snowflake-ml-python 1.6.4__py3-none-any.whl → 1.7.1__py3-none-any.whl

snowflake/ml/monitoring/entities/model_monitor_config.py

@@ -1,17 +1,19 @@
 from dataclasses import dataclass
-from typing import List
+from typing import List, Optional
 
 from snowflake.ml.model._client.model import model_version_impl
-from snowflake.ml.monitoring.entities import model_monitor_interval
 
 
 @dataclass
-class ModelMonitorTableConfig:
-    source_table: str
+class ModelMonitorSourceConfig:
+    source: str
     timestamp_column: str
-    prediction_columns: List[str]
-    label_columns: List[str]
     id_columns: List[str]
+    prediction_score_columns: Optional[List[str]] = None
+    prediction_class_columns: Optional[List[str]] = None
+    actual_score_columns: Optional[List[str]] = None
+    actual_class_columns: Optional[List[str]] = None
+    baseline: Optional[str] = None
 
 
 @dataclass
@@ -22,7 +24,5 @@ class ModelMonitorConfig:
     model_function_name: str
     background_compute_warehouse_name: str
     # TODO: Add support for pythonic notion of time.
-    refresh_interval: str = model_monitor_interval.ModelMonitorRefreshInterval.DAILY
-    aggregation_window: model_monitor_interval.ModelMonitorAggregationWindow = (
-        model_monitor_interval.ModelMonitorAggregationWindow.WINDOW_1_DAY
-    )
+    refresh_interval: str = "1 hour"
+    aggregation_window: str = "1 day"

snowflake/ml/monitoring/model_monitor.py

@@ -0,0 +1,37 @@
+from snowflake.ml._internal import telemetry
+from snowflake.ml._internal.utils import sql_identifier
+from snowflake.ml.monitoring._client import model_monitor_sql_client
+
+
+class ModelMonitor:
+    """Class to manage instrumentation of Model Monitoring and Observability"""
+
+    name: sql_identifier.SqlIdentifier
+    _model_monitor_client: model_monitor_sql_client.ModelMonitorSQLClient
+
+    statement_params = telemetry.get_statement_params(
+        telemetry.TelemetryProject.MLOPS.value,
+        telemetry.TelemetrySubProject.MONITORING.value,
+    )
+
+    def __init__(self) -> None:
+        raise RuntimeError("ModelMonitor's initializer is not meant to be used.")
+
+    @classmethod
+    def _ref(
+        cls,
+        model_monitor_client: model_monitor_sql_client.ModelMonitorSQLClient,
+        name: sql_identifier.SqlIdentifier,
+    ) -> "ModelMonitor":
+        self: "ModelMonitor" = object.__new__(cls)
+        self.name = name
+        self._model_monitor_client = model_monitor_client
+        return self
+
+    def suspend(self) -> None:
+        """Suspend pipeline for ModelMonitor"""
+        self._model_monitor_client.suspend_monitor(self.name, statement_params=self.statement_params)
+
+    def resume(self) -> None:
+        """Resume pipeline for ModelMonitor"""
+        self._model_monitor_client.resume_monitor(self.name, statement_params=self.statement_params)

snowflake/ml/registry/_manager/model_manager.py

@@ -3,10 +3,11 @@ from typing import Any, Dict, List, Optional, Union
 
 import pandas as pd
 from absl.logging import logging
+from packaging import version
 
 from snowflake.ml._internal import telemetry
 from snowflake.ml._internal.human_readable_id import hrid_generator
-from snowflake.ml._internal.utils import sql_identifier
+from snowflake.ml._internal.utils import snowflake_env, sql_identifier
 from snowflake.ml.model import model_signature, type_hints as model_types
 from snowflake.ml.model._client.model import model_impl, model_version_impl
 from snowflake.ml.model._client.ops import metadata_ops, model_ops, service_ops
@@ -45,6 +46,7 @@ class ModelManager:
         metrics: Optional[Dict[str, Any]] = None,
         conda_dependencies: Optional[List[str]] = None,
         pip_requirements: Optional[List[str]] = None,
+        target_platforms: Optional[List[model_types.SupportedTargetPlatformType]] = None,
         python_version: Optional[str] = None,
         signatures: Optional[Dict[str, model_signature.ModelSignature]] = None,
         sample_input_data: Optional[model_types.SupportedDataType] = None,
@@ -85,6 +87,7 @@ class ModelManager:
             metrics=metrics,
             conda_dependencies=conda_dependencies,
             pip_requirements=pip_requirements,
+            target_platforms=target_platforms,
             python_version=python_version,
             signatures=signatures,
             sample_input_data=sample_input_data,
@@ -105,6 +108,7 @@ class ModelManager:
         metrics: Optional[Dict[str, Any]] = None,
         conda_dependencies: Optional[List[str]] = None,
         pip_requirements: Optional[List[str]] = None,
+        target_platforms: Optional[List[model_types.SupportedTargetPlatformType]] = None,
         python_version: Optional[str] = None,
         signatures: Optional[Dict[str, model_signature.ModelSignature]] = None,
         sample_input_data: Optional[model_types.SupportedDataType] = None,
@@ -143,6 +147,15 @@ class ModelManager:
             statement_params=statement_params,
         )
 
