dslighting 1.7.8__py3-none-any.whl → 1.7.16__py3-none-any.whl

dslighting/__init__.py

@@ -26,7 +26,7 @@ Advanced Usage:
 For more information, see: https://github.com/usail-hkust/dslighting
 """
 
-__version__ = "1.7.8"
+__version__ = "1.7.16"
 __author__ = "DSLighting Team"
 
 # Core API classes

dslighting/core/agent.py

@@ -377,8 +377,9 @@ class Agent:
                         print(f"[DEBUG 9] Registry not available, skipping grading")
                     else:
                         print(f"[DEBUG 10] Registry available, proceeding with initialization")
-                        # Initialize benchmark with registry
-                        registry_kwargs = {"data_dir": data_dir_path, "registry_dir": str(registry_dir)}
+                        # Initialize benchmark with registry (keep as Path objects)
+                        registry_dir_path = Path(registry_dir) if not isinstance(registry_dir, Path) else registry_dir
+                        registry_kwargs = {"data_dir": data_dir_path, "registry_dir": registry_dir_path}
 
                         try:
                             print(f"[DEBUG 11] Creating Registry with kwargs: {registry_kwargs}")
@@ -392,17 +393,113 @@ class Agent:
 
                             # Create simple wrapper class
                             class SimpleMLEBenchmark:
-                                def __init__(self, registry_instance):
+                                def __init__(self, registry_instance, logger, registry_dir, data_dir, task_id):
                                     self.registry = registry_instance
                                     self.problems = [{"competition_id": task_id}]
+                                    self.logger = logger
+                                    self.registry_dir = registry_dir
+                                    self.data_dir = data_dir
+                                    self.task_id = task_id
 
                                 async def grade(self, submission_path):
                                     """Grade submission using custom registry."""
                                     try:
-                                        competition = self.registry.get_competition(task_id)
+                                        # Read config directly from registry
+                                        import yaml
+                                        config_path = self.registry_dir / self.task_id / "config.yaml"
+
+                                        if not config_path.exists():
+                                            self.logger.warning(f"  Config not found: {config_path}")
+                                            return 0.0
+
+                                        with open(config_path) as f:
+                                            config = yaml.safe_load(f)
+
+                                        # Resolve paths relative to data_dir (from config.yaml)
+                                        # self.data_dir is the parent directory (e.g., /path/to/competitions)
+                                        # config["dataset"]["answers"] is relative path like "bike-sharing-demand/prepared/private/test_answer.csv"
+                                        answers_rel_path = config.get("dataset", {}).get("answers", "")
+                                        answers_path = self.data_dir / answers_rel_path
+
+                                        if not answers_path.exists():
+                                            self.logger.warning(f"  Answers file not found: {answers_path}")
+                                            self.logger.warning(f"  Looking for: {answers_path}")
+                                            return 0.0
+
+                                        self.logger.info(f"  Found answers file: {answers_path}")
+
+                                        # Import the actual Competition class from mlebench
+                                        from mlebench.registry import Competition
+                                        from mlebench.grade_helpers import Grader
+
+                                        # Load grader
+                                        grader_config = config.get("grader", {})
+                                        grader_name = grader_config.get("name", "rmsle")
+
+                                        # Import grade function if specified
+                                        grade_fn = None
+                                        if "grade_fn" in grader_config:
+                                            # Parse grade_fn format: mlebench.competitions.bike_sharing_demand.grade:grade
+                                            fn_str = grader_config["grade_fn"]
+                                            if ":" in fn_str:
+                                                module_path, fn_name = fn_str.rsplit(":", 1)
+                                                # Convert to file import if needed
+                                                if not module_path.startswith("file:"):
+                                                    fn_file = self.registry_dir / self.task_id / "grade.py"
+                                                    if fn_file.exists():
+                                                        fn_str = f"file:{fn_file}:{fn_name}"
+                                                else:
+                                                    # Try to import from mlebench
+                                                    try:
+                                                        import importlib
+                                                        importlib.import_module(module_path)
+                                                    except:
+                                                        pass
+
+                                        # Create a simple grader
+                                        if grade_fn or fn_str:
+                                            grader = Grader(
+                                                name=grader_name,
+                                                grade_fn=fn_str if fn_str else grade_fn,
+                                            )
+                                        else:
+                                            # Default RMSLE grader
+                                            grader = Grader(name="rmsle", grade_fn=None)
+
+                                        # Resolve paths
+                                        competition_dir = self.data_dir / self.task_id
+                                        private_dir = competition_dir / "prepared" / "private"
+                                        public_dir = competition_dir / "prepared" / "public"
+                                        raw_dir = competition_dir / "raw"
+                                        checksums = competition_dir / "checksums.txt"
+                                        leaderboard = competition_dir / "leaderboard.csv"
+
+                                        # Create placeholder prepare_fn
+                                        def dummy_prepare_fn(a, b, c):
+                                            return private_dir
+
+                                        # Create actual Competition object with all required fields
+                                        simple_comp = Competition(
+                                            id=config["id"],
+                                            name=config["name"],
+                                            description=config.get("description", ""),
+                                            grader=grader,
+                                            answers=answers_path,
+                                            gold_submission=answers_path,  # Use same as answers for grading
+                                            sample_submission=public_dir / "sampleSubmission.csv",
+                                            competition_type=config.get("competition_type", "standard"),
+                                            prepare_fn=dummy_prepare_fn,
+                                            raw_dir=raw_dir,
+                                            private_dir=private_dir,
+                                            public_dir=public_dir,
+                                            checksums=checksums,
+                                            leaderboard=leaderboard,
+                                        )
+
+                                        # Grade using mlebench's grade_csv
                                         report = grade_csv(
                                             submission_path,
-                                            competition,
+                                            simple_comp,
                                         )
                                         # Return the score (float), not the entire report
                                         score = report.score if report.score is not None else 0.0
@@ -410,9 +507,11 @@ class Agent:
                                         return score
                                     except Exception as e:
                                         self.logger.warning(f"  Grading failed: {e}")
+                                        import traceback
+                                        self.logger.warning(f"  Traceback: {traceback.format_exc()}")
                                         return 0.0
 
