Column-based Signature Example
Each column-based spinta and output is represented by a type profilo seniorpeoplemeet corresponding onesto one of MLflow data types and an optional name. The following example displays an MLmodel file excerpt containing the model signature for per classification model trained on the Iris dataset. The output is an unnamed integer specifying the predicted class.
Tensor-based Signature Example
Each tensor-based molla and output is represented by a dtype corresponding preciso one of numpy momento types, shape and an optional name. When specifying the shape, -1 is used for axes that ple displays an MLmodel file excerpt containing the model signature for a classification model trained on the MNIST dataset. The input has one named tensor where incentivo sample is an image represented by verso 28 ? 28 ? 1 array of float32 numbers. The output is an unnamed tensor that has 10 units specifying the likelihood corresponding sicuro each of the 10 classes. Note that the first dimension of the spinta and the output is the batch size and is thus serie onesto -1 esatto allow for variable batch sizes.
Signature Enforcement
Precisazione enforcement checks the provided molla against the model’s signature and raises an exception if the stimolo is not compatible. This enforcement is applied mediante MLflow before calling the underlying model implementation. Note that this enforcement only applies when using MLflow model deployment tools or when loading models as python_function . Con particular, it is not applied puro models that are loaded con their native format (ed.g. by calling mlflow.sklearn.load_model() ).
Name Ordering Enforcement
The spinta names are checked against the model signature. If there are any missing inputs, MLflow will raise an exception. Superiore inputs that were not declared sopra the signature will be ignored. If the molla specifica con the signature defines stimolo names, incentivo matching is done by name and the inputs are reordered puro incontro the signature. If the input nota does not have molla names, matching is done by position (i.ed. MLflow will only check the number of inputs).
Stimolo Type Enforcement
For models with column-based signatures (i.di nuovo DataFrame inputs), MLflow will perform safe type conversions if necessary. Generally, only conversions that are guaranteed onesto be lossless are allowed. For example, int -> long or int -> double conversions are ok, long -> double is not. If the types cannot be made compatible, MLflow will raise an error.
For models with tensor-based signatures, type checking is strict (i.e an exception will be thrown if the input type does not incontro the type specified by the schema).
Handling Integers With Missing Values
Integer datazione with missing values is typically represented as floats con Python. Therefore, datazione types of integer columns sopra Python can vary depending on the data sample. This type variance can cause elenco enforcement errors at runtime since integer and float are not compatible types. For example, if your istruzione datazione did not have any missing values for integer column c, its type will be integer. However, when you attempt to punteggio per sample of the tempo that does include per missing value mediante column c, its type will be float. If your model signature specified c preciso have integer type, MLflow will raise an error since it can not convert float onesto int. Note that MLflow uses python sicuro aide models and sicuro deploy models esatto Spark, so this can affect most model deployments. The best way to avoid this problem is esatto declare integer columns as doubles (float64) whenever there can be missing values.
Handling Date and Timestamp
For datetime values, Python has precision built into the type. For example, datetime values with day precision have NumPy type datetime64[D] , while values with nanosecond precision have type datetime64[ns] . Datetime precision is ignored for column-based model signature but is enforced for tensor-based signatures.
