflow Trainer Free Open-source project; requires Kubernetes >= 1.31 for control-plane installation Read 11 Oct 2026 · trainer.kubeflow.org
Where Specifiction places it
- Deep Learning Software#16 of 357.7