Update data-pipelines.md

content/docs/start/data-pipelines.md

@@ -292,17 +292,20 @@ prepare:
 DVC pipelines (`dvc.yaml` file, `dvc run`, and `dvc repro` commands) solve a few
 important problems:
 
-- _Automation_ - run sequence of steps in a "smart" way that makes iterating on
-  the project faster. It automatically determines which parts of a project need
-  to be run, it caches "runs" and results — all to avoid running the same stage
-  again.
-- _Reproducibility_ - it can describe and capture what data should be used and
-  what commands to run to produce an ML model, for example. It's described and
-  captured in way that is easy to put into Git. It means that it's easy to
-  version and share.
-- _Continuous Delivery and Continuous Integration (CI/CD) for ML_ - describing
-  project in way that it can be reproduced (built) is the fist necessary step
-  before introducing CI/CD systems.
+- _Automation_ - run a sequence of steps in a "smart" way to make iterating on
+  your project faster. DVC caches "runs" and results in stages and automatically
+  determines which parts of a project need to be run to avoid unnecessary
+  re-runs.
+- _Reproducibility_ - `dvc.yaml` and `dvc.lock` files describe and capture what
+  data to use and what commands to run to produce an ML model. Storing these
+  files in Git makes it easy to version and share.
+- _Continuous Delivery and Continuous Integration (CI/CD) for ML_ - reproducible
+  ML pipelines allow CI/CD systems to retrain
+  ([repro](https://dvc.org/doc/command-reference/repro)) models on fresh
+  datasets with identical preprocessing and training stages, version upstream
+  models and datasets, and easily
+  [compare](https://dvc.org/doc/start/experiments#comparing-experiments) metrics
+  with currently deployed models.
 
 ## Visualize