snowflake-ml-python 1.14.0__py3-none-any.whl → 1.15.0__py3-none-any.whl

snowflake/ml/model/_packager/model_handlers/_utils.py

@@ -1,5 +1,6 @@
 import importlib
 import json
+import logging
 import os
 import pathlib
 import warnings
@@ -8,7 +9,6 @@ from typing import Any, Callable, Iterable, Optional, Sequence, cast
 import numpy as np
 import numpy.typing as npt
 import pandas as pd
-from absl import logging
 
 import snowflake.snowpark.dataframe as sp_df
 from snowflake.ml._internal import env
@@ -23,6 +23,8 @@ from snowflake.ml.model._signatures import (
 )
 from snowflake.snowpark import DataFrame as SnowparkDataFrame
 
+logger = logging.getLogger(__name__)
+
 EXPLAIN_BACKGROUND_DATA_ROWS_COUNT_LIMIT = 1000
 
 
@@ -257,7 +259,7 @@ def validate_model_task(passed_model_task: model_types.Task, inferred_model_task
             )
         return inferred_model_task
     elif inferred_model_task != model_types.Task.UNKNOWN:
-        logging.info(f"Inferred Task: {inferred_model_task.name} is used as task for this model " f"version")
+        logger.info(f"Inferred Task: {inferred_model_task.name} is used as task for this model " f"version")
         return inferred_model_task
     return passed_model_task
 

snowflake/ml/model/_packager/model_handlers/huggingface_pipeline.py

@@ -43,7 +43,6 @@ DEFAULT_CHAT_TEMPLATE = "{% for message in messages %}{{'<|im_start|>' + message
 def get_requirements_from_task(task: str, spcs_only: bool = False) -> list[model_env.ModelDependency]:
     # Text
     if task in [
-        "conversational",
         "fill-mask",
         "ner",
         "token-classification",
@@ -521,6 +520,7 @@ class HuggingFacePipelineHandler(
                                 input_data = X[signature.inputs[0].name].to_list()
                             temp_res = getattr(raw_model, target_method)(input_data)
                     else:
+                        # TODO: remove conversational pipeline code
                         # For others, we could offer the whole dataframe as a list.
                         # Some of them may need some conversion
                         if hasattr(transformers, "ConversationalPipeline") and isinstance(
@@ -759,11 +759,13 @@ class HuggingFaceOpenAICompatibleModel:
             eos_token_id=self.tokenizer.eos_token_id,
             stop_strings=stop_strings,
             stream=stream,
-            repetition_penalty=frequency_penalty,
-            diversity_penalty=presence_penalty if n > 1 else None,
             num_return_sequences=n,
-            num_beams=max(2, n),  # must be >1
-            num_beam_groups=max(2, n) if presence_penalty else 1,
+            num_beams=max(1, n),  # must be >1
+            repetition_penalty=frequency_penalty,
+            # TODO: Handle diversity_penalty and num_beam_groups
+            # not all models support them making it hard to support any huggingface model
+            # diversity_penalty=presence_penalty if n > 1 else None,
+            # num_beam_groups=max(2, n) if presence_penalty else 1,
             do_sample=False,
         )
 

snowflake/ml/model/_packager/model_packager.py

@@ -1,9 +1,8 @@
+import logging
 import os
 from types import ModuleType
 from typing import Optional
 
-from absl import logging
-
 from snowflake.ml._internal.exceptions import (
     error_codes,
     exceptions as snowml_exceptions,
@@ -12,6 +11,8 @@ from snowflake.ml.model import custom_model, model_signature, type_hints as mode
 from snowflake.ml.model._packager import model_handler
 from snowflake.ml.model._packager.model_meta import model_meta
 
+logger = logging.getLogger(__name__)
+
 
 class ModelPackager:
     """Top-level class to save/load and manage a Snowflake Native formatted model.
@@ -96,7 +97,7 @@ class ModelPackager:
                 **options,
             )
             if signatures is None:
-                logging.info(f"Model signatures are auto inferred as:\n\n{meta.signatures}")
+                logger.info(f"Model signatures are auto inferred as:\n\n{meta.signatures}")
 
         self.model = model
         self.meta = meta

snowflake/ml/model/_packager/model_runtime/_snowml_inference_alternative_requirements.py

@@ -2,7 +2,6 @@
 # Generate by running 'bazel run --config=pre_build //bazel/requirements:sync_requirements'
 
