workbench 0.8.177__tar.gz → 0.8.198__tar.gz

{workbench-0.8.177 → workbench-0.8.198}/.github/workflows/python-lint.yml

@@ -12,9 +12,9 @@ jobs:
     - uses: actions/checkout@v4
 
     - name: Set up Python
-      uses: actions/setup-python@v4
+      uses: actions/setup-python@v5
       with:
-        python-version: '3.x'
+        python-version: '3.13'
 
     - name: Install dependencies
       run: |
@@ -27,4 +27,3 @@ jobs:
     - name: Run Flake8 (Linting)
       run: |
         flake8 --exclude '*generated*' src/workbench applications tests
-

workbench-0.8.198/Dockerfile

@@ -0,0 +1,7 @@
+FROM python:3.12-slim
+
+ARG WORKBENCH_VERSION=
+RUN apt-get update && apt-get upgrade -y && \
+    apt-get install -y --no-install-recommends libexpat1 libnghttp2-dev && \
+    apt-get clean && rm -rf /var/lib/apt/lists/* && \
+    pip install --no-cache-dir workbench${WORKBENCH_VERSION:+==$WORKBENCH_VERSION}

{workbench-0.8.177 → workbench-0.8.198}/LICENSE

@@ -1,6 +1,6 @@
 MIT License
 
-Copyright (c) 2021-2024 SuperCowPowers LLC
+Copyright (c) 2021-2026 SuperCowPowers LLC
 
 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to deal

{workbench-0.8.177 → workbench-0.8.198}/PKG-INFO

@@ -1,15 +1,37 @@
 Metadata-Version: 2.4
 Name: workbench
-Version: 0.8.177
+Version: 0.8.198
 Summary: Workbench: A Dashboard and Python API for creating and deploying AWS SageMaker Model Pipelines
 Author-email: SuperCowPowers LLC <support@supercowpowers.com>
-License-Expression: MIT
+License: MIT License
+        
+        Copyright (c) 2021-2026 SuperCowPowers LLC
+        
+        Permission is hereby granted, free of charge, to any person obtaining a copy
+        of this software and associated documentation files (the "Software"), to deal
+        in the Software without restriction, including without limitation the rights
+        to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+        copies of the Software, and to permit persons to whom the Software is
+        furnished to do so, subject to the following conditions:
+        
+        The above copyright notice and this permission notice shall be included in all
+        copies or substantial portions of the Software.
+        
+        THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+        IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+        FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+        AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+        LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+        OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+        SOFTWARE.
+        
 Project-URL: Homepage, https://github.com/SuperCowPowers/workbench
 Keywords: SageMaker,Machine Learning,AWS,Python,Utilities
 Classifier: Development Status :: 4 - Beta
 Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
 Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
 Classifier: Topic :: Scientific/Engineering
 Requires-Python: >=3.10
 Description-Content-Type: text/markdown

{workbench-0.8.177 → workbench-0.8.198}/applications/aws_dashboard/Dockerfile

@@ -7,7 +7,7 @@ EXPOSE 8000
 ENV PYTHONHASHSEED=0
 
 # Define workbench version
-ARG WORKBENCH_VERSION=0.8.165
+ARG WORKBENCH_VERSION=0.8.196
 
 # Install system packages and gunicorn in one layer
 RUN apt-get update && apt-get install -y --no-install-recommends \

{workbench-0.8.177 → workbench-0.8.198}/aws_setup/workbench_compute/workbench_compute/workbench_compute_stack.py

@@ -9,6 +9,7 @@ from aws_cdk import (
     aws_sqs as sqs,
     aws_lambda as lambda_,
     aws_lambda_event_sources as lambda_events,
+    aws_logs as logs,
     Duration,
     Size,
 )
@@ -95,6 +96,7 @@ class WorkbenchComputeStack(Stack):
             compute_environment_name="workbench-compute-env",
             vpc=vpc,
             vpc_subnets=vpc_subnets,
+            maxv_cpus=16,  # Limit to 16 vCPU to minimize AWS Throttling issues
         )
 
     def create_batch_job_queue(self) -> batch.JobQueue:
@@ -118,7 +120,7 @@ class WorkbenchComputeStack(Stack):
                 "MLPipelineRepo",
                 repository_arn=f"arn:aws:ecr:{self.region}:507740646243:repository/aws-ml-images/py312-ml-pipelines",
             ),
-            tag="0.1",
+            tag="0.2",
         )
 
         # Job Definition Tiers
@@ -192,6 +194,7 @@ class WorkbenchComputeStack(Stack):
             handler="index.lambda_handler",
             code=lambda_.Code.from_inline(self._get_lambda_code()),
             timeout=Duration.minutes(5),
+            log_retention=logs.RetentionDays.ONE_WEEK,
             environment={
                 "WORKBENCH_BUCKET": self.workbench_bucket,
                 "JOB_QUEUE": self.batch_job_queue.job_queue_name,
@@ -205,7 +208,7 @@ class WorkbenchComputeStack(Stack):
         # Connect SQS to Lambda
         batch_trigger_lambda.add_event_source(
             lambda_events.SqsEventSource(
-                self.ml_pipeline_queue, batch_size=1, max_concurrency=5  # Limit parallel Batch job submissions
+                self.ml_pipeline_queue, batch_size=1  # One message at a time for precise error handling
             )
         )
         return batch_trigger_lambda
@@ -241,8 +244,6 @@ class WorkbenchComputeStack(Stack):
                         script_path = message['script_path']  # s3://bucket/path/to/script.py
                         size = message.get('size', 'small')
                         extra_env = message.get('environment', {})
-                        model_outputs = message.get('MODEL_OUTPUTS', [])
-                        endpoint_outputs = message.get('ENDPOINT_OUTPUTS', [])
 
                         script_name = Path(script_path).stem
                         job_name = f"workbench_{script_name}_{datetime.now():%Y%m%d_%H%M%S}"
@@ -257,14 +258,7 @@ class WorkbenchComputeStack(Stack):
                             *[{'name': k, 'value': v} for k, v in extra_env.items()]
                         ]
 
-                        # Add MODEL_OUTPUTS as comma-separated list if provided
-                        if model_outputs:
-                            env_vars.append({'name': 'MODEL_OUTPUTS', 'value': ','.join(model_outputs)})
-
-                        # Add ENDPOINT_OUTPUTS as comma-separated list if provided
-                        if endpoint_outputs:
-                            env_vars.append({'name': 'ENDPOINT_OUTPUTS', 'value': ','.join(endpoint_outputs)})
-
+                        # Submit Batch job to the queue
                         response = batch.submit_job(
                             jobName=job_name,
                             jobQueue=JOB_QUEUE,

{workbench-0.8.177 → workbench-0.8.198}/aws_setup/workbench_dashboard_full/app.py

@@ -12,7 +12,7 @@ print(f"Account: {aws_account}")
 print(f"Region: {aws_region}")
 
 # When you want a different docker image change this line
-dashboard_image = "public.ecr.aws/m6i5k1r2/workbench_dashboard:v0_8_165_amd64"
+dashboard_image = "public.ecr.aws/m6i5k1r2/workbench_dashboard:v0_8_196_amd64"
 
 # Workbench Configuration
 try:

{workbench-0.8.177 → workbench-0.8.198}/aws_setup/workbench_dashboard_lite/app.py

@@ -13,7 +13,7 @@ print(f"Account: {aws_account}")
 print(f"Region: {aws_region}")
 
 # When you want a different docker image change this line
-dashboard_image = "public.ecr.aws/m6i5k1r2/workbench_dashboard:v0_8_165_amd64"
+dashboard_image = "public.ecr.aws/m6i5k1r2/workbench_dashboard:v0_8_196_amd64"
 
