dragon-ml-toolbox 12.13.0__py3-none-any.whl → 13.0.0__py3-none-any.whl

{dragon_ml_toolbox-12.13.0.dist-info → dragon_ml_toolbox-13.0.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dragon-ml-toolbox
-Version: 12.13.0
+Version: 13.0.0
 Summary: A collection of tools for data science and machine learning projects.
 Author-email: "Karl L. Loza Vidaurre" <luigiloza@gmail.com>
 License-Expression: MIT

{dragon_ml_toolbox-12.13.0.dist-info → dragon_ml_toolbox-13.0.0.dist-info}/RECORD

@@ -1,19 +1,19 @@
-dragon_ml_toolbox-12.13.0.dist-info/licenses/LICENSE,sha256=L35WDmmLZNTlJvxF6Vy7Uy4SYNi6rCfWUqlTHpoRMoU,1081
-dragon_ml_toolbox-12.13.0.dist-info/licenses/LICENSE-THIRD-PARTY.md,sha256=iy2r_R7wjzsCbz_Q_jMsp_jfZ6oP8XW9QhwzRBH0mGY,1904
+dragon_ml_toolbox-13.0.0.dist-info/licenses/LICENSE,sha256=L35WDmmLZNTlJvxF6Vy7Uy4SYNi6rCfWUqlTHpoRMoU,1081
+dragon_ml_toolbox-13.0.0.dist-info/licenses/LICENSE-THIRD-PARTY.md,sha256=iy2r_R7wjzsCbz_Q_jMsp_jfZ6oP8XW9QhwzRBH0mGY,1904
 ml_tools/ETL_cleaning.py,sha256=2VBRllV8F-ZiPylPp8Az2gwn5ztgazN0BH5OKnRUhV0,20402
 ml_tools/ETL_engineering.py,sha256=KfYqgsxupAx6e_TxwO1LZXeu5mFkIhVXJrNjP3CzIZc,54927
 ml_tools/GUI_tools.py,sha256=Va6ig-dHULPVRwQYYtH3fvY5XPIoqRcJpRW8oXC55Hw,45413
 ml_tools/MICE_imputation.py,sha256=X273Qlgoqqg7KTmoKd75YDyAPB0UIbTzGP3xsCmRh3E,11717
-ml_tools/ML_callbacks.py,sha256=2ZazJjlbClP-ALc8q0ru2oalkugbhO3TFwPg4RFZpck,14056
+ml_tools/ML_callbacks.py,sha256=elD2Yr030sv_6gX_m9GVd6HTyrbmt34nFS8lrgS4HtM,15808
 ml_tools/ML_datasetmaster.py,sha256=kedCGneR3S2zui0_JFZN6TBL5e69XWkdpkE_QohyqSM,31433
-ml_tools/ML_evaluation.py,sha256=h7fAtk0lS4gTqQ46fiVjucTvFlX4rsufKnEtate6Nu0,18381
-ml_tools/ML_evaluation_multi.py,sha256=Kn9n5lfxo7A0TvgIDMx8UHZCvzTqv1ViezzwJBF-ypM,15970
-ml_tools/ML_inference.py,sha256=ymFvncFsU10PExq87xnEj541DKV5ck0nMuK8ToJHzVQ,23067
+ml_tools/ML_evaluation.py,sha256=3u5dOhS77gn3kAshKr2GwSa5xZBF0YM77ZkFevqNPvA,18528
+ml_tools/ML_evaluation_multi.py,sha256=L6Ub_uObXsI7ToVCF6DtmAFekHRcga5wWMOnRYRR-BY,16121
+ml_tools/ML_inference.py,sha256=yq2gdN6s_OUYC5ZLQrIJC5BA5H33q8UKODXwb-_0M2c,23549
 ml_tools/ML_models.py,sha256=G64NPhYZfYvHTIUwkIrMrNLgfDTKJwqdc8jwesPqB9E,28090
 ml_tools/ML_optimization.py,sha256=es3TlQbY7RYgJMZnznkjYGbUxFnAqzZxE_g3_qLK9Q8,22960
 ml_tools/ML_scaler.py,sha256=tw6onj9o8_kk3FQYb930HUzvv1zsFZe2YZJdF3LtHkU,7538
 ml_tools/ML_simple_optimization.py,sha256=W2mce1XFCuiOHTOjOsCNbETISHn5MwYlYsTIXH5hMMo,18177
-ml_tools/ML_trainer.py,sha256=UmCuKr_GzQGYqhEZ-kaRv9Buj44DsNyuOzmOM7Fw8N0,24569
+ml_tools/ML_trainer.py,sha256=9BP6JFClqGfe7GL-FGG3n5e-no9ssjEOLol7P6baGrI,29019
 ml_tools/ML_utilities.py,sha256=EnKpPTnJ2qjZmz7kvows4Uu5CfSA7ByRmI1v2-KarKw,9337
 ml_tools/PSO_optimization.py,sha256=fVHeemqilBS0zrGV25E5yKwDlGdd2ZKa18d8CZ6Q6Fk,22961
 ml_tools/RNN_forecast.py,sha256=Qa2KoZfdAvSjZ4yE78N4BFXtr3tTr0Gx7tQJZPotsh0,1967
@@ -29,13 +29,13 @@ ml_tools/ensemble_evaluation.py,sha256=FGHSe8LBI8_w8LjNeJWOcYQ1UK_mc6fVah8gmSvNV
 ml_tools/ensemble_inference.py,sha256=0yLmLNj45RVVoSCLH1ZYJG9IoAhTkWUqEZmLOQTFGTY,9348
 ml_tools/ensemble_learning.py,sha256=vsIED7nlheYI4w2SBzP6SC1AnNeMfn-2A1Gqw5EfxsM,21964
 ml_tools/handle_excel.py,sha256=pfdAPb9ywegFkM9T54bRssDOsX-K7rSeV0RaMz7lEAo,14006
-ml_tools/keys.py,sha256=FDpbS3Jb0pjrVvvp2_8nZi919mbob_-xwuy5OOtKM_A,1848
+ml_tools/keys.py,sha256=eJ4St5fl8uHstEGO1XVdP8G-ddwjOxV9zqG0D6W8pCI,2124
 ml_tools/math_utilities.py,sha256=PxoOrnuj6Ntp7_TJqyDWi0JX03WpAO5iaFNK2Oeq5I4,8800
 ml_tools/optimization_tools.py,sha256=P074YCuZzkqkONnAsM-Zb9DTX_i8cRkkJLpwAWz6CRw,13521
 ml_tools/path_manager.py,sha256=CyDU16pOKmC82jPubqJPT6EBt-u-3rGVbxyPIZCvDDY,18432
 ml_tools/serde.py,sha256=ll2mVC0sO2jIEdG3K6xMcgEN13N4YSb8VjviGvw_ers,4949
 ml_tools/utilities.py,sha256=OcAyV1tEcYAfOWlGjRgopsjDLxU3DcI5EynzvWV4q3A,15754
-dragon_ml_toolbox-12.13.0.dist-info/METADATA,sha256=p3-oOSqq1hhJj13KjIXeFnwBu3UTfBJu5mTDL9MCpdU,6167
-dragon_ml_toolbox-12.13.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-dragon_ml_toolbox-12.13.0.dist-info/top_level.txt,sha256=wm-oxax3ciyez6VoO4zsFd-gSok2VipYXnbg3TH9PtU,9
-dragon_ml_toolbox-12.13.0.dist-info/RECORD,,
+dragon_ml_toolbox-13.0.0.dist-info/METADATA,sha256=trY1fFyTTXLS6TZdrJXxq4_YMPjEZhKCilzCg6qFxzw,6166
+dragon_ml_toolbox-13.0.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+dragon_ml_toolbox-13.0.0.dist-info/top_level.txt,sha256=wm-oxax3ciyez6VoO4zsFd-gSok2VipYXnbg3TH9PtU,9
+dragon_ml_toolbox-13.0.0.dist-info/RECORD,,