 REQUIREMENTS = [
-    "absl-py>=0.15,<2",
     "aiohttp!=4.0.0a0, !=4.0.0a1",
     "anyio>=3.5.0,<5",
     "cachetools>=3.1.1,<6",

snowflake/ml/model/_signatures/utils.py

@@ -110,27 +110,6 @@ def huggingface_pipeline_signature_auto_infer(
 ) -> Optional[core.ModelSignature]:
     # Text
 
-    # https://huggingface.co/docs/transformers/en/main_classes/pipelines#transformers.ConversationalPipeline
-    # Needs to convert to conversation object.
-    if task == "conversational":
-        warnings.warn(
-            (
-                "Conversational pipeline is removed from transformers since 4.42.0. "
-                "Support will be removed from snowflake-ml-python soon."
-            ),
-            category=DeprecationWarning,
-            stacklevel=1,
-        )
-        return core.ModelSignature(
-            inputs=[
-                core.FeatureSpec(name="user_inputs", dtype=core.DataType.STRING, shape=(-1,)),
-                core.FeatureSpec(name="generated_responses", dtype=core.DataType.STRING, shape=(-1,)),
-            ],
-            outputs=[
-                core.FeatureSpec(name="generated_responses", dtype=core.DataType.STRING, shape=(-1,)),
-            ],
-        )
-
     # https://huggingface.co/docs/transformers/en/main_classes/pipelines#transformers.TokenClassificationPipeline
     if task == "fill-mask":
         return core.ModelSignature(

snowflake/ml/model/models/huggingface_pipeline.py

@@ -8,6 +8,7 @@ from snowflake import snowpark
 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.model._client.model import inference_engine_utils
 from snowflake.ml.model._client.ops import service_ops
 from snowflake.snowpark import async_job, session
 
@@ -77,6 +78,15 @@ class HuggingFacePipelineModel:
         framework = kwargs.get("framework", None)
         feature_extractor = kwargs.get("feature_extractor", None)
 
+        _can_download_snapshot = False
+        if download_snapshot:
+            try:
+                import huggingface_hub as hf_hub
+
+                _can_download_snapshot = True
+            except ImportError:
+                pass
+
         # ==== Start pipeline logic from transformers ====
         if model_kwargs is None:
             model_kwargs = {}
@@ -141,22 +151,23 @@ class HuggingFacePipelineModel:
         # Instantiate config if needed
         config_obj = None
 
-        if isinstance(config, str):
-            config_obj = transformers.AutoConfig.from_pretrained(
-                config, _from_pipeline=task, **hub_kwargs, **model_kwargs
-            )
-            hub_kwargs["_commit_hash"] = config_obj._commit_hash
-        elif config is None and isinstance(model, str):
-            config_obj = transformers.AutoConfig.from_pretrained(
-                model, _from_pipeline=task, **hub_kwargs, **model_kwargs
-            )
-            hub_kwargs["_commit_hash"] = config_obj._commit_hash
-        # We only support string as config argument.
-        elif config is not None and not isinstance(config, str):
-            raise RuntimeError(
-                "Impossible to use non-string config as input for HuggingFacePipelineModel. Use transformers.Pipeline"
-                " object if required."
-            )
+        if not _can_download_snapshot:
+            if isinstance(config, str):
+                config_obj = transformers.AutoConfig.from_pretrained(
+                    config, _from_pipeline=task, **hub_kwargs, **model_kwargs
+                )
+                hub_kwargs["_commit_hash"] = config_obj._commit_hash
+            elif config is None and isinstance(model, str):
+                config_obj = transformers.AutoConfig.from_pretrained(
+                    model, _from_pipeline=task, **hub_kwargs, **model_kwargs
+                )
+                hub_kwargs["_commit_hash"] = config_obj._commit_hash
+            # We only support string as config argument.
+            elif config is not None and not isinstance(config, str):
+                raise RuntimeError(
+                    "Impossible to use non-string config as input for HuggingFacePipelineModel. "
+                    "Use transformers.Pipeline object if required."
+                )
 
         # ==== Start pipeline logic (Task) from transformers ====
 
@@ -208,7 +219,7 @@ class HuggingFacePipelineModel:
                 "Using a pipeline without specifying a model name and revision in production is not recommended.",
                 stacklevel=2,
             )
-            if config is None and isinstance(model, str):
+            if not _can_download_snapshot and config is None and isinstance(model, str):
                 config_obj = transformers.AutoConfig.from_pretrained(
                     model, _from_pipeline=task, **hub_kwargs, **model_kwargs
                 )
@@ -228,11 +239,10 @@ class HuggingFacePipelineModel:
                 )
 
         repo_snapshot_dir: Optional[str] = None
-        if download_snapshot:
+        if _can_download_snapshot:
             try:
-                from huggingface_hub import snapshot_download
 