+        platforms = None
+        # TODO(jbahk): Remove the version check after Snowflake 8.40.0 release
+        # User specified target platforms are defaulted to None and will not show up in the generated manifest.
+        # In the backend, we attempt to create a model for all platforms (WH, SPCS) regardless by default.
+        if snowflake_env.get_current_snowflake_version(self._model_ops._session) >= version.parse("8.40.0"):
+            # Convert any string target platforms to TargetPlatform objects
+            if target_platforms:
+                platforms = [model_types.TargetPlatform(platform) for platform in target_platforms]
+
         logger.info("Start packaging and uploading your model. It might take some time based on the size of the model.")
 
         mc = model_composer.ModelComposer(
@@ -155,6 +168,7 @@ class ModelManager:
             sample_input_data=sample_input_data,
             conda_dependencies=conda_dependencies,
             pip_requirements=pip_requirements,
+            target_platforms=platforms,
             python_version=python_version,
             code_paths=code_paths,
             ext_modules=ext_modules,

snowflake/ml/registry/registry.py

@@ -14,11 +14,8 @@ from snowflake.ml.model import (
     type_hints as model_types,
 )
 from snowflake.ml.model._client.model import model_version_impl
-from snowflake.ml.monitoring._client import (
-    model_monitor,
-    model_monitor_manager,
-    model_monitor_version,
-)
+from snowflake.ml.monitoring import model_monitor, model_monitor_version
+from snowflake.ml.monitoring._manager import model_monitor_manager
 from snowflake.ml.monitoring.entities import model_monitor_config
 from snowflake.ml.registry._manager import model_manager
 from snowflake.snowpark import session
@@ -26,6 +23,11 @@ from snowflake.snowpark import session
 _TELEMETRY_PROJECT = "MLOps"
 _MODEL_TELEMETRY_SUBPROJECT = "ModelManagement"
 
+_MODEL_MONITORING_UNIMPLEMENTED_ERROR = "Model Monitoring is not implemented in python yet."
+_MODEL_MONITORING_DISABLED_ERROR = (
+    """Must enable monitoring to use this method. Please set `options={"enable_monitoring": True}` in the Registry"""
+)
+
 
 class Registry:
     def __init__(
@@ -87,7 +89,6 @@ class Registry:
                 session=session,
                 database_name=self._database_name,
                 schema_name=self._schema_name,
-                create_if_not_exists=True,  # TODO: Support static setup method to configure schema for monitoring.
                 statement_params=monitor_statement_params,
             )
 
