mlflow-onboarding

To get started with MLflow for LLM and RAG pipelines, determine your GenAI use case and follow the tailored quickstart tutorials to integrate features like tracing for observability, LLM judge evaluations, and prompt management.

The best way to set up MLflow tracking for traditional ML models using scikit-learn, PyTorch, TensorFlow, or XGBoost is to follow the traditional ML quickstart tutorials to log parameters, metrics, artifacts, and deploy models.

Yes, MLflow onboarding supports both traditional ML and GenAI workflows. It guides you through specific integration steps and quickstart tutorials based on whether you are building LLM-powered apps or traditional deep learning models.

To use MLflow tracing for observability in GenAI applications, identify your use case as GenAI, then follow the guided quickstart tutorials which cover integrating observability features for chatbots, tool-calling agents, and RAG pipelines.

Yes, you can log parameters, metrics, and artifacts for PyTorch and TensorFlow models. By selecting the traditional ML onboarding path, you receive guidance on integrating experiment tracking and model logging into your workflow.

No, an experiment ID is not strictly required to start integrating MLflow, but providing one as input helps tailor the onboarding guidance to your specific experimentation and tracking setup.