Skip to content

[Guide]: Executing VLM models in Opik #49

Description

@LeoRoccoBreedt

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

  • Tracing / observability

Metadata

Metadata

Labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions