ins-pricing 0.5.0__tar.gz → 0.5.1__tar.gz

{ins_pricing-0.5.0 → ins_pricing-0.5.1}/PKG-INFO

@@ -1,182 +1,182 @@
-Metadata-Version: 2.4
-Name: ins_pricing
-Version: 0.5.0
-Summary: Reusable modelling, pricing, governance, and reporting utilities.
-Author: meishi125478
-License: Proprietary
-Keywords: pricing,insurance,bayesopt,ml
-Classifier: Programming Language :: Python :: 3
-Classifier: Programming Language :: Python :: 3 :: Only
-Classifier: Programming Language :: Python :: 3.9
-Classifier: License :: Other/Proprietary License
-Classifier: Operating System :: OS Independent
-Classifier: Intended Audience :: Developers
-Requires-Python: >=3.9
-Description-Content-Type: text/markdown
-Requires-Dist: numpy>=1.20
-Requires-Dist: pandas>=1.4
-Provides-Extra: bayesopt
-Requires-Dist: torch>=1.13; extra == "bayesopt"
-Requires-Dist: optuna>=3.0; extra == "bayesopt"
-Requires-Dist: xgboost>=1.6; extra == "bayesopt"
-Requires-Dist: scikit-learn>=1.1; extra == "bayesopt"
-Requires-Dist: statsmodels>=0.13; extra == "bayesopt"
-Requires-Dist: joblib>=1.2; extra == "bayesopt"
-Requires-Dist: matplotlib>=3.5; extra == "bayesopt"
-Provides-Extra: plotting
-Requires-Dist: matplotlib>=3.5; extra == "plotting"
-Requires-Dist: scikit-learn>=1.1; extra == "plotting"
-Provides-Extra: explain
-Requires-Dist: torch>=1.13; extra == "explain"
-Requires-Dist: shap>=0.41; extra == "explain"
-Requires-Dist: scikit-learn>=1.1; extra == "explain"
-Provides-Extra: geo
-Requires-Dist: contextily>=1.3; extra == "geo"
-Requires-Dist: matplotlib>=3.5; extra == "geo"
-Provides-Extra: gnn
-Requires-Dist: torch>=1.13; extra == "gnn"
-Requires-Dist: pynndescent>=0.5; extra == "gnn"
-Requires-Dist: torch-geometric>=2.3; extra == "gnn"
-Provides-Extra: all
-Requires-Dist: torch>=1.13; extra == "all"
-Requires-Dist: optuna>=3.0; extra == "all"
-Requires-Dist: xgboost>=1.6; extra == "all"
-Requires-Dist: scikit-learn>=1.1; extra == "all"
-Requires-Dist: statsmodels>=0.13; extra == "all"
-Requires-Dist: joblib>=1.2; extra == "all"
-Requires-Dist: matplotlib>=3.5; extra == "all"
-Requires-Dist: shap>=0.41; extra == "all"
-Requires-Dist: contextily>=1.3; extra == "all"
-Requires-Dist: pynndescent>=0.5; extra == "all"
-Requires-Dist: torch-geometric>=2.3; extra == "all"
-
-# Insurance-Pricing
-
-A reusable toolkit for insurance modeling, pricing, governance, and reporting.
-
-## Overview
-
-Insurance-Pricing (ins_pricing) is an enterprise-grade Python library designed for machine learning
-model training, pricing calculations, and model governance workflows in the insurance industry.
-
-### Core Modules
-
-| Module | Description |
-|--------|-------------|
-| modelling | ML model training (GLM, XGBoost, ResNet, FT-Transformer, GNN) and model interpretability |
-| pricing | Factor table construction, numeric binning, premium calibration, exposure calculation, PSI monitoring |
-| production | Model prediction, batch scoring, data drift detection, production metrics monitoring |
-| governance | Model registry, version management, approval workflows, audit logging |
-| reporting | Report generation (Markdown format), report scheduling |
-| utils | Data validation, performance profiling, device management, logging configuration |
-
-### Quick Start
-
-```python
-# Model training with Bayesian optimization
-from ins_pricing import bayesopt as ropt
-
-model = ropt.BayesOptModel(
-    train_data, test_data,
-    model_name='my_model',
-    resp_nme='target',
-    weight_nme='weight',
-    factor_nmes=feature_list,
-    cate_list=categorical_features,
-)
-model.bayesopt_xgb(max_evals=100)      # Train XGBoost
-model.bayesopt_resnet(max_evals=50)    # Train ResNet
-model.bayesopt_ft(max_evals=50)        # Train FT-Transformer
-
-# Pricing: build factor table
-from ins_pricing.pricing import build_factor_table
-factors = build_factor_table(
-    df,
-    factor_col='age_band',
-    loss_col='claim_amount',
-    exposure_col='exposure',
-)
-
-# Production: batch scoring
-from ins_pricing.production import batch_score
-scores = batch_score(model.trainers['xgb'].predict, df)
