Workflow: Enhance metric parsing and increase evaluation timeout

Author: mx-Liu123

pylib/default_graph.json

@@ -284,11 +284,11 @@
             },
             {
               "config": {
-                "command": "cd {current_exp_path} && {eval_cmd} > eval_out.txt 2>&1; cat eval_out.txt",
+                "command": "cd {current_exp_path} && ( {eval_cmd} ) > eval_out.txt 2>&1; cat eval_out.txt",
                 "output_vars": [
                   "test_output"
                 ],
-                "timeout": 12000
+                "timeout": 86400
               },
               "id": "step4_eval",
               "label": "4. Run Evaluator",
@@ -300,7 +300,7 @@
             },
             {
               "config": {
-                "code": "\nimport re\nlogger.log(\"\ud83d\udc1b DEBUG: Step 4.5 Parse Metric Started\")\noutput = context.get('test_output', '')\nlogger.log(f'\ud83d\udc1b DEBUG: test_output length: {len(output)}')\nif len(output) < 200: logger.log(f'\ud83d\udc1b DEBUG: test_output content: {output}')\n\nmetric = None\ntry:\n    match = re.search(r\"Best metric:\\s*([-\\d\\.]+)\", output)\n    if match:\n        metric = float(match.group(1))\n        logger.log(f'\ud83d\udc1b DEBUG: Matched Best metric: {metric}')\n    else:\n        logger.log(\"\ud83d\udc1b DEBUG: No 'Best metric' found.\")\nexcept Exception as e:\n    logger.log(f'\ud83d\udc1b DEBUG: Regex Error: {e}')\n\nif metric is None:\n    logger.log(\"\u26a0\ufe0f Could not parse metric. Setting current_metric to None.\")\n    context['current_metric'] = None # Explicitly set to None if not found\nelse:\n    context['current_metric'] = metric\n    logger.log(f'\u2705 Metric Parsed and Set: {metric}')\n"
+                "code": "\nimport re\nlogger.log(\"\ud83d\udc1b DEBUG: Step 4.5 Parse Metric Started\")\noutput = context.get('test_output', '')\nlogger.log(f'\ud83d\udc1b DEBUG: test_output length: {len(output)}')\nif len(output) < 200: logger.log(f'\ud83d\udc1b DEBUG: test_output content: {output}')\n\nmetric = None\ntry:\n    match = re.search(r\"Best metric.*?:\\s*([-\\d\\.]+)\", output)\n    if match:\n        metric = float(match.group(1))\n        logger.log(f'\ud83d\udc1b DEBUG: Matched Best metric: {metric}')\n    else:\n        logger.log(\"\ud83d\udc1b DEBUG: No 'Best metric' found.\")\nexcept Exception as e:\n    logger.log(f'\ud83d\udc1b DEBUG: Regex Error: {e}')\n\nif metric is None:\n    logger.log(\"\u26a0\ufe0f Could not parse metric. Setting current_metric to None.\")\n    context['current_metric'] = None # Explicitly set to None if not found\nelse:\n    context['current_metric'] = metric\n    logger.log(f'\u2705 Metric Parsed and Set: {metric}')\n"
               },
               "id": "step4_5_parse",
               "label": "4.5 Parse Metric",