 # Workbench Configuration
 try:

{workbench-0.8.177 → workbench-0.8.198}/docs/api_classes/endpoint.md

@@ -11,19 +11,15 @@
 **Run Inference on an Endpoint**
 
 ```py title="endpoint_inference.py"
-from workbench.api.feature_set import FeatureSet
-from workbench.api.model import Model
-from workbench.api.endpoint import Endpoint
+from workbench.api import Endpoint
+from workbench.utils.endpoint_utils import get_evaluation_data
 
 # Grab an existing Endpoint
 endpoint = Endpoint("abalone-regression-end")
 
 # Workbench has full ML Pipeline provenance, so we can backtrack the inputs,
 # get a DataFrame of data (not used for training) and run inference
-model = Model(endpoint.get_input())
-fs = FeatureSet(model.get_input())
-athena_table = fs.view("training").table
-df = fs.query(f"SELECT * FROM {athena_table} where training = FALSE")
+df = get_evaluation_data(endpoint)
 
 # Run inference/predictions on the Endpoint
 results_df = endpoint.inference(df)

{workbench-0.8.177 → workbench-0.8.198}/docs/api_classes/overview.md

@@ -17,10 +17,7 @@ These classes provide high-level APIs for the Workbench package, they enable you
 ## Example ML Pipline
 
 ```py title="full_ml_pipeline.py"
-from workbench.api.data_source import DataSource
-from workbench.api.feature_set import FeatureSet
-from workbench.api.model import Model, ModelType
-from workbench.api.endpoint import Endpoint
+from workbench.api import DataSource, FeatureSet, Model, ModelType, Endpoint
 
 # Create the abalone_data DataSource
 ds = DataSource("s3://workbench-public-data/common/abalone.csv")
@@ -46,9 +43,7 @@ model.to_endpoint(name="abalone-regression-end", tags=["abalone", "regression"])
 endpoint = Endpoint("abalone-regression-end")
 
