cleaning 516 jan

Author: R-Palazzo

.github/workflows/run_benchmark.yml

@@ -6,10 +6,6 @@ on:
       modality:
         required: true
         type: string
-      sdgym_ref:
-        required: false
-        type: string
-        default: issue-516-add-workflows
     secrets:
       SDV_ENTERPRISE_USERNAME:
         required: true
@@ -55,8 +51,7 @@ jobs:
               license_key='${LICENSE_KEY}',
               package='sdv-enterprise',
           )
-          "
-          python -m pip install "sdgym[all] @ git+https://github.com/sdv-dev/SDGym.git@${{ inputs.sdgym_ref }}"
+          python -m pip install sdgym[all]
 
           echo "VIRTUAL_ENV=$(pwd)/venv" >> $GITHUB_ENV
           echo "$(pwd)/venv/bin" >> $GITHUB_PATH

.github/workflows/run_benchmark_multi_table.yaml

@@ -4,9 +4,6 @@ on:
   workflow_dispatch:
   schedule:
     - cron: "0 5 1 * *"
-  push:
-    branches:
-      - issue-516-add-workflows
 
 jobs:
   call-run-sdgym-benchmark:

.github/workflows/run_benchmark_single_table.yml

@@ -4,9 +4,6 @@ on:
   workflow_dispatch:
   schedule:
     - cron: "0 5 1 * *"
-  push:
-    branches:
-      - issue-516-add-workflows
 
 jobs:
   call-run-sdgym-benchmark:

.github/workflows/upload_benchmark_results_multi_table.yml

@@ -7,9 +7,6 @@ on:
   workflow_dispatch:
   schedule:
     - cron: "0 6 * * *"
-  push:
-    branches:
-      - issue-516-add-workflows
 
 jobs:
   call-upload-benchmark-results:

sdgym/benchmark.py

@@ -1644,7 +1644,7 @@ def _get_user_data_script(access_key, secret_key, region_name, script_content):
 
         echo "======== Install Dependencies in venv ============"
         pip install --upgrade pip
-        pip install "sdgym[all] @ git+https://github.com/sdv-dev/SDGym.git@feature_branch/mutli_table_benchmark"
+        pip install sdgym[all]
         pip install s3fs
 
         echo "======== Write Script ==========="

sdgym/run_benchmark/upload_benchmark_results.py

@@ -34,6 +34,8 @@
     'Column': 'top center',
     'CopulaGAN': 'top center',
     'RealTabFormer': 'bottom center',
+    'HSA': 'bottom center',
+    'Independent': 'top center',
 }
 RESULT_FILENAME = 'SDGym Monthly Run.xlsx'
 

sdgym/run_benchmark/utils.py

@@ -10,8 +10,7 @@
 
 from sdgym.s3 import parse_s3_path
 
-OUTPUT_DESTINATION_AWS = 's3://sdgym-benchmark/Debug/GCP_Github_2/'
-UPLOAD_DESTINATION_AWS = 's3://sdgym-benchmark/Debug/GCP/'
+OUTPUT_DESTINATION_AWS = 's3://sdgym-benchmark/Benchmarks/'
 DEBUG_SLACK_CHANNEL = 'sdv-alerts-debug'
 SLACK_CHANNEL = 'sdv-alerts'
 KEY_DATE_FILE = '_BENCHMARK_DATES.json'
@@ -49,7 +48,7 @@
 ]
 MODALITY_TO_GDRIVE_LINK = {
     'single_table': 'https://docs.google.com/spreadsheets/d/1W3tsGOOtbtTw3g0EVE0irLgY_TN_cy2W4ONiZQ57OPo/edit?usp=sharing',
-    'multi_table': 'https://docs.google.com/spreadsheets/d/1R13RktVvKnxRecYIge07OBpbX1vbEkE2D1_2idNAKSY/edit?usp=sharing',
+    'multi_table': 'https://docs.google.com/spreadsheets/d/1srmXx2ddq025hqzAE4JRdebuoBfro_7wbgeUHUkMEMM/edit?usp=sharing',
 }
 
 # The synthesizers inside the same list will be run by the same ec2 instance
@@ -102,7 +101,7 @@ def post_slack_message(channel, text):
 
 def post_benchmark_launch_message(date_str, compute_service='AWS', modality='single_table'):
     """Post a message to the SDV Alerts Slack channel when the benchmark is launched."""
-    channel = DEBUG_SLACK_CHANNEL
+    channel = SLACK_CHANNEL
     folder_name = get_result_folder_name(date_str)
     bucket, prefix = parse_s3_path(OUTPUT_DESTINATION_AWS)
     url_link = get_s3_console_link(bucket, f'{prefix}{modality}/{folder_name}/')
@@ -114,7 +113,7 @@ def post_benchmark_launch_message(date_str, compute_service='AWS', modality='sin
 
 def post_benchmark_uploaded_message(folder_name, commit_url=None, modality='single_table'):
     """Post benchmark uploaded message to sdv-alerts slack channel."""
-    channel = DEBUG_SLACK_CHANNEL
+    channel = SLACK_CHANNEL
     bucket, prefix = parse_s3_path(OUTPUT_DESTINATION_AWS)
     modality_text = modality.replace('_', '-')
     url_link = get_s3_console_link(bucket, quote_plus(f'{prefix}{modality}/SDGym Monthly Run.xlsx'))

tests/unit/run_benchmark/test_upload_benchmark_result.py

@@ -195,7 +195,7 @@ def test_upload_results(
         '10_01_2023_Detailed_results': 'results',
         '10_01_2023_plot_data': 'df_to_plot',
     }
-    local_path = str(Path('/tmp/sdgym_results/SDGym Monthly Run.xlsx'))
+    local_path = str(Path('/tmp/sdgym_results/[single_table] SDGym Monthly Run.xlsx'))
     mock_extract_google_file_id.return_value = 'google_file_id'
 
     # Run

tests/unit/run_benchmark/test_utils.py

@@ -107,7 +107,7 @@ def test_post_benchmark_launch_message(
     mock_get_s3_console_link.assert_called_once_with(
         'my-bucket', f'my-prefix/single_table/{folder_name}/'
     )
-    mock_post_slack_message.assert_called_once_with('sdv-alerts-debug', expected_body)
+    mock_post_slack_message.assert_called_once_with('sdv-alerts', expected_body)
 
 
 @patch('sdgym.run_benchmark.utils.post_slack_message')
@@ -135,7 +135,7 @@ def test_post_benchmark_uploaded_message(
     post_benchmark_uploaded_message(folder_name)
 
     # Assert
-    mock_post_slack_message.assert_called_once_with('sdv-alerts-debug', expected_body)
+    mock_post_slack_message.assert_called_once_with('sdv-alerts', expected_body)
     mock_parse_s3_path.assert_called_once_with(OUTPUT_DESTINATION_AWS)
     mock_get_s3_console_link.assert_called_once_with(
         'my-bucket', 'my-prefix%2Fsingle_table%2FSDGym+Monthly+Run.xlsx'
@@ -169,7 +169,7 @@ def test_post_benchmark_uploaded_message_with_commit(
     post_benchmark_uploaded_message(folder_name, commit_url)
 
     # Assert
-    mock_post_slack_message.assert_called_once_with('sdv-alerts-debug', expected_body)
+    mock_post_slack_message.assert_called_once_with('sdv-alerts', expected_body)
     mock_parse_s3_path.assert_called_once_with(OUTPUT_DESTINATION_AWS)
     mock_get_s3_console_link.assert_called_once_with(
         'my-bucket', 'my-prefix%2Fsingle_table%2FSDGym+Monthly+Run.xlsx'