-
-# Model governance
-from ins_pricing.governance import ModelRegistry
-registry = ModelRegistry('models.json')
-registry.register(model_name, version, metrics=metrics)
-```
-
-### Project Structure
-
-```
-ins_pricing/
-  cli/                    # Command-line entry points
-  modelling/
-    core/bayesopt/         # ML model training core
-    explain/               # Model interpretability
-    plotting/              # Model visualization
-  pricing/                 # Insurance pricing module
-  production/              # Production deployment module
-  governance/              # Model governance
-  reporting/               # Report generation
-  utils/                   # Utilities
-  tests/                   # Test suite
-```
-
-### Installation
-
-```bash
-# Basic installation
-pip install ins_pricing
-
-# Full installation (all optional dependencies)
-pip install ins_pricing[all]
-
-# Install specific extras
-pip install ins_pricing[bayesopt]    # Model training
-pip install ins_pricing[explain]     # Model explanation
-pip install ins_pricing[plotting]    # Visualization
-pip install ins_pricing[gnn]         # Graph neural networks
-```
-
-#### Multi-platform and GPU notes
-
-- Install the correct PyTorch build for your platform/GPU before installing extras.
-- Torch Geometric requires platform-specific wheels; follow the official PyG install guide.
-- Multi-GPU uses torch.distributed/DataParallel where supported; Windows disables CUDA DDP.
-
----
-## PyPI Upload (scripts)
-
-This repo includes upload scripts for Windows and Linux/macOS.
-
-### Windows
-
-```cmd
-set TWINE_PASSWORD=your_pypi_token_here
-python -m build
-upload_to_pypi.bat
-```
-
-### Linux / macOS
-
-```bash
-chmod +x upload_to_pypi.sh
-export TWINE_PASSWORD='your_pypi_token_here'
-python -m build
-./upload_to_pypi.sh
-```
-
-### Makefile (if make is available)
-
-```bash
-make build
-make upload
-```
-
-### Tips
-
-- Never commit tokens to version control.
-- Use environment variables or secret managers to store credentials.
-- Test with TestPyPI before publishing when needed.
+Metadata-Version: 2.4
+Name: ins_pricing
+Version: 0.5.1
+Summary: Reusable modelling, pricing, governance, and reporting utilities.
+Author: meishi125478
+License: Proprietary
+Keywords: pricing,insurance,bayesopt,ml
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3 :: Only
+Classifier: Programming Language :: Python :: 3.9
+Classifier: License :: Other/Proprietary License
+Classifier: Operating System :: OS Independent
+Classifier: Intended Audience :: Developers
+Requires-Python: >=3.9
+Description-Content-Type: text/markdown
+Requires-Dist: numpy>=1.20
+Requires-Dist: pandas>=1.4
+Provides-Extra: bayesopt
+Requires-Dist: torch>=1.13; extra == "bayesopt"
+Requires-Dist: optuna>=3.0; extra == "bayesopt"
+Requires-Dist: xgboost>=1.6; extra == "bayesopt"
+Requires-Dist: scikit-learn>=1.1; extra == "bayesopt"
+Requires-Dist: statsmodels>=0.13; extra == "bayesopt"
+Requires-Dist: joblib>=1.2; extra == "bayesopt"
+Requires-Dist: matplotlib>=3.5; extra == "bayesopt"
+Provides-Extra: plotting
+Requires-Dist: matplotlib>=3.5; extra == "plotting"
+Requires-Dist: scikit-learn>=1.1; extra == "plotting"
+Provides-Extra: explain
+Requires-Dist: torch>=1.13; extra == "explain"
+Requires-Dist: shap>=0.41; extra == "explain"
+Requires-Dist: scikit-learn>=1.1; extra == "explain"
+Provides-Extra: geo
+Requires-Dist: contextily>=1.3; extra == "geo"
+Requires-Dist: matplotlib>=3.5; extra == "geo"
+Provides-Extra: gnn
+Requires-Dist: torch>=1.13; extra == "gnn"
+Requires-Dist: pynndescent>=0.5; extra == "gnn"
+Requires-Dist: torch-geometric>=2.3; extra == "gnn"
+Provides-Extra: all
+Requires-Dist: torch>=1.13; extra == "all"
+Requires-Dist: optuna>=3.0; extra == "all"
+Requires-Dist: xgboost>=1.6; extra == "all"
+Requires-Dist: scikit-learn>=1.1; extra == "all"
+Requires-Dist: statsmodels>=0.13; extra == "all"
+Requires-Dist: joblib>=1.2; extra == "all"
+Requires-Dist: matplotlib>=3.5; extra == "all"
+Requires-Dist: shap>=0.41; extra == "all"