 # Run inference on the Endpoint
-athena_table = fs.view("training").table
-df = fs.query(f"SELECT * FROM {athena_table} where training = FALSE")
-results = endpoint.predict(df)
+results = endpoint.auto_inference()
 print(results[["class_number_of_rings", "prediction"]])
 ```
 

{workbench-0.8.177 → workbench-0.8.198}/docs/api_classes/views.md

@@ -49,10 +49,10 @@ from workbench.api.feature_set import FeatureSet
 
 # Grab a FeatureSet View
 fs = FeatureSet("abalone_features")
-t_view = fs.view("training")
+d_view = fs.view("display")
 
 # Make some queries using the Athena backend
-df = t_view(f"select * from {t_view.table} where height > .3")
+df = d_view(f"select * from {d_view.table} where height > .3")
 print(df.head())
 
 df = t_view.query("select * from abalone_features where class_number_of_rings < 3")

{workbench-0.8.177 → workbench-0.8.198}/docs/core_classes/views/overview.md

@@ -57,10 +57,10 @@ from workbench.api.feature_set import FeatureSet
 
 # Grab a FeatureSet View
 fs = FeatureSet("abalone_features")
-t_view = fs.view("training")
+d_view = fs.view("display")
 
 # Make some queries using the Athena backend
-df = t_view(f"select * from {t_view.table} where height > .3")
+df = t_view(f"select * from {d_view.table} where height > .3")
 print(df.head())
 
 df = t_view.query("select * from abalone_features where class_number_of_rings < 3")

{workbench-0.8.177 → workbench-0.8.198}/examples/endpoint/endpoint_inference.py

@@ -1,16 +1,12 @@
-from workbench.api.feature_set import FeatureSet
-from workbench.api.model import Model
-from workbench.api.endpoint import Endpoint
+from workbench.api import Endpoint
+from workbench.utils.endpoint_utils import get_evaluation_data
 
 # Grab an existing Endpoint
 endpoint = Endpoint("abalone-regression")
 
 # Workbench has full ML Pipeline provenance, so we can backtrack the inputs,
 # get a DataFrame of data (not used for training) and run inference
-model = Model(endpoint.get_input())
-fs = FeatureSet(model.get_input())
-athena_table = fs.view("training").table
-df = fs.query(f"SELECT * FROM {athena_table} where training = FALSE")
+df = get_evaluation_data(endpoint)
 
 # Run inference/predictions on the Endpoint
 results_df = endpoint.inference(df)

{workbench-0.8.177 → workbench-0.8.198}/examples/full_ml_pipeline.py

@@ -44,7 +44,5 @@ if __name__ == "__main__":
     endpoint = Endpoint("abalone-regression")
 
     # Run inference on the Endpoint
-    athena_table = fs.view("training").table
-    df = fs.query(f"SELECT * FROM {athena_table} where training = FALSE")
-    results = endpoint.inference(df)
+    results = endpoint.auto_inference()
     print(results[["class_number_of_rings", "prediction"]])

{workbench-0.8.177 → workbench-0.8.198}/experiments/chemprop_model.py

@@ -427,12 +427,12 @@ if __name__ == "__main__":
                 target: label_names,
                 "precision": scores[0],
                 "recall": scores[1],
-                "fscore": scores[2],
+                "f1": scores[2],
                 "support": scores[3],
             }
         )
         # We need to get creative with the Classification Metrics
-        metrics = ["precision", "recall", "fscore", "support"]
+        metrics = ["precision", "recall", "f1", "support"]
         for t in label_names:
             for m in metrics:
                 value = score_df.loc[score_df[target] == t, m].iloc[0]

{workbench-0.8.177 → workbench-0.8.198}/experiments/pytorch_model.py

@@ -526,13 +526,13 @@ if __name__ == "__main__":
                 target: label_names,
                 "precision": scores[0],
                 "recall": scores[1],
-                "fscore": scores[2],
+                "f1": scores[2],
                 "support": scores[3],
             }
         )
 
         # We need to get creative with the Classification Metrics
-        metrics = ["precision", "recall", "fscore", "support"]
+        metrics = ["precision", "recall", "f1", "support"]
         for t in label_names:
             for m in metrics:
                 value = score_df.loc[score_df[target] == t, m].iloc[0]

