Bucket
Guide (how-to for an Opik workflow)
What & why
Show how to trace vision-language model (image + text) inference in Opik, capturing multimodal calls as traces. The focus is the nuances of visualising the inputs — the images sent to the model — alongside the text outputs in the trace view, so image-in / text-out interactions are easy to inspect. (Distinct from guides/multimodal_online_evaluation, which focuses on online LLM-as-judge evaluation; this guide is about tracing and input/output visualisation.)
Opik products / features involved
Bucket
Guide (how-to for an Opik workflow)
What & why
Show how to trace vision-language model (image + text) inference in Opik, capturing multimodal calls as traces. The focus is the nuances of visualising the inputs — the images sent to the model — alongside the text outputs in the trace view, so image-in / text-out interactions are easy to inspect. (Distinct from
guides/multimodal_online_evaluation, which focuses on online LLM-as-judge evaluation; this guide is about tracing and input/output visualisation.)Opik products / features involved