+Requires-Dist: contextily>=1.3; extra == "all"
+Requires-Dist: pynndescent>=0.5; extra == "all"
+Requires-Dist: torch-geometric>=2.3; extra == "all"
+
+# Insurance-Pricing
+
+A reusable toolkit for insurance modeling, pricing, governance, and reporting.
+
+## Overview
+
+Insurance-Pricing (ins_pricing) is an enterprise-grade Python library designed for machine learning
+model training, pricing calculations, and model governance workflows in the insurance industry.
+
+### Core Modules
+
+| Module | Description |
+|--------|-------------|
+| modelling | ML model training (GLM, XGBoost, ResNet, FT-Transformer, GNN) and model interpretability |
+| pricing | Factor table construction, numeric binning, premium calibration, exposure calculation, PSI monitoring |
+| production | Model prediction, batch scoring, data drift detection, production metrics monitoring |
+| governance | Model registry, version management, approval workflows, audit logging |
+| reporting | Report generation (Markdown format), report scheduling |
+| utils | Data validation, performance profiling, device management, logging configuration |
+
+### Quick Start
+
+```python
+# Model training with Bayesian optimization
+from ins_pricing import bayesopt as ropt
+
+model = ropt.BayesOptModel(
+    train_data, test_data,
+    model_name='my_model',
+    resp_nme='target',
+    weight_nme='weight',
+    factor_nmes=feature_list,
+    cate_list=categorical_features,
+)
+model.bayesopt_xgb(max_evals=100)      # Train XGBoost
+model.bayesopt_resnet(max_evals=50)    # Train ResNet
+model.bayesopt_ft(max_evals=50)        # Train FT-Transformer
+
+# Pricing: build factor table
+from ins_pricing.pricing import build_factor_table
+factors = build_factor_table(
+    df,
+    factor_col='age_band',
+    loss_col='claim_amount',
+    exposure_col='exposure',
+)
+
+# Production: batch scoring
+from ins_pricing.production import batch_score
+scores = batch_score(model.trainers['xgb'].predict, df)
+
+# Model governance
+from ins_pricing.governance import ModelRegistry
+registry = ModelRegistry('models.json')
+registry.register(model_name, version, metrics=metrics)
+```
+
+### Project Structure
+
+```
+ins_pricing/
+  cli/                    # Command-line entry points
+  modelling/
+    core/bayesopt/         # ML model training core
+    explain/               # Model interpretability
+    plotting/              # Model visualization
+  pricing/                 # Insurance pricing module
+  production/              # Production deployment module
+  governance/              # Model governance
+  reporting/               # Report generation
+  utils/                   # Utilities
+  tests/                   # Test suite
+```
+
+### Installation
+
+```bash
+# Basic installation
+pip install ins_pricing
+
+# Full installation (all optional dependencies)
+pip install ins_pricing[all]
+
+# Install specific extras
+pip install ins_pricing[bayesopt]    # Model training
+pip install ins_pricing[explain]     # Model explanation
+pip install ins_pricing[plotting]    # Visualization
+pip install ins_pricing[gnn]         # Graph neural networks
+```
+
+#### Multi-platform and GPU notes
+
+- Install the correct PyTorch build for your platform/GPU before installing extras.
+- Torch Geometric requires platform-specific wheels; follow the official PyG install guide.
+- Multi-GPU uses torch.distributed/DataParallel where supported; Windows disables CUDA DDP.
+
+---
+## PyPI Upload (scripts)
+
+This repo includes upload scripts for Windows and Linux/macOS.
+
+### Windows
+
+```cmd
+set TWINE_PASSWORD=your_pypi_token_here
+python -m build
+upload_to_pypi.bat
+```
+
+### Linux / macOS
+
+```bash
+chmod +x upload_to_pypi.sh
+export TWINE_PASSWORD='your_pypi_token_here'
+python -m build
+./upload_to_pypi.sh
+```
+
+### Makefile (if make is available)
+
+```bash
+make build
+make upload
+```
+
+### Tips
+
+- Never commit tokens to version control.
+- Use environment variables or secret managers to store credentials.
+- Test with TestPyPI before publishing when needed.