workbench-0.8.198/ml_pipelines/hyper_uq.py

@@ -0,0 +1,50 @@
+import os
+from pprint import pprint
+from workbench.api.feature_set import FeatureSet
+from workbench.api.model import Model, ModelType
+
+hyperparameters = {
+    "n_estimators": 200,  # Number of trees
+    "max_depth": 6,  # Maximum tree depth
+    "learning_rate": 0.05,  # Boosting learning rate
+    "subsample": 0.7,  # Row sampling ratio
+    "colsample_bytree": 0.3,  # Feature sampling per tree
+    "colsample_bylevel": 0.5,  # Feature sampling per level
+    "min_child_weight": 5,  # Minimum samples in leaf
+    "gamma": 0.2,  # Minimum loss reduction for split
+    "reg_alpha": 0.5,  # L1 regularization
+    "reg_lambda": 2.0,  # L2 regularization
+    "scale_pos_weight": 1,  # Class weight balance
+}
+
+# Grab a FeatureSet
+my_features = FeatureSet("aqsol_features")
+
+# Grab our AQSol Regression Model
+model = Model("aqsol-regression")
+
+# Get Feature and Target Columns from the existing given Model
+features = model.features()
+target = model.target()
+
+# Create the UQ Model from our FeatureSet
+fs = FeatureSet(model.get_input())
+my_model = fs.to_model(
+    name="aqsol-uq-hyper",
+    model_type=ModelType.UQ_REGRESSOR,
+    feature_list=features,
+    target_column=target,
+    description="Mapie + XGB Model",
+    tags=["aqsol", "mapie", "regression"],
+    hyperparameters=hyperparameters,
+)
+# Print the details of the created model
+pprint(my_model.details())
+my_model = Model("aqsol-uq-hyper")
+
+# Deploy an Endpoint for the Model
+endpoint = my_model.to_endpoint()
+
+# Run auto-inference on the Endpoint
+endpoint.auto_inference(capture=True)
+endpoint.cross_fold_inference()

workbench-0.8.198/model_docker_images/inference/requirements.txt

@@ -0,0 +1,9 @@
+fastapi==0.121.0       # FastAPI for serving the model
+uvicorn==0.34.0        # ASGI server for running FastAPI
+scikit-learn==1.7.2
+xgboost==3.1.1
+lightgbm==4.6.0
+ngboost==0.5.6
+pandas==2.3.3
+awswrangler==3.14.0
+joblib==1.5.2

{workbench-0.8.177 → workbench-0.8.198}/model_docker_images/ml_pipelines/Dockerfile

@@ -29,10 +29,10 @@ COPY --from=builder /usr/local/lib/python3.12/site-packages /usr/local/lib/pytho
 COPY --from=builder /usr/local/bin /usr/local/bin
 
 # Install workbench only
-RUN pip install --no-cache-dir workbench==0.8.167
+RUN pip install --no-cache-dir workbench==0.8.196
 
 # Add the script runner
 COPY ml_pipeline_runner.py /app/ml_pipeline_runner.py
 WORKDIR /app
 
-ENTRYPOINT ["python", "/app/ml_pipeline_runner.py"]
+ENTRYPOINT ["python", "/app/ml_pipeline_runner.py"]

workbench-0.8.198/model_docker_images/ml_pipelines/requirements.txt

@@ -0,0 +1,16 @@
+# This file is used to install the required packages for the ML Pipeline Docker image.
+--index-url https://download.pytorch.org/whl/cpu
+--extra-index-url https://pypi.org/simple/
+pandas==2.3.3
+awswrangler==3.14.0
+torch==2.9.1
+pytorch-tabular==1.1.1
+requests==2.32.5
+joblib==1.5.2
+boto3==1.40.72
+botocore==1.40.72
+redis==7.0.1
+scikit-learn==1.7.2
+xgboost==3.1.1
+rdkit==2025.9.1
+mordredcommunity==2.0.6

{workbench-0.8.177 → workbench-0.8.198}/model_docker_images/tests/example_model_script.py

@@ -244,13 +244,13 @@ if __name__ == "__main__":
                 target: label_names,
                 "precision": scores[0],
                 "recall": scores[1],
-                "fscore": scores[2],
+                "f1": scores[2],
                 "support": scores[3],
             }
         )
 
         # We need to get creative with the Classification Metrics
-        metrics = ["precision", "recall", "fscore", "support"]
+        metrics = ["precision", "recall", "f1", "support"]
         for t in label_names:
             for m in metrics:
                 value = score_df.loc[score_df[target] == t, m].iloc[0]

workbench-0.8.198/model_docker_images/training/requirements.txt

@@ -0,0 +1,7 @@
+scikit-learn==1.7.2
+xgboost==3.1.1
+lightgbm==4.6.0
+ngboost==0.5.6
+pandas==2.3.3
+awswrangler==3.14.0
+joblib==1.5.2

{workbench-0.8.177 → workbench-0.8.198}/notebooks/Inference_On_Legacy_Models.ipynb

@@ -63,8 +63,8 @@
