Stage 2 · Lesson 3 of 7
Reproducible stages with dvc.yaml
can also describe how outputs are made. In dvc.yaml you declare , each with a command, its dependencies (input data and code files), its from params.yaml and its outputs. dvc repro runs the stages in order and skips any whose inputs have not changed, and dvc.lock records the exact hashes of everything used.
Metrics files declared in a stage can be compared across commits with dvc metrics diff. Together this proves which data, code and parameters produced a model, and rebuilding it is one command. It is a lightweight for one machine; later stages add an for scheduling and retries.
Common trapleaving a script out of a stage's dependencies, so a code change does not trigger a rerun and stale outputs look current.
A question to check yourself, an AI practice session and 1 cheat sheet entry come with the full roadmap. Unlock
Resources for this stage
Start with
- CourseFreeMade With ML
Goku Mohandas · Advanced
Free: the versioning and data testing lessons, built around one real project