-                            benchmark = SimpleMLEBenchmark(custom_registry)
+                            benchmark = SimpleMLEBenchmark(custom_registry, self.logger, registry_dir_path, data_dir_path, task_id)
                             runner = self.get_runner()
                             runner.benchmark = benchmark
                             print(f"[DEBUG 13] ✓ Benchmark set successfully for task: {task_id}")

{dslighting-1.7.8.dist-info → dslighting-1.7.16.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dslighting
-Version: 1.7.8
+Version: 1.7.16
 Summary: Simplified API for Data Science Agent Automation
 Author: DSLighting Team
 License: AGPL-3.0
@@ -32,6 +32,10 @@ Requires-Dist: torch>=2.0.0
 Requires-Dist: scikit-learn>=1.0.0
 Requires-Dist: diskcache
 Requires-Dist: tenacity
+Requires-Dist: appdirs
+Requires-Dist: pyyaml
+Requires-Dist: tqdm
+Requires-Dist: py7zr
 Requires-Dist: nbformat
 Requires-Dist: nbclient
 Provides-Extra: full

{dslighting-1.7.8.dist-info → dslighting-1.7.16.dist-info}/RECORD

@@ -66,9 +66,9 @@ dsat/workflows/search/aide_workflow.py,sha256=mxIGXcueZGXpv1RXsQJ0YPWtvzICaFQeJo
 dsat/workflows/search/automind_workflow.py,sha256=b2JzqUDnDOt_SQdtAvC0fBCJzgTadLylbpgmpaS63Ls,12573
 dsat/workflows/templates/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 dsat/workflows/templates/basic_kaggle_loop.py,sha256=e6YLEpCArgWfKViwoti7SdygHsHp43sqP6VyMqnOJaA,3128
-dslighting/__init__.py,sha256=Q-q-kEjPM7UmlYlViXZX4Q8re3_Pg5eUssl-8sMbR_E,5152
+dslighting/__init__.py,sha256=EB2LmdQzrd-rZB6fEyW7QRG2q58I2X9QonTsMihkii4,5153
 dslighting/core/__init__.py,sha256=T4yYs0RQoz6DBarjOk12PeZq9YoPYrfl3Os0CPlzcB0,252
-dslighting/core/agent.py,sha256=nZEQcCqG4KTc3Iqz2yGdxYIJ_n6UPzyAK_Fmaa71wP8,31388
+dslighting/core/agent.py,sha256=qiAqJwrhry3T_HBfEndkghtRCy02uGUiHod4hpLkm9w,38211
 dslighting/core/config_builder.py,sha256=JMDW0JFj6PRCeP70D1Td8FYXC1bIupe4H6m8J1caO7Y,10294
 dslighting/core/data_loader.py,sha256=HdcDN-10FYfViu8Z0RSYjA2ne6VXVbfKEOZO7EpxLIc,25003
 dslighting/core/task_detector.py,sha256=xOYAV9yiboC8lDeDIEtxvucaGi6fENfeycrowWs-kP0,16300
@@ -2446,8 +2446,8 @@ mlebench/competitions/vinbigdata-chest-xray-abnormalities-detection/prepare_val.
 mlebench/competitions/whale-categorization-playground/grade.py,sha256=Wl1fNvpapAmc_Cdy2Cp68nEqeHB2XKnN91U9t-YLlXQ,1562
 mlebench/competitions/whale-categorization-playground/prepare.py,sha256=huo8fDBcGR413JF8m8Js8l8gkFCZpWL1st7Yd57Rjg8,4199
 mlebench/competitions/whale-categorization-playground/prepare_val.py,sha256=XvdnOvEJyTxovXQsoKLMKA_J6EMdxkpOe_SIgEdoOx8,7486
-dslighting-1.7.8.dist-info/METADATA,sha256=_aT7IWO_N5aRJYEj0RAIJVdiZSPA89UHKiiUEUzgh1M,18239
-dslighting-1.7.8.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-dslighting-1.7.8.dist-info/entry_points.txt,sha256=1xqIWL9-EG9t7nkERVeNRtMoRyIQe-6CJZJN-rdCYFQ,91
-dslighting-1.7.8.dist-info/top_level.txt,sha256=cmZU3ri0tz1IjiTKrz85Ih9614QlCrYqSqzlPMgaSM0,25
-dslighting-1.7.8.dist-info/RECORD,,
+dslighting-1.7.16.dist-info/METADATA,sha256=CqHn1grvJF-lyY06Q3GB9-0VKnwpyQSK98MsZBEvPEk,18326
+dslighting-1.7.16.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+dslighting-1.7.16.dist-info/entry_points.txt,sha256=1xqIWL9-EG9t7nkERVeNRtMoRyIQe-6CJZJN-rdCYFQ,91
+dslighting-1.7.16.dist-info/top_level.txt,sha256=cmZU3ri0tz1IjiTKrz85Ih9614QlCrYqSqzlPMgaSM0,25
+dslighting-1.7.16.dist-info/RECORD,,