     "# Grab the evaluation data\n",
     "from workbench.api.feature_set import FeatureSet\n",
     "fs = FeatureSet(\"solubility_featurized_fs\")\n",
-    "training_view = fs.view(\"training\").table\n",
-    "eval_data = fs.query(f\"select * from {training_view} where training=0\")"
+    "table = model.training_view().table\n",
+    "eval_data = fs.query(f\"select * from {table} where training=0\")"
    ]
   },
   {

{workbench-0.8.177 → workbench-0.8.198}/notebooks/ML_Pipeline_with_Workbench_2.ipynb

@@ -316,7 +316,7 @@
    "id": "a12b00ea",
    "metadata": {},
    "source": [
-    "table = fs.view(\"training\").table\n",
+    "table = model.training_view().table\n",
     "test_df = fs.query(f\"select * from {table} where training=0\")\n",
     "test_df.head()"
    ],

{workbench-0.8.177 → workbench-0.8.198}/pyproject.toml

@@ -11,7 +11,7 @@ name = "workbench"
 dynamic = ["version"]
 description = "Workbench: A Dashboard and Python API for creating and deploying AWS SageMaker Model Pipelines"
 readme = "README.md"
-license = "MIT"
+license = {file = "LICENSE"}
 requires-python = ">=3.10"
 keywords = ["SageMaker", "Machine Learning", "AWS", "Python", "Utilities"]
 classifiers = [
@@ -19,6 +19,7 @@ classifiers = [
     "Programming Language :: Python :: 3.10",
     "Programming Language :: Python :: 3.11",
     "Programming Language :: Python :: 3.12",
+    "Programming Language :: Python :: 3.13",
     "Topic :: Scientific/Engineering",
 ]
 authors = [{ name = "SuperCowPowers LLC", email = "support@supercowpowers.com" }]
@@ -92,6 +93,7 @@ workbench_config = "workbench.scripts.show_config:main"
 glue_launcher = "workbench.scripts.glue_launcher:main"
 ml_pipeline_batch = "workbench.scripts.ml_pipeline_batch:main"
 ml_pipeline_sqs = "workbench.scripts.ml_pipeline_sqs:main"
+lambda_launcher = "workbench.scripts.lambda_launcher:main"
 
 [tool.pytest.ini_options]
 addopts = "-v --cov-report term-missing"

workbench-0.8.198/scripts/delete_data_capture.py

@@ -0,0 +1,72 @@
+#!/usr/bin/env python3
+import boto3
+import argparse
+from collections import defaultdict
+
+# Workbench Imports
+from workbench.utils.config_manager import ConfigManager
+
+
+def list_and_delete_data_capture(bucket, preview=True):
+    s3 = boto3.client("s3")
+    prefix = "endpoints/"
+
+    # Get all endpoint prefixes
+    paginator = s3.get_paginator("list_objects_v2")
+    endpoints = set()
+
+    for page in paginator.paginate(Bucket=bucket, Prefix=prefix, Delimiter="/"):
+        for common_prefix in page.get("CommonPrefixes", []):
+            endpoint_prefix = common_prefix["Prefix"]
+            if endpoint_prefix != prefix:
+                endpoints.add(endpoint_prefix)
+
+    # Count and optionally delete files per endpoint
+    counts = defaultdict(int)
+    to_delete = []
+
+    for endpoint_prefix in endpoints:
+        data_capture_prefix = f"{endpoint_prefix}data_capture/"
+
+        for page in paginator.paginate(Bucket=bucket, Prefix=data_capture_prefix):
+            for obj in page.get("Contents", []):
+                if obj["Key"].endswith(".jsonl"):
+                    endpoint_name = endpoint_prefix.split("/")[-2]
+                    counts[endpoint_name] += 1
+                    to_delete.append(obj["Key"])
+
+    # Display results
+    if counts:
+        print(f"\n{'Endpoint':<50} {'Files'}")
+        print("-" * 60)
+        for endpoint, count in sorted(counts.items()):
+            print(f"{endpoint:<50} {count:>6}")
+        print("-" * 60)
+        print(f"{'TOTAL':<50} {sum(counts.values()):>6}\n")
+    else:
+        print("No .jsonl files found in data_capture directories")
+        return
+
+    # Delete if not preview mode
+    if not preview:
+        print("Deleting files...")
+        # Delete in batches of 1000
+        for i in range(0, len(to_delete), 1000):
+            batch = to_delete[i : i + 1000]
+            s3.delete_objects(Bucket=bucket, Delete={"Objects": [{"Key": k} for k in batch]})
+        print(f"Deleted {len(to_delete)} files")
+    else:
+        print("PREVIEW MODE - no files deleted")