@@ -107,6 +108,7 @@ class Registry:
         metrics: Optional[Dict[str, Any]] = None,
         conda_dependencies: Optional[List[str]] = None,
         pip_requirements: Optional[List[str]] = None,
+        target_platforms: Optional[List[model_types.SupportedTargetPlatformType]] = None,
         python_version: Optional[str] = None,
         signatures: Optional[Dict[str, model_signature.ModelSignature]] = None,
         sample_input_data: Optional[model_types.SupportedDataType] = None,
@@ -128,14 +130,17 @@ class Registry:
             metrics: A JSON serializable dictionary containing metrics linked to the model version. Defaults to None.
             signatures: Model data signatures for inputs and outputs for various target methods. If it is None,
                 sample_input_data would be used to infer the signatures for those models that cannot automatically
-                infer the signature. If not None, sample_input_data should not be specified. Defaults to None.
-            sample_input_data: Sample input data to infer model signatures from. Defaults to None.
+                infer the signature. Defaults to None.
+            sample_input_data: Sample input data to infer model signatures from.
+                It would also be used as background data in explanation and to capture data lineage. Defaults to None.
             conda_dependencies: List of Conda package specifications. Use "[channel::]package [operator version]" syntax
                 to specify a dependency. It is a recommended way to specify your dependencies using conda. When channel
                 is not specified, Snowflake Anaconda Channel will be used. Defaults to None.
             pip_requirements: List of Pip package specifications. Defaults to None.
                 Currently it is not supported since Model can only executed in Snowflake Warehouse where all
                 dependencies are required to be retrieved from Snowflake Anaconda Channel.
+            target_platforms: List of target platforms to run the model. The only acceptable inputs are a combination of
+                {"WAREHOUSE", "SNOWPARK_CONTAINER_SERVICES"}. Defaults to None.
             python_version: Python version in which the model is run. Defaults to None.
             code_paths: List of directories containing code to import. Defaults to None.
             ext_modules: List of external modules to pickle with the model object.
@@ -190,6 +195,7 @@ class Registry:
             "metrics",
             "conda_dependencies",
             "pip_requirements",
+            "target_platforms",
             "python_version",
             "signatures",
         ],
@@ -204,6 +210,7 @@ class Registry:
         metrics: Optional[Dict[str, Any]] = None,
         conda_dependencies: Optional[List[str]] = None,
         pip_requirements: Optional[List[str]] = None,
+        target_platforms: Optional[List[model_types.SupportedTargetPlatformType]] = None,
         python_version: Optional[str] = None,
         signatures: Optional[Dict[str, model_signature.ModelSignature]] = None,
         sample_input_data: Optional[model_types.SupportedDataType] = None,
@@ -229,13 +236,16 @@ class Registry:
             signatures: Model data signatures for inputs and outputs for various target methods. If it is None,
                 sample_input_data would be used to infer the signatures for those models that cannot automatically
                 infer the signature. If not None, sample_input_data should not be specified. Defaults to None.
-            sample_input_data: Sample input data to infer model signatures from. Defaults to None.
+            sample_input_data: Sample input data to infer model signatures from.
+                It would also be used as background data in explanation and to capture data lineage. Defaults to None.
             conda_dependencies: List of Conda package specifications. Use "[channel::]package [operator version]" syntax
                 to specify a dependency. It is a recommended way to specify your dependencies using conda. When channel
                 is not specified, Snowflake Anaconda Channel will be used. Defaults to None.
             pip_requirements: List of Pip package specifications. Defaults to None.
                 Currently it is not supported since Model can only executed in Snowflake Warehouse where all
                 dependencies are required to be retrieved from Snowflake Anaconda Channel.
+            target_platforms: List of target platforms to run the model. The only acceptable inputs are a combination of
+                {"WAREHOUSE", "SNOWPARK_CONTAINER_SERVICES"}. Defaults to None.
             python_version: Python version in which the model is run. Defaults to None.
             code_paths: List of directories containing code to import. Defaults to None.
             ext_modules: List of external modules to pickle with the model object.
@@ -287,6 +297,7 @@ class Registry:
             metrics=metrics,
             conda_dependencies=conda_dependencies,
             pip_requirements=pip_requirements,
+            target_platforms=target_platforms,
             python_version=python_version,
             signatures=signatures,
             sample_input_data=sample_input_data,
@@ -374,34 +385,25 @@ class Registry:
     def add_monitor(
         self,
         name: str,
-        table_config: model_monitor_config.ModelMonitorTableConfig,
+        source_config: model_monitor_config.ModelMonitorSourceConfig,
         model_monitor_config: model_monitor_config.ModelMonitorConfig,
-        *,
-        add_dashboard_udtfs: bool = False,
     ) -> model_monitor.ModelMonitor:
         """Add a Model Monitor to the Registry
 
         Args:
             name: Name of Model Monitor to create
-            table_config: Configuration options of table for ModelMonitor.
+            source_config: Configuration options of table for ModelMonitor.
             model_monitor_config: Configuration options of ModelMonitor.
-            add_dashboard_udtfs: Add UDTFs useful for creating a dashboard.
 
         Returns:
             The newly added ModelMonitor object.
 