ml_tools/ML_callbacks.py

@@ -5,7 +5,7 @@ from typing import Union, Literal, Optional
 from pathlib import Path
 
 from .path_manager import make_fullpath, sanitize_filename
-from .keys import PyTorchLogKeys
+from .keys import PyTorchLogKeys, PyTorchCheckpointKeys
 from ._logger import _LOGGER
 from ._script_info import _script_info
 
@@ -189,7 +189,7 @@ class EarlyStopping(Callback):
 
 class ModelCheckpoint(Callback):
     """
-    Saves the model weights to a directory with automated filename generation and rotation. 
+    Saves the model weights, optimizer state, LR scheduler state (if any), and epoch number to a directory with automated filename generation and rotation. 
     """
     def __init__(self, save_dir: Union[str,Path], checkpoint_name: Optional[str]=None, monitor: str = PyTorchLogKeys.VAL_LOSS,
                  save_best_only: bool = True, mode: Literal['auto', 'min', 'max']= 'auto', verbose: int = 0):
@@ -200,7 +200,7 @@ class ModelCheckpoint(Callback):
         Args:
             save_dir (str): Directory where checkpoint files will be saved.
             checkpoint_name (str| None): If None, the filename will include the epoch and score.
-            monitor (str): Metric to monitor for `save_best_only=True`.
+            monitor (str): Metric to monitor.
             save_best_only (bool): If true, save only the best model.
             mode (str): One of {'auto', 'min', 'max'}.
             verbose (int): Verbosity mode.
@@ -270,15 +270,29 @@ class ModelCheckpoint(Callback):
             if self.verbose > 0:
                 _LOGGER.info(f"Epoch {epoch}: {self.monitor} improved from {old_best_str} to {current:.4f}, saving model to {new_filepath}")
             