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(description="Remove data capture files from SageMaker endpoints")
+    parser.add_argument("--delete", action="store_true", help="Actually delete files (default is preview)")
+    args = parser.parse_args()
+
+    # Get the default S3 bucket from Workbench config
+    config = ConfigManager()
+    bucket = config.get_config("WORKBENCH_BUCKET")
+    print(f"Using S3 bucket: {bucket}")
+
+    list_and_delete_data_capture(bucket, preview=not args.delete)

workbench-0.8.198/scripts/sagemaker_feature_group_issue.py

@@ -0,0 +1,61 @@
+import pandas as pd
+import sagemaker
+from sagemaker.feature_store.feature_group import FeatureGroup
+import time
+import multiprocessing
+import multiprocess
+
+# Toggle this flag to test spawn mode fix
+USE_SPAWN_MODE = True  # Set to True to fix the Tahoe hang issue
+
+if __name__ == "__main__":
+    if USE_SPAWN_MODE:
+        print("Using SPAWN mode (fix for Tahoe)")
+        multiprocessing.set_start_method("spawn", force=True)
+        multiprocess.set_start_method("spawn", force=True)
+    else:
+        print("Using default fork mode (will hang on Tahoe)")
+
+    # Create fake data
+    data = pd.DataFrame(
+        {
+            "record_id": [f"id_{i}" for i in range(10)],
+            "feature_1": [float(i) for i in range(10)],
+            "feature_2": [float(i * 2) for i in range(10)],
+            "event_time": [time.time()] * 10,
+        }
+    )
+
+    # Setup SageMaker session
+    sagemaker_session = sagemaker.Session()
+
+    # Define feature group
+    feature_group_name = "temp_delete_me"
+    feature_group = FeatureGroup(name=feature_group_name, sagemaker_session=sagemaker_session)
+
+    # Create feature definitions
+    feature_group.load_feature_definitions(data_frame=data)
+
+    # Create feature group
+    print("Creating feature group...")
+    feature_group.create(
+        s3_uri=f"s3://{sagemaker_session.default_bucket()}/featurestore",
+        record_identifier_name="record_id",
+        event_time_feature_name="event_time",
+        role_arn=sagemaker.get_execution_role(),
+        enable_online_store=True,
+    )
+
+    # Wait for feature group to be created (can take 1-2 minutes)
+    print("Waiting for feature group to be ready...")
+    status = feature_group.describe().get("FeatureGroupStatus")
+    while status == "Creating":
+        print(f"Status: {status}... waiting 10 seconds")
+        time.sleep(10)
+        status = feature_group.describe().get("FeatureGroupStatus")
+    print(f"Feature group status: {status}")
+
+    # This will hang on macOS Tahoe with USE_SPAWN_MODE=False
+    print("Starting ingest...")
+    feature_group.ingest(data_frame=data, max_workers=2, max_processes=2, wait=True)
+    print("Ingest completed!")

workbench-0.8.198/scripts/test_feature_set_creation.py

@@ -0,0 +1,20 @@
+import awswrangler as wr
+
+# Workbench Imports
+from workbench.core.transforms.pandas_transforms import PandasToFeatures
+
+import multiprocessing
+import multiprocess
+
+if __name__ == "__main__":
+    multiprocessing.set_start_method("spawn", force=True)
+    multiprocess.set_start_method("spawn", force=True)
+
+    s3_path = "s3://workbench-public-data/comp_chem/aqsol_public_data.csv"
+    aqsol_data = wr.s3.read_csv(s3_path)
+    aqsol_data.columns = aqsol_data.columns.str.lower()
+
+    to_features = PandasToFeatures("temp_delete_me")
+    to_features.set_input(aqsol_data, id_column="id")
+    to_features.set_output_tags(["aqsol", "public"])
+    to_features.transform()

{workbench-0.8.177 → workbench-0.8.198}/src/workbench/algorithms/dataframe/__init__.py

@@ -5,14 +5,13 @@ These classes provide functionality for Pandas Dataframes
 - TBD: TBD
 """
 
-from .proximity import Proximity, ProximityType
+from .proximity import Proximity
 from .feature_space_proximity import FeatureSpaceProximity
 from .fingerprint_proximity import FingerprintProximity
 from .projection_2d import Projection2D
 
 __all__ = [
     "Proximity",
-    "ProximityType",
     "FeatureSpaceProximity",
     "FingerprintProximity",
     "Projection2D",