         Raises:
-            ValueError: If monitoring feature flag is not enabled.
+            ValueError: If monitoring is not enabled in the Registry.
         """
         if not self.enable_monitoring:
-            raise ValueError(
-                "Must enable monitoring in Registry to use this method. Please set the `enable_monitoring=True` option"
-            )
-
-        # TODO: Change to fully qualified source table reference to allow table to live in different DB.
-        return self._model_monitor_manager.add_monitor(
-            name, table_config, model_monitor_config, add_dashboard_udtfs=add_dashboard_udtfs
-        )
+            raise ValueError(_MODEL_MONITORING_DISABLED_ERROR)
+        return self._model_monitor_manager.add_monitor(name, source_config, model_monitor_config)
 
     @overload
     def get_monitor(self, model_version: model_version_impl.ModelVersion) -> model_monitor.ModelMonitor:
@@ -439,17 +441,14 @@ class Registry:
             The fetched ModelMonitor.
 
         Raises:
-            ValueError: If monitoring feature flag is not enabled.
-            ValueError: If neither name nor model_version specified.
+            ValueError: If monitoring is not enabled in the Registry.
         """
         if not self.enable_monitoring:
-            raise ValueError(
-                "Must enable monitoring in Registry to use this method. Please set the `enable_monitoring=True` option"
-            )
+            raise ValueError(_MODEL_MONITORING_DISABLED_ERROR)
         if name is not None:
             return self._model_monitor_manager.get_monitor(name=name)
         elif model_version is not None:
-            return self._model_monitor_manager.get_monitor_by_model_version(model_version=model_version)
+            return self._model_monitor_manager.get_monitor_by_model_version(model_version)
         else:
             raise ValueError("Must provide either `name` or `model_version` to get ModelMonitor")
 
@@ -465,12 +464,10 @@ class Registry:
             List of snowpark.Row containing metadata for each model monitor.
 
         Raises:
-            ValueError: If monitoring feature flag is not enabled.
+            ValueError: If monitoring is not enabled in the Registry.
         """
         if not self.enable_monitoring:
-            raise ValueError(
-                "Must enable monitoring in Registry to use this method. Please set the `enable_monitoring=True` option"
-            )
+            raise ValueError(_MODEL_MONITORING_DISABLED_ERROR)
         return self._model_monitor_manager.show_model_monitors()
 
     @telemetry.send_api_usage_telemetry(
@@ -485,10 +482,8 @@ class Registry:
             name: Name of the Model Monitor to delete.
 
         Raises:
-            ValueError: If monitoring feature flag is not enabled.
+            ValueError: If monitoring is not enabled in the registry.
         """
         if not self.enable_monitoring:
-            raise ValueError(
-                "Must enable monitoring in Registry to use this method. Please set the `enable_monitoring=True` option"
-            )
+            raise ValueError(_MODEL_MONITORING_DISABLED_ERROR)
         self._model_monitor_manager.delete_monitor(name)

snowflake/ml/version.py

@@ -1 +1 @@
-VERSION="1.6.4"
+VERSION="1.7.1"