{ins_pricing-0.5.0 → ins_pricing-0.5.1}/ins_pricing/cli/BayesOpt_entry.py

@@ -11,11 +11,21 @@ import json
 import os
 import sys
 
-if __package__ in {None, ""}:
-    if importlib.util.find_spec("ins_pricing") is None:
-        repo_root = Path(__file__).resolve().parents[2]
-        if str(repo_root) not in sys.path:
-            sys.path.insert(0, str(repo_root))
+def _ensure_repo_root() -> None:
+    if __package__ not in {None, ""}:
+        return
+    if importlib.util.find_spec("ins_pricing") is not None:
+        return
+    bootstrap_path = Path(__file__).resolve().parents[1] / "utils" / "bootstrap.py"
+    spec = importlib.util.spec_from_file_location("ins_pricing.cli.utils.bootstrap", bootstrap_path)
+    if spec is None or spec.loader is None:
+        return
+    module = importlib.util.module_from_spec(spec)
+    spec.loader.exec_module(module)
+    module.ensure_repo_root()
+
+
+_ensure_repo_root()
 
 def _apply_env_from_config(argv: list[str]) -> None:
     if "--config-json" not in argv:

{ins_pricing-0.5.0 → ins_pricing-0.5.1}/ins_pricing/cli/BayesOpt_incremental.py

@@ -20,11 +20,22 @@ from pathlib import Path
 import importlib.util
 import sys
 
-if __package__ in {None, ""}:
-    if importlib.util.find_spec("ins_pricing") is None:
-        repo_root = Path(__file__).resolve().parents[2]
-        if str(repo_root) not in sys.path:
-            sys.path.insert(0, str(repo_root))
+
+def _ensure_repo_root() -> None:
+    if __package__ not in {None, ""}:
+        return
+    if importlib.util.find_spec("ins_pricing") is not None:
+        return
+    bootstrap_path = Path(__file__).resolve().parents[1] / "utils" / "bootstrap.py"
+    spec = importlib.util.spec_from_file_location("ins_pricing.cli.utils.bootstrap", bootstrap_path)
+    if spec is None or spec.loader is None:
+        return
+    module = importlib.util.module_from_spec(spec)
+    spec.loader.exec_module(module)
+    module.ensure_repo_root()
+
+
+_ensure_repo_root()
 
