Fine-Tune Llama Models with Kubeflow Training Operator

In this repo, you will find an example showing how to fine-tune Large Language Models with the Kubeflow Training Operator on OpenShift AI. This example uses tools like HuggingFace SFTTrainer, PEFT (LoRA) and PyTorch FSDP, for Llama 3.1 and 3.3 models. This approach is built for Kubernetes and provides a simple way to train large models across multiple GPUs (Nvidia & AMD)

Openshift AI workbench setup with all required components and shared storage:

TensorBoard dashboard showing successful model finetuning using NVIDIA GPUs:

TensorBoard dashboard showing successful model finetuning on AMD GPUs:

---

1. [https://github.com/opendatahub-io/distributed-workloads/tree/main/examples/kfto-sft-llm] ↩