What benefit does logging an input_example with mlflow.sklearn.log_model() provide?
Select an answer to reveal the explanation.
Short Explanation and Infographic
An input example is a sample row that gets attached to your model artifact — it documents what format the model expects and powers the 'Try it out' feature in Model Serving.
Full explanation below image
![Infographic 1024x768 showing inputexample benefits: logmodel call with inputexample=Xtrain[:5] parameter highlighted. Three outputs shown: 1) Model Serving UI showing sample JSON payload (curl example), 2) Inferred signature schema from example types, 3) inputexample.json artifact file stored in run. Model artifact directory tree showing MLmodel, conda.yaml, model.pkl, inputexample.json files.](/media/databricks/dbml/DBML-q058-infographic.webp)
Full Explanation
mlflow.sklearn.log_model(model, 'model', input_example=X_train[:5]) logs the first 5 training rows as a JSON artifact alongside the model. Benefits: 1) API documentation — the Databricks Model Serving UI shows the example in a curl command or JSON input field so users know exactly what format to send. 2) Signature inference — if signature is not explicitly provided, MLflow can infer it from input_example plus the model's prediction on that example. 3) Testing — the example can be used to verify the model loads correctly after serialization. The input_example is stored as a JSON file in the artifact directory. It does NOT store the full training dataset (only a small sample), perform data validation, or trigger retraining. Providing both signature and input_example is best practice for production models.