{snowflake_ml_python-1.6.4.dist-info → snowflake_ml_python-1.7.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: snowflake-ml-python
-Version: 1.6.4
+Version: 1.7.1
 Summary: The machine learning client library that is used for interacting with Snowflake to build machine learning solutions.
 Author-email: "Snowflake, Inc" <support@snowflake.com>
 License: 
@@ -220,7 +220,6 @@ Classifier: Intended Audience :: Information Technology
 Classifier: Intended Audience :: System Administrators
 Classifier: License :: OSI Approved :: Apache Software License
 Classifier: Operating System :: OS Independent
-Classifier: Programming Language :: Python :: 3.8
 Classifier: Programming Language :: Python :: 3.9
 Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
@@ -230,38 +229,39 @@ Classifier: Topic :: Software Development :: Libraries
 Classifier: Topic :: Software Development :: Libraries :: Application Frameworks
 Classifier: Topic :: Software Development :: Libraries :: Python Modules
 Classifier: Topic :: Scientific/Engineering :: Information Analysis
-Requires-Python: <3.12,>=3.8
+Requires-Python: <3.12,>=3.9
 Description-Content-Type: text/markdown
 License-File: LICENSE.txt
 Requires-Dist: absl-py <2,>=0.15
 Requires-Dist: anyio <4,>=3.5.0
 Requires-Dist: cachetools <6,>=3.1.1
 Requires-Dist: cloudpickle >=2.0.0
+Requires-Dist: cryptography
 Requires-Dist: fsspec[http] <2024,>=2022.11
 Requires-Dist: importlib-resources <7,>=6.1.1
 Requires-Dist: numpy <2,>=1.23
-Requires-Dist: packaging <24,>=20.9
+Requires-Dist: packaging <25,>=20.9
 Requires-Dist: pandas <3,>=1.0.0
 Requires-Dist: pyarrow
 Requires-Dist: pytimeparse <2,>=1.1.8
 Requires-Dist: pyyaml <7,>=6.0
 Requires-Dist: retrying <2,>=1.3.3
 Requires-Dist: s3fs <2024,>=2022.11
-Requires-Dist: scikit-learn <1.6,>=1.2.1
+Requires-Dist: scikit-learn <1.6,>=1.4
 Requires-Dist: scipy <2,>=1.9
 Requires-Dist: snowflake-connector-python[pandas] <4,>=3.5.0
 Requires-Dist: snowflake-snowpark-python <2,>=1.17.0
 Requires-Dist: sqlparse <1,>=0.4
 Requires-Dist: typing-extensions <5,>=4.1.0
-Requires-Dist: xgboost <2.1,>=1.7.3
+Requires-Dist: xgboost <3,>=1.7.3
 Provides-Extra: all
 Requires-Dist: catboost <2,>=1.2.0 ; extra == 'all'
-Requires-Dist: lightgbm <5,>=3.3.5 ; extra == 'all'
+Requires-Dist: lightgbm <5,>=4.1.0 ; extra == 'all'
 Requires-Dist: mlflow <2.4,>=2.1.0 ; extra == 'all'
 Requires-Dist: peft <1,>=0.5.0 ; extra == 'all'
 Requires-Dist: sentence-transformers <3,>=2.2.2 ; extra == 'all'
 Requires-Dist: sentencepiece <1,>=0.1.95 ; extra == 'all'
-Requires-Dist: shap <1,>=0.42.0 ; extra == 'all'
+Requires-Dist: shap <1,>=0.46.0 ; extra == 'all'
 Requires-Dist: tensorflow <3,>=2.10 ; extra == 'all'
 Requires-Dist: tokenizers <1,>=0.10 ; extra == 'all'
 Requires-Dist: torch <2.3.0,>=2.0.1 ; extra == 'all'
@@ -270,13 +270,13 @@ Requires-Dist: transformers <5,>=4.32.1 ; extra == 'all'
 Provides-Extra: catboost
 Requires-Dist: catboost <2,>=1.2.0 ; extra == 'catboost'
 Provides-Extra: lightgbm
-Requires-Dist: lightgbm <5,>=3.3.5 ; extra == 'lightgbm'
+Requires-Dist: lightgbm <5,>=4.1.0 ; extra == 'lightgbm'
 Provides-Extra: llm
 Requires-Dist: peft <1,>=0.5.0 ; extra == 'llm'
 Provides-Extra: mlflow
 Requires-Dist: mlflow <2.4,>=2.1.0 ; extra == 'mlflow'
 Provides-Extra: shap
-Requires-Dist: shap <1,>=0.42.0 ; extra == 'shap'
+Requires-Dist: shap <1,>=0.46.0 ; extra == 'shap'
 Provides-Extra: tensorflow
 Requires-Dist: tensorflow <3,>=2.10 ; extra == 'tensorflow'
 Provides-Extra: torch
@@ -336,7 +336,7 @@ If you don't have a Snowflake account yet, you can [sign up for a 30-day free tr
 Follow the [installation instructions](https://docs.snowflake.com/en/developer-guide/snowpark-ml/index#installing-snowpark-ml)
 in the Snowflake documentation.
 
-Python versions 3.8 to 3.11 are supported. You can use [miniconda](https://docs.conda.io/en/latest/miniconda.html) or
+Python versions 3.9 to 3.11 are supported. You can use [miniconda](https://docs.conda.io/en/latest/miniconda.html) or
 [anaconda](https://www.anaconda.com/) to create a Conda environment (recommended),
 or [virtualenv](https://docs.python.org/3/tutorial/venv.html) to create a virtual environment.
 
@@ -373,7 +373,96 @@ be compatibility issues. Server-side functionality that `snowflake-ml-python` de
 