 import argparse
 import json
@@ -422,9 +433,20 @@ class IncrementalUpdateRunner:
         self.prediction_cache_dir = runtime_cfg["prediction_cache_dir"]
         self.prediction_cache_format = runtime_cfg["prediction_cache_format"]
         self.plot_path_style = runtime_cfg["plot_path_style"]
-        self.xgb_max_depth_max = runtime_cfg["xgb_max_depth_max"]
-        self.xgb_n_estimators_max = runtime_cfg["xgb_n_estimators_max"]
-        self.optuna_storage = runtime_cfg["optuna_storage"]
+        self.xgb_max_depth_max = runtime_cfg["xgb_max_depth_max"]
+        self.xgb_n_estimators_max = runtime_cfg["xgb_n_estimators_max"]
+        self.xgb_gpu_id = runtime_cfg["xgb_gpu_id"]
+        self.xgb_cleanup_per_fold = runtime_cfg["xgb_cleanup_per_fold"]
+        self.xgb_cleanup_synchronize = runtime_cfg["xgb_cleanup_synchronize"]
+        self.xgb_use_dmatrix = runtime_cfg["xgb_use_dmatrix"]
+        self.ft_cleanup_per_fold = runtime_cfg["ft_cleanup_per_fold"]
+        self.ft_cleanup_synchronize = runtime_cfg["ft_cleanup_synchronize"]
+        self.resn_cleanup_per_fold = runtime_cfg["resn_cleanup_per_fold"]
+        self.resn_cleanup_synchronize = runtime_cfg["resn_cleanup_synchronize"]
+        self.gnn_cleanup_per_fold = runtime_cfg["gnn_cleanup_per_fold"]
+        self.gnn_cleanup_synchronize = runtime_cfg["gnn_cleanup_synchronize"]
+        self.optuna_cleanup_synchronize = runtime_cfg["optuna_cleanup_synchronize"]
+        self.optuna_storage = runtime_cfg["optuna_storage"]
         self.optuna_study_prefix = runtime_cfg["optuna_study_prefix"]
         self.best_params_files = runtime_cfg["best_params_files"]
         self.reuse_best_params = runtime_cfg["reuse_best_params"]
@@ -593,8 +615,19 @@ class IncrementalUpdateRunner:
             use_ft_ddp=self.cfg.get("use_ft_ddp", False),
             use_gnn_ddp=self.cfg.get("use_gnn_ddp", False),
             output_dir=str(self.output_root) if self.output_root else None,
-            xgb_max_depth_max=self.xgb_max_depth_max,
-            xgb_n_estimators_max=self.xgb_n_estimators_max,
+            xgb_max_depth_max=self.xgb_max_depth_max,
+            xgb_n_estimators_max=self.xgb_n_estimators_max,
+            xgb_gpu_id=self.xgb_gpu_id,
+            xgb_cleanup_per_fold=self.xgb_cleanup_per_fold,
+            xgb_cleanup_synchronize=self.xgb_cleanup_synchronize,
+            xgb_use_dmatrix=self.xgb_use_dmatrix,
+            ft_cleanup_per_fold=self.ft_cleanup_per_fold,
+            ft_cleanup_synchronize=self.ft_cleanup_synchronize,
+            resn_cleanup_per_fold=self.resn_cleanup_per_fold,
+            resn_cleanup_synchronize=self.resn_cleanup_synchronize,
+            gnn_cleanup_per_fold=self.gnn_cleanup_per_fold,
+            gnn_cleanup_synchronize=self.gnn_cleanup_synchronize,
+            optuna_cleanup_synchronize=self.optuna_cleanup_synchronize,
             resn_weight_decay=self.cfg.get("resn_weight_decay"),
             final_ensemble=bool(self.cfg.get("final_ensemble", False)),
             final_ensemble_k=int(self.cfg.get("final_ensemble_k", 3)),

{ins_pricing-0.5.0 → ins_pricing-0.5.1}/ins_pricing/cli/Explain_Run.py

@@ -5,11 +5,22 @@ from typing import Optional
 import importlib.util
 import sys
 
-if __package__ in {None, ""}:
-    if importlib.util.find_spec("ins_pricing") is None:
-        repo_root = Path(__file__).resolve().parents[2]
-        if str(repo_root) not in sys.path:
-            sys.path.insert(0, str(repo_root))
+
+def _ensure_repo_root() -> None:
+    if __package__ not in {None, ""}:
+        return
+    if importlib.util.find_spec("ins_pricing") is not None:
+        return
+    bootstrap_path = Path(__file__).resolve().parents[1] / "utils" / "bootstrap.py"
+    spec = importlib.util.spec_from_file_location("ins_pricing.cli.utils.bootstrap", bootstrap_path)
+    if spec is None or spec.loader is None:
+        return
+    module = importlib.util.module_from_spec(spec)
+    spec.loader.exec_module(module)
+    module.ensure_repo_root()
+
+
+_ensure_repo_root()
 
 try:
     from ins_pricing.cli.utils.notebook_utils import run_from_config, run_from_config_cli  # type: ignore