-                repo_snapshot_dir = snapshot_download(
+                repo_snapshot_dir = hf_hub.snapshot_download(
                     repo_id=model,
                     revision=revision,
                     token=token,
@@ -268,7 +278,7 @@ class HuggingFacePipelineModel:
         ],
     )
     @snowpark._internal.utils.private_preview(version="1.9.1")
-    def create_service(
+    def log_model_and_create_service(
         self,
         *,
         session: session.Session,
@@ -293,6 +303,7 @@ class HuggingFacePipelineModel:
         force_rebuild: bool = False,
         build_external_access_integrations: Optional[list[str]] = None,
         block: bool = True,
+        experimental_options: Optional[dict[str, Any]] = None,
     ) -> Union[str, async_job.AsyncJob]:
         """Logs a Hugging Face model and creates a service in Snowflake.
 
@@ -319,6 +330,10 @@ class HuggingFacePipelineModel:
             force_rebuild: Whether to force rebuild the image. Defaults to False.
             build_external_access_integrations: External access integrations for building the image. Defaults to None.
             block: Whether to block the operation. Defaults to True.
+            experimental_options: Experimental options for the service creation with custom inference engine.
+                Currently, only `inference_engine` and `inference_engine_args_override` are supported.
+                `inference_engine` is the name of the inference engine to use.
+                `inference_engine_args_override` is a list of string arguments to pass to the inference engine.
 
         Raises:
             ValueError: if database and schema name is not provided and session doesn't have a
@@ -360,6 +375,24 @@ class HuggingFacePipelineModel:
         )
         logger.info(f"A service job is going to register the hf model as: {model_name}.{version_name}")
 
+        # Check if model is HuggingFace text-generation before doing inference engine checks
+        inference_engine_args = None
+        if experimental_options:
+            if self.task != "text-generation":
+                raise ValueError(
+                    "Currently, InferenceEngine using experimental_options is only supported for "
+                    "HuggingFace text-generation models."
+                )
+
+            inference_engine_args = inference_engine_utils._get_inference_engine_args(experimental_options)
+
+            # Enrich inference engine args if inference engine is specified
+            if inference_engine_args is not None:
+                inference_engine_args = inference_engine_utils._enrich_inference_engine_args(
+                    inference_engine_args,
+                    gpu_requests,
+                )
+
         from snowflake.ml.model import event_handler
         from snowflake.snowpark import exceptions
 
@@ -412,6 +445,8 @@ class HuggingFacePipelineModel:
                         # TODO: remove warehouse in the next release
                         warehouse=session.get_current_warehouse(),
                     ),
+                    # inference engine
+                    inference_engine_args=inference_engine_args,
                 )
                 status.update(label="HuggingFace model service created successfully", state="complete", expanded=False)
                 return result

snowflake/ml/registry/_manager/model_manager.py

@@ -1,8 +1,8 @@
+import logging
 from types import ModuleType
 from typing import TYPE_CHECKING, Any, Optional, Union
 
 import pandas as pd
-from absl.logging import logging
 
 from snowflake.ml._internal import platform_capabilities, telemetry
 from snowflake.ml._internal.exceptions import error_codes, exceptions

snowflake/ml/registry/_manager/model_parameter_reconciler.py

@@ -1,8 +1,8 @@
+import logging
 import warnings
 from dataclasses import dataclass
 from typing import Any, Optional
 
-from absl.logging import logging
 from packaging import requirements
 
 from snowflake.ml import version as snowml_version
@@ -221,7 +221,7 @@ class ModelParameterReconciler:
             ).get(env_utils.SNOWPARK_ML_PKG_NAME, [])
 
             if len(snowml_matched_versions) < 1 and not options.get("embed_local_ml_library", False):
-                logging.info(
+                logger.info(
                     f"Local snowflake-ml-python library has version {snowml_version.VERSION},"
                     " which is not available in the Snowflake server, embedding local ML library automatically."
                 )

snowflake/ml/utils/connection_params.py

@@ -1,13 +1,15 @@
 import configparser
+import logging
 import os
 from typing import Optional, Union
 
-from absl import logging
 from cryptography.hazmat import backends
 from cryptography.hazmat.primitives import serialization
 
 from snowflake import snowpark
 
+logger = logging.getLogger(__name__)
+
 _DEFAULT_CONNECTION_FILE = "~/.snowsql/config"
 