+            # Update best score *before* saving
+            self.best = current
+
+            # Create a comprehensive checkpoint dictionary
+            checkpoint_data = {
+                PyTorchCheckpointKeys.EPOCH: epoch,
+                PyTorchCheckpointKeys.MODEL_STATE: self.trainer.model.state_dict(), # type: ignore
+                PyTorchCheckpointKeys.OPTIMIZER_STATE: self.trainer.optimizer.state_dict(), # type: ignore
+                PyTorchCheckpointKeys.BEST_SCORE: self.best, 
+            }
+            
+            # Check for scheduler
+            if hasattr(self.trainer, 'scheduler') and self.trainer.scheduler is not None: # type: ignore
+                checkpoint_data[PyTorchCheckpointKeys.SCHEDULER_STATE] = self.trainer.scheduler.state_dict() # type: ignore
+            
             # Save the new best model
-            torch.save(self.trainer.model.state_dict(), new_filepath) # type: ignore
+            torch.save(checkpoint_data, new_filepath)
 
             # Delete the old best model file
             if self.last_best_filepath and self.last_best_filepath.exists():
                 self.last_best_filepath.unlink()
             
             # Update state
-            self.best = current
             self.last_best_filepath = new_filepath
 
     def _save_rolling_checkpoints(self, epoch, logs):
@@ -292,7 +306,19 @@ class ModelCheckpoint(Callback):
         
         if self.verbose > 0:
             _LOGGER.info(f'Epoch {epoch}: saving model to {filepath}')
-        torch.save(self.trainer.model.state_dict(), filepath) # type: ignore
+
+        # Create a comprehensive checkpoint dictionary
+        checkpoint_data = {
+            PyTorchCheckpointKeys.EPOCH: epoch,
+            PyTorchCheckpointKeys.MODEL_STATE: self.trainer.model.state_dict(), # type: ignore
+            PyTorchCheckpointKeys.OPTIMIZER_STATE: self.trainer.optimizer.state_dict(), # type: ignore
+            PyTorchCheckpointKeys.BEST_SCORE: self.best, # Save the current best score
+        }
+        
+        if hasattr(self.trainer, 'scheduler') and self.trainer.scheduler is not None: # type: ignore
+            checkpoint_data[PyTorchCheckpointKeys.SCHEDULER_STATE] = self.trainer.scheduler.state_dict() # type: ignore
+        
+        torch.save(checkpoint_data, filepath)
 
         self.saved_checkpoints.append(filepath)
 
@@ -309,19 +335,25 @@ class LRScheduler(Callback):
     """
     Callback to manage a PyTorch learning rate scheduler.
     """
-    def __init__(self, scheduler, monitor: Optional[str] = None):
+    def __init__(self, scheduler, monitor: Optional[str] = PyTorchLogKeys.VAL_LOSS):
         """
         This callback automatically calls the scheduler's `step()` method at the
         end of each epoch. It also logs a message when the learning rate changes.
 
         Args:
             scheduler: An initialized PyTorch learning rate scheduler.
-            monitor (str, optional): The metric to monitor for schedulers that require it, like `ReduceLROnPlateau`. Should match a key in the logs (e.g., 'val_loss').
+            monitor (str): The metric to monitor for schedulers that require it, like `ReduceLROnPlateau`. Should match a key in the logs (e.g., 'val_loss').
         """
         super().__init__()
         self.scheduler = scheduler
         self.monitor = monitor
         self.previous_lr = None
+        
+    def set_trainer(self, trainer):
+        """This is called by the Trainer to associate itself with the callback."""
+        super().set_trainer(trainer)
+        # Register the scheduler with the trainer so it can be added to the checkpoint
+        self.trainer.scheduler = self.scheduler # type: ignore
 
     def on_train_begin(self, logs=None):
         """Store the initial learning rate."""

