def sclack 1

Author: R-Palazzo

sdgym/run_benchmark/run_benchmark.py

@@ -10,8 +10,9 @@
 from sdgym.run_benchmark.utils import (
     KEY_DATE_FILE,
     OUTPUT_DESTINATION_AWS,
-    SYNTHESIZERS,
+    SYNTHESIZERS_SPLIT,
     get_result_folder_name,
+    post_benchmark_launch_message,
 )
 from sdgym.s3 import get_s3_client, parse_s3_path
 
@@ -45,16 +46,18 @@ def main():
     aws_access_key_id = os.getenv('AWS_ACCESS_KEY_ID')
     aws_secret_access_key = os.getenv('AWS_SECRET_ACCESS_KEY')
     date_str = datetime.now(timezone.utc).strftime('%Y-%m-%d')
-    for synthesizer in SYNTHESIZERS:
+    for synthesizer_group in SYNTHESIZERS_SPLIT[:2]:
         benchmark_single_table_aws(
             output_destination=OUTPUT_DESTINATION_AWS,
+            dataset=['expedia_hotel_logs', 'fake_companies'],
             aws_access_key_id=aws_access_key_id,
             aws_secret_access_key=aws_secret_access_key,
-            synthesizers=[synthesizer],
+            synthesizers=synthesizer_group,
             compute_privacy_score=False,
         )
 
     append_benchmark_run(aws_access_key_id, aws_secret_access_key, date_str)
+    post_benchmark_launch_message()
 
 
 if __name__ == '__main__':

sdgym/run_benchmark/utils.py

@@ -1,15 +1,24 @@
 """Utils file for the run_benchmark module."""
 
+import os
 from datetime import datetime
 
-from sdgym.benchmark import SDV_SINGLE_TABLE_SYNTHESIZERS
+from slack_sdk import WebClient
 
 OUTPUT_DESTINATION_AWS = 's3://sdgym-benchmark/Debug/Issue_425/'
 UPLOAD_DESTINATION_AWS = 's3://sdgym-benchmark/Debug/Issue_425/'
 DEBUG_SLACK_CHANNEL = 'sdv-alerts-debug'
 SLACK_CHANNEL = 'sdv-alerts'
 KEY_DATE_FILE = '_BENCHMARK_DATES.json'
-SYNTHESIZERS = SDV_SINGLE_TABLE_SYNTHESIZERS
+
+# The synthesizers inside the same list will be run by the same ec2 instance
+SYNTHESIZERS_SPLIT = [
+    ['UniformSynthesizer', 'ColumnSynthesizer', 'GaussianCopulaSynthesizer'],
+    ['TVAESynthesizer'],
+    ['CopulaGANSynthesizer'],
+    ['CTGANSynthesizer'],
+    ['RealTabFormerSynthesizer'],
+]
 
 
 def get_result_folder_name(date_str):
@@ -20,3 +29,41 @@ def get_result_folder_name(date_str):
         raise ValueError(f'Invalid date format: {date_str}. Expected YYYY-MM-DD.')
 
     return f'SDGym_results_{date.month:02d}_{date.day:02d}_{date.year}'
+
+
+def _get_slack_client():
+    """Create an authenticated Slack client.
+
+    Returns:
+        WebClient:
+            An authenticated Slack WebClient instance.
+    """
+    token = os.getenv('SLACK_TOKEN')
+    client = WebClient(token=token)
+    return client
+
+
+def post_slack_message(channel, text):
+    """Post a message to a Slack channel."""
+    client = _get_slack_client()
+    client.chat_postMessage(channel=channel, text=text)
+
+
+def post_benchmark_launch_message():
+    """Post a message to the SDV Alerts Slack channel when the benchmark is launched."""
+    channel = DEBUG_SLACK_CHANNEL
+    body = 'SDGym benchmark has been launched! Results will be available soon.'
+    post_slack_message(channel, body)
+
+
+def post_run_summary(folder_name):
+    """Post run summary to sdv-alerts slack channel."""
+    channel = DEBUG_SLACK_CHANNEL
+    body = ''
+    body += f'SDGym benchmark results for {folder_name} are available!\n'
+    body += (
+        f'Check the results <{OUTPUT_DESTINATION_AWS}{folder_name}/{folder_name}_summary'
+        '.csv|here>.\n'
+    )
+
+    post_slack_message(channel, body)

tasks.py

@@ -206,9 +206,9 @@ def rmdir(c, path):
 @task
 def run_sdgym_benchmark(c):
     """Run the SDGym benchmark."""
-    c.run('python sdgym/_run_benchmark/run_benchmark.py')
+    c.run('python sdgym/run_benchmark/run_benchmark.py')
 
 @task
 def upload_benchmark_results(c, date=None):
     """Upload the benchmark results to S3."""
-    c.run(f'python sdgym/_run_benchmark/upload_benchmark_results.py {date}')
+    c.run(f'python sdgym/run_benchmark/upload_benchmark_results.py {date}')