 
@@ -108,7 +110,7 @@ def _load_from_snowsql_config_file(connection_name: str, login_file: str = "") -
     """Loads the dictionary from snowsql config file."""
     snowsql_config_file = login_file if login_file else os.path.expanduser(_DEFAULT_CONNECTION_FILE)
     if not os.path.exists(snowsql_config_file):
-        logging.error(f"Connection name given but snowsql config file is not found at: {snowsql_config_file}")
+        logger.error(f"Connection name given but snowsql config file is not found at: {snowsql_config_file}")
         raise Exception("Snowflake SnowSQL config not found.")
 
     config = configparser.ConfigParser(inline_comment_prefixes="#")
@@ -124,7 +126,7 @@ def _load_from_snowsql_config_file(connection_name: str, login_file: str = "") -
         # See https://docs.snowflake.com/en/user-guide/snowsql-start.html#configuring-default-connection-settings
         connection_name = "connections"
 
-    logging.info(f"Reading {snowsql_config_file} for connection parameters defined as {connection_name}")
+    logger.info(f"Reading {snowsql_config_file} for connection parameters defined as {connection_name}")
     config.read(snowsql_config_file)
     conn_params = dict(config[connection_name])
     # Remap names to appropriate args in Python Connector API

snowflake/ml/version.py

@@ -1,2 +1,2 @@
 # This is parsed by regex in conda recipe meta file. Make sure not to break it.
-VERSION = "1.14.0"
+VERSION = "1.15.0"

{snowflake_ml_python-1.14.0.dist-info → snowflake_ml_python-1.15.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: snowflake-ml-python
-Version: 1.14.0
+Version: 1.15.0
 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: 
@@ -233,7 +233,6 @@ Classifier: Topic :: Scientific/Engineering :: Information Analysis
 Requires-Python: <3.13,>=3.9
 Description-Content-Type: text/markdown
 License-File: LICENSE.txt
-Requires-Dist: absl-py<2,>=0.15
 Requires-Dist: anyio<5,>=3.5.0
 Requires-Dist: cachetools<6,>=3.1.1
 Requires-Dist: cloudpickle>=2.0.0
@@ -302,21 +301,21 @@ Requires-Dist: torch<3,>=2.0.1; extra == "transformers"
 Requires-Dist: transformers!=4.51.3,<5,>=4.39.3; extra == "transformers"
 Dynamic: license-file
 
-# Snowpark ML
+# Snowflake ML Python
 
-Snowpark ML is a set of tools including SDKs and underlying infrastructure to build and deploy machine learning models.
-With Snowpark ML, you can pre-process data, train, manage and deploy ML models all within Snowflake, using a single SDK,
+Snowflake ML Python is a set of tools including SDKs and underlying infrastructure to build and deploy machine learning models.
+With Snowflake ML Python, you can pre-process data, train, manage and deploy ML models all within Snowflake,
 and benefit from Snowflake’s proven performance, scalability, stability and governance at every stage of the Machine
 Learning workflow.
 
-## Key Components of Snowpark ML
+## Key Components of Snowflake ML Python
 
-The Snowpark ML Python SDK provides a number of APIs to support each stage of an end-to-end Machine Learning development
-and deployment process, and includes two key components.
+The Snowflake ML Python SDK provides a number of APIs to support each stage of an end-to-end Machine Learning development
+and deployment process.
 
-### Snowpark ML Development
+### Snowflake ML Model Development
 
-[Snowpark ML Development](https://docs.snowflake.com/en/developer-guide/snowpark-ml/index#ml-modeling)
+[Snowflake ML Model Development](https://docs.snowflake.com/developer-guide/snowflake-ml/overview#ml-modeling)
 provides a collection of python APIs enabling efficient ML model development directly in Snowflake:
 
 1. Modeling API (`snowflake.ml.modeling`) for data preprocessing, feature engineering and model training in Snowflake.
@@ -327,19 +326,16 @@ model development classes based on sklearn, xgboost, and lightgbm.
 1. Framework Connectors: Optimized, secure and performant data provisioning for Pytorch and Tensorflow frameworks in
 their native data loader formats.
 
