Task-oriented examples for doing something specific with Opik — combining Opik features, using Comet ecosystem tools together, or solving a non-obvious workflow problem.
| Guide | Description |
|---|---|
| annotation_queues_with_context/ | Structure RAG traces for Opik annotation queues — clean answer in output, context in metadata, full detail in child spans |
| multimodal_online_evaluation/ | Run an online LLM-as-judge eval over multimodal (text + image) traces — create the rule in the UI and with the SDK |
| prompt_agent_optimization/ | A-to-Z guide to prompt & agent optimization with Opik — one escalating RAG example, from an exact-match metric to LLM-judge, multi-objective, and agent/tool optimization; doubles as a live workshop (Part 1) |
| tracing_finetuned_models/ | Fine-tune a model, register it to the CometML Model Registry, then fetch and trace inference in Opik |