{ins_pricing-0.5.0 → ins_pricing-0.5.1}/ins_pricing/cli/Explain_entry.py

@@ -6,11 +6,21 @@ from pathlib import Path
 import importlib.util
 import sys
 
-if __package__ in {None, ""}:
-    if importlib.util.find_spec("ins_pricing") is None:
-        repo_root = Path(__file__).resolve().parents[2]
-        if str(repo_root) not in sys.path:
-            sys.path.insert(0, str(repo_root))
+def _ensure_repo_root() -> None:
+    if __package__ not in {None, ""}:
+        return
+    if importlib.util.find_spec("ins_pricing") is not None:
+        return
+    bootstrap_path = Path(__file__).resolve().parents[1] / "utils" / "bootstrap.py"
+    spec = importlib.util.spec_from_file_location("ins_pricing.cli.utils.bootstrap", bootstrap_path)
+    if spec is None or spec.loader is None:
+        return
+    module = importlib.util.module_from_spec(spec)
+    spec.loader.exec_module(module)
+    module.ensure_repo_root()
+
+
+_ensure_repo_root()
 
 import argparse
 import json
@@ -460,9 +470,20 @@ def explain_from_config(args: argparse.Namespace) -> None:
             "epochs": int(runtime_cfg["epochs"]),
             "use_gpu": bool(cfg.get("use_gpu", True)),
             "output_dir": output_dir,
-            "xgb_max_depth_max": runtime_cfg["xgb_max_depth_max"],
-            "xgb_n_estimators_max": runtime_cfg["xgb_n_estimators_max"],
-            "resn_weight_decay": cfg.get("resn_weight_decay"),
+            "xgb_max_depth_max": runtime_cfg["xgb_max_depth_max"],
+            "xgb_n_estimators_max": runtime_cfg["xgb_n_estimators_max"],
+            "xgb_gpu_id": runtime_cfg["xgb_gpu_id"],
+            "xgb_cleanup_per_fold": runtime_cfg["xgb_cleanup_per_fold"],
+            "xgb_cleanup_synchronize": runtime_cfg["xgb_cleanup_synchronize"],
+            "xgb_use_dmatrix": runtime_cfg["xgb_use_dmatrix"],
+            "ft_cleanup_per_fold": runtime_cfg["ft_cleanup_per_fold"],
+            "ft_cleanup_synchronize": runtime_cfg["ft_cleanup_synchronize"],
+            "resn_cleanup_per_fold": runtime_cfg["resn_cleanup_per_fold"],
+            "resn_cleanup_synchronize": runtime_cfg["resn_cleanup_synchronize"],
+            "gnn_cleanup_per_fold": runtime_cfg["gnn_cleanup_per_fold"],
+            "gnn_cleanup_synchronize": runtime_cfg["gnn_cleanup_synchronize"],
+            "optuna_cleanup_synchronize": runtime_cfg["optuna_cleanup_synchronize"],
+            "resn_weight_decay": cfg.get("resn_weight_decay"),
             "final_ensemble": bool(cfg.get("final_ensemble", False)),
             "final_ensemble_k": int(cfg.get("final_ensemble_k", 3)),
             "final_refit": bool(cfg.get("final_refit", True)),

{ins_pricing-0.5.0 → ins_pricing-0.5.1}/ins_pricing/cli/Pricing_Run.py

@@ -5,11 +5,22 @@ from typing import Optional
 import importlib.util
 import sys
 
-if __package__ in {None, ""}:
-    if importlib.util.find_spec("ins_pricing") is None:
-        repo_root = Path(__file__).resolve().parents[2]
-        if str(repo_root) not in sys.path:
-            sys.path.insert(0, str(repo_root))
+
+def _ensure_repo_root() -> None:
+    if __package__ not in {None, ""}:
+        return
+    if importlib.util.find_spec("ins_pricing") is not None:
+        return
+    bootstrap_path = Path(__file__).resolve().parents[1] / "utils" / "bootstrap.py"
+    spec = importlib.util.spec_from_file_location("ins_pricing.cli.utils.bootstrap", bootstrap_path)
+    if spec is None or spec.loader is None:
+        return
+    module = importlib.util.module_from_spec(spec)
+    spec.loader.exec_module(module)
+    module.ensure_repo_root()
+
+
+_ensure_repo_root()
 
 try:
     from ins_pricing.cli.utils.notebook_utils import run_from_config, run_from_config_cli  # type: ignore