-1. FileSet API: FileSet provides a Python fsspec-compliant API for materializing data into a Snowflake internal stage
-from a query or Snowpark Dataframe along with a number of convenience APIs.
+### Snowflake ML Ops
 
-### Snowflake MLOps
+Snowflake ML Python contains a suite of MLOps tools. It complements
+the Snowflake Modeling API, and provides end to end development to deployment within Snowflake.
+The Snowflake ML Ops suite consists of:
 
-Snowflake MLOps contains suit of tools and objects to make ML development cycle. It complements
-the Snowpark ML Development API, and provides end to end development to deployment within Snowflake.
-Currently, the API consists of:
-
-1. [Registry](https://docs.snowflake.com/en/developer-guide/snowpark-ml/index#snowflake-model-registry): A python API
+1. [Registry](https://docs.snowflake.com/developer-guide/snowflake-ml/overview#snowflake-model-registry): A python API
   allows secure deployment and management of models in Snowflake, supporting models trained both inside and outside of
   Snowflake.
-2. [Feature Store](https://docs.snowflake.com/en/developer-guide/snowpark-ml/index#snowflake-feature-store): A fully
+2. [Feature Store](https://docs.snowflake.com/developer-guide/snowflake-ml/overview#snowflake-feature-store): A fully
   integrated solution for defining, managing, storing and discovering ML features derived from your data. The
   Snowflake Feature Store supports automated, incremental refresh from batch and streaming data sources, so that
   feature pipelines need be defined only once to be continuously updated with new data.
@@ -348,12 +344,19 @@ Currently, the API consists of:
 
 ## Getting started
 
+Learn about all Snowflake ML feature offerings in the [Developer Guide](https://docs.snowflake.com/developer-guide/snowflake-ml/overview).
+
 ### Have your Snowflake account ready
 
 If you don't have a Snowflake account yet, you can [sign up for a 30-day free trial account](https://signup.snowflake.com/).
 
 ### Installation
 
+Snowflake ML Python is pre-installed in Container Runtime notebook environments.
+[Learn more](https://docs.snowflake.com/en/developer-guide/snowflake-ml/notebooks-on-spcs).
+
+In Snowflake Warehouse notebook environments, snowflake-ml-python can be installed using the "Packages" drop-down menu.
+
 Follow the [installation instructions](https://docs.snowflake.com/en/developer-guide/snowpark-ml/index#installing-snowpark-ml)
 in the Snowflake documentation.
 
@@ -363,8 +366,8 @@ or [virtualenv](https://docs.python.org/3/tutorial/venv.html) to create a virtua
 
 ### Conda channels
 
-The [Snowflake Conda Channel](https://repo.anaconda.com/pkgs/snowflake/) contains the official snowpark ML package releases.
-The recommended approach is to install `snowflake-ml-python` this conda channel:
+The [Snowflake Anaconda Channel](https://repo.anaconda.com/pkgs/snowflake/) contains the official snowflake-ml-python package
+releases. To install `snowflake-ml-python` from this conda channel:
 
 ```sh
 conda install \
@@ -373,25 +376,18 @@ conda install \
   snowflake-ml-python
 ```
 
-See [the developer guide](https://docs.snowflake.com/en/developer-guide/snowpark-ml/index) for installation instructions.
-
-The latest version of the `snowpark-ml-python` package is also published in a conda channel in this repository. Package versions
-in this channel may not yet be present in the official Snowflake conda channel.
+See [the developer guide](https://docs.snowflake.com/en/developer-guide/snowpark-ml/index) for detailed installation instructions.
 
-Install `snowflake-ml-python` from this channel with the following (being sure to replace `<version_specifier>` with the
-desired version, e.g. `1.0.10`):
+The snowflake-ml-python package is also published in [conda-forge](https://anaconda.org/conda-forge/snowflake-ml-python).
+To install `snowflake-ml-python` from conda forge:
 
-```bash
+```sh
 conda install \
-  -c https://raw.githubusercontent.com/snowflakedb/snowflake-ml-python/conda/releases/  \
-  -c https://repo.anaconda.com/pkgs/snowflake \
+  -c https://conda.anaconda.org/conda-forge/ \
   --override-channels \
-  snowflake-ml-python==<version_specifier>
+  snowflake-ml-python
 ```
 
-Note that until a `snowflake-ml-python` package version is available in the official Snowflake conda channel, there may
-be compatibility issues. Server-side functionality that `snowflake-ml-python` depends on may not yet be released.
-
 ### Verifying the package
 
 1. Install cosign.
@@ -410,7 +406,18 @@ NOTE: Version 1.7.0 is used as example here. Please choose the the latest versio
 
 # Release History
 
-## 1.14.0
+## 1.15.0
+
+### Bug Fixes
+
+### Behavior Changes
+
+* Registry: Dropping support for deprecated `conversational` task type for Huggingface models.
+  To read more <https://github.com/huggingface/transformers/pull/31165>
+
+### New Features
+
+## 1.14.0 (09-18-2025)
 
 ### Bug Fixes