 # Release History
 
-## 1.6.4
+## 1.7.1
+
+### Bug Fixes
+
+- Registry: Null value is now allowed in the dataframe used in model signature inference. Null values will be ignored
+ and others will be used to infer the signature.
+- Registry: Pandas Extension DTypes (`pandas.StringDType()`, `pandas.BooleanDType()`, etc.) are now supported in model
+signature inference.
+- Registry: Null value is now allowed in the dataframe used to predict.
+- Data: Fix missing `snowflake.ml.data.*` module exports in wheel
+- Dataset: Fix missing `snowflake.ml.dataset.*` module exports in wheel.
+- Registry: Fix the issue that `tf_keras.Model` is not recognized as keras model when logging.
+
+### Behavior Changes
+
+### New Features
+
+- Registry: Option to `enable_monitoring` set to False by default.  This will gate access to preview features of Model Monitoring.
+- Model Monitoring: `show_model_monitors` Registry method.  This feature is still in Private Preview.
+- Registry: Support `pd.Series` in input and output data.
+- Model Monitoring: `add_monitor` Registry method.  This feature is still in Private Preview.
+- Model Monitoring: `resume` and `suspend` ModelMonitor.  This feature is still in Private Preview.
+- Model Monitoring: `get_monitor` Registry method.  This feature is still in Private Preview.
+- Model Monitoring: `delete_monitor` Registry method.  This feature is still in Private Preview.
+
+## 1.7.0 (10-22-2024)
+
+### Behavior Change
+
+- Generic: Require python >= 3.9.
+- Data Connector: Update `to_torch_dataset` and `to_torch_datapipe` to add a dimension for scalar data.
+This allows for more seamless integration with PyTorch `DataLoader`, which creates batches by stacking inputs of each batch.
+
+Examples:
+
+```python
+ds = connector.to_torch_dataset(shuffle=False, batch_size=3)
+```
+
+- Input: "col1": [10, 11, 12]
+  - Previous batch: array([10., 11., 12.]) with shape (3,)
+  - New batch: array([[10.], [11.], [12.]]) with shape (3, 1)
+
+- Input: "col2": [[0, 100], [1, 110], [2, 200]]
+  - Previous batch: array([[  0, 100], [  1, 110], [  2, 200]]) with shape (3,2)
+  - New batch: No change
+
+- Model Registry: External access integrations are optional when creating a model inference service in
+  Snowflake >= 8.40.0.
+- Model Registry: Deprecate `build_external_access_integration` with `build_external_access_integrations` in
+  `ModelVersion.create_service()`.
+
+### Bug Fixes
+
+- Registry: Updated `log_model` API to accept both signature and sample_input_data parameters.
+- Feature Store: ExampleHelper uses fully qualified path for table name. change weather features aggregation from 1d to 1h.
+- Data Connector: Return numpy array with appropriate object type instead of list for multi-dimensional
+data from `to_torch_dataset` and `to_torch_datapipe`
+- Model explainability: Incompatibility between SHAP 0.42.1 and XGB 2.1.1 resolved by using latest SHAP 0.46.0.
+
+### New Features
+
+- Registry: Provide pass keyworded variable length of arguments to class ModelContext. Example usage:
+
+```python
+mc = custom_model.ModelContext(
+    config = 'local_model_dir/config.json',
+    m1 = model1
+)
+
+class ExamplePipelineModel(custom_model.CustomModel):
+    def __init__(self, context: custom_model.ModelContext) -> None:
+      super().__init__(context)
+      v = open(self.context['config']).read()
+      self.bias = json.loads(v)['bias']
+
+    @custom_model.inference_api
+    def predict(self, input: pd.DataFrame) -> pd.DataFrame:
+      model_output = self.context['m1'].predict(input)
+      return pd.DataFrame({'output': model_output + self.bias})
+```
+
+- Model Development: Upgrade scikit-learn in UDTF backend for log_loss metric. As a result, `eps` argument is now ignored.
+- Data Connector: Add the option of passing a `None` sized batch to `to_torch_dataset` for better
+interoperability with PyTorch DataLoader.
+- Model Registry: Support [pandas.CategoricalDtype](https://pandas.pydata.org/docs/reference/api/pandas.CategoricalDtype.html#pandas-categoricaldtype)
+- Registry: It is now possible to pass `signatures` and `sample_input_data` at the same time to capture background
+data from explainablity and data lineage.
+
+## 1.6.4 (2024-10-17)
 
 ### Bug Fixes
 
@@ -391,6 +480,9 @@ be compatibility issues. Server-side functionality that `snowflake-ml-python` de
 - Registry: Fix a bug that `ModelVersion.run` is called in a nested way.
 - Registry: Fix an issue that leads to `log_model` failure when local package version contains parts other than
   base version.
+- Fix issue where `sample_weights` were not being applied to search estimators.
+- Model explainability: Fix bug which creates explain as a function instead of table function when enabling by default.
+- Model explainability: Update lightgbm binary classification to return non-json values, from customer feedback.
 
 ### New Features