{ins_pricing-0.5.0 → ins_pricing-0.5.1}/ins_pricing/cli/bayesopt_entry_runner.py

@@ -15,11 +15,22 @@ from pathlib import Path
 import importlib.util
 import sys
 
-if __package__ in {None, ""}:
-    if importlib.util.find_spec("ins_pricing") is None:
-        repo_root = Path(__file__).resolve().parents[2]
-        if str(repo_root) not in sys.path:
-            sys.path.insert(0, str(repo_root))
+
+def _ensure_repo_root() -> None:
+    if __package__ not in {None, ""}:
+        return
+    if importlib.util.find_spec("ins_pricing") is not None:
+        return
+    bootstrap_path = Path(__file__).resolve().parents[1] / "utils" / "bootstrap.py"
+    spec = importlib.util.spec_from_file_location("ins_pricing.cli.utils.bootstrap", bootstrap_path)
+    if spec is None or spec.loader is None:
+        return
+    module = importlib.util.module_from_spec(spec)
+    spec.loader.exec_module(module)
+    module.ensure_repo_root()
+
+
+_ensure_repo_root()
 
 import argparse
 import hashlib
@@ -1114,10 +1125,21 @@ def train_from_config(args: argparse.Namespace) -> None:
         output_override=args.output_dir,
     )
     output_dir = output_cfg["output_dir"]
-    reuse_best_params = bool(
-        args.reuse_best_params or runtime_cfg["reuse_best_params"])
-    xgb_max_depth_max = runtime_cfg["xgb_max_depth_max"]
-    xgb_n_estimators_max = runtime_cfg["xgb_n_estimators_max"]
+    reuse_best_params = bool(
+        args.reuse_best_params or runtime_cfg["reuse_best_params"])
+    xgb_max_depth_max = runtime_cfg["xgb_max_depth_max"]
+    xgb_n_estimators_max = runtime_cfg["xgb_n_estimators_max"]
+    xgb_gpu_id = runtime_cfg["xgb_gpu_id"]
+    xgb_cleanup_per_fold = runtime_cfg["xgb_cleanup_per_fold"]
+    xgb_cleanup_synchronize = runtime_cfg["xgb_cleanup_synchronize"]
+    xgb_use_dmatrix = runtime_cfg["xgb_use_dmatrix"]
+    ft_cleanup_per_fold = runtime_cfg["ft_cleanup_per_fold"]
+    ft_cleanup_synchronize = runtime_cfg["ft_cleanup_synchronize"]
+    resn_cleanup_per_fold = runtime_cfg["resn_cleanup_per_fold"]
+    resn_cleanup_synchronize = runtime_cfg["resn_cleanup_synchronize"]
+    gnn_cleanup_per_fold = runtime_cfg["gnn_cleanup_per_fold"]
+    gnn_cleanup_synchronize = runtime_cfg["gnn_cleanup_synchronize"]
+    optuna_cleanup_synchronize = runtime_cfg["optuna_cleanup_synchronize"]
     optuna_storage = runtime_cfg["optuna_storage"]
     optuna_study_prefix = runtime_cfg["optuna_study_prefix"]
     best_params_files = runtime_cfg["best_params_files"]
@@ -1249,9 +1271,20 @@ def train_from_config(args: argparse.Namespace) -> None:
             "use_ft_ddp": use_ft_ddp,
             "use_gnn_ddp": use_gnn_ddp,
             "output_dir": output_dir,
-            "xgb_max_depth_max": xgb_max_depth_max,
-            "xgb_n_estimators_max": xgb_n_estimators_max,
-            "resn_weight_decay": cfg.get("resn_weight_decay"),
+            "xgb_max_depth_max": xgb_max_depth_max,
+            "xgb_n_estimators_max": xgb_n_estimators_max,
+            "xgb_gpu_id": xgb_gpu_id,
+            "xgb_cleanup_per_fold": xgb_cleanup_per_fold,
+            "xgb_cleanup_synchronize": xgb_cleanup_synchronize,
+            "xgb_use_dmatrix": xgb_use_dmatrix,
+            "ft_cleanup_per_fold": ft_cleanup_per_fold,
+            "ft_cleanup_synchronize": ft_cleanup_synchronize,
+            "resn_cleanup_per_fold": resn_cleanup_per_fold,
+            "resn_cleanup_synchronize": resn_cleanup_synchronize,
+            "gnn_cleanup_per_fold": gnn_cleanup_per_fold,
+            "gnn_cleanup_synchronize": gnn_cleanup_synchronize,
+            "optuna_cleanup_synchronize": optuna_cleanup_synchronize,
+            "resn_weight_decay": cfg.get("resn_weight_decay"),
             "final_ensemble": bool(cfg.get("final_ensemble", False)),
             "final_ensemble_k": int(cfg.get("final_ensemble_k", 3)),
             "final_refit": bool(cfg.get("final_refit", True)),