ml_tools/ML_evaluation.py

@@ -19,6 +19,7 @@ import torch
 import shap
 from pathlib import Path
 from typing import Union, Optional, List, Literal
+import warnings
 
 from .path_manager import make_fullpath
 from ._logger import _LOGGER
@@ -298,8 +299,11 @@ def shap_summary_plot(model,
 
         background_data = background_data.to(device)
         instances_to_explain = instances_to_explain.to(device)
-
-        explainer = shap.DeepExplainer(model, background_data)
+        
+        with warnings.catch_warnings():
+            warnings.simplefilter("ignore", category=UserWarning)
+            explainer = shap.DeepExplainer(model, background_data)
+            
         # print("Calculating SHAP values with DeepExplainer...")
         shap_values = explainer.shap_values(instances_to_explain)
         instances_to_explain_np = instances_to_explain.cpu().numpy()

ml_tools/ML_evaluation_multi.py

@@ -20,6 +20,7 @@ from sklearn.metrics import (
 )
 from pathlib import Path
 from typing import Union, List, Literal
+import warnings
 
 from .path_manager import make_fullpath, sanitize_filename
 from ._logger import _LOGGER
@@ -273,9 +274,12 @@ def multi_target_shap_summary_plot(
 
         background_data = background_data.to(device)
         instances_to_explain = instances_to_explain.to(device)
-
-        explainer = shap.DeepExplainer(model, background_data)
-        print("Calculating SHAP values with DeepExplainer...")
+        
+        with warnings.catch_warnings():
+            warnings.simplefilter("ignore", category=UserWarning)
+            explainer = shap.DeepExplainer(model, background_data)
+            
+        # print("Calculating SHAP values with DeepExplainer...")
         # DeepExplainer returns a list of arrays for multi-output models
         shap_values_list = explainer.shap_values(instances_to_explain)
         instances_to_explain_np = instances_to_explain.cpu().numpy()
@@ -304,7 +308,7 @@ def multi_target_shap_summary_plot(
             return output.cpu().numpy() # Return full multi-output array
 
         explainer = shap.KernelExplainer(prediction_wrapper, background_summary)
-        print("Calculating SHAP values with KernelExplainer...")
+        # print("Calculating SHAP values with KernelExplainer...")
         # KernelExplainer also returns a list of arrays for multi-output models
         shap_values_list = explainer.shap_values(instances_to_explain_np, l1_reg="aic")
         # instances_to_explain_np is already set

ml_tools/ML_inference.py

@@ -9,7 +9,7 @@ from .ML_scaler import PytorchScaler
 from ._script_info import _script_info
 from ._logger import _LOGGER
 from .path_manager import make_fullpath
-from .keys import PyTorchInferenceKeys
+from .keys import PyTorchInferenceKeys, PyTorchCheckpointKeys
 
 
 __all__ = [
@@ -56,11 +56,21 @@ class _BaseInferenceHandler(ABC):
         model_p = make_fullpath(state_dict, enforce="file")
 
         try:
-            # Load the state dictionary and apply it to the model structure
-            self.model.load_state_dict(torch.load(model_p, map_location=self.device))
+            # Load whatever is in the file
+            loaded_data = torch.load(model_p, map_location=self.device)
+
+            # Check if it's the new checkpoint dictionary or an old weights-only file
+            if isinstance(loaded_data, dict) and PyTorchCheckpointKeys.MODEL_STATE in loaded_data:
+                # It's a new training checkpoint, extract the weights
+                self.model.load_state_dict(loaded_data[PyTorchCheckpointKeys.MODEL_STATE])
+            else:
+                # It's an old-style file (or just a state_dict), load it directly
+                self.model.load_state_dict(loaded_data)
+            
+            _LOGGER.info(f"Model state loaded from '{model_p.name}'.")
+                
             self.model.to(self.device)
             self.model.eval()  # Set the model to evaluation mode
-            _LOGGER.info(f"Model state loaded from '{model_p.name}' and set to evaluation mode.")
         except Exception as e:
             _LOGGER.error(f"Failed to load model state from '{model_p}': {e}")
             raise

ml_tools/ML_trainer.py

@@ -5,12 +5,13 @@ import torch
 from torch import nn
 import numpy as np
 
-from .ML_callbacks import Callback, History, TqdmProgressBar
+from .ML_callbacks import Callback, History, TqdmProgressBar, ModelCheckpoint
 from .ML_evaluation import classification_metrics, regression_metrics, plot_losses, shap_summary_plot, plot_attention_importance
 from .ML_evaluation_multi import multi_target_regression_metrics, multi_label_classification_metrics, multi_target_shap_summary_plot
 from ._script_info import _script_info
-from .keys import PyTorchLogKeys
+from .keys import PyTorchLogKeys, PyTorchCheckpointKeys
 from ._logger import _LOGGER
+from .path_manager import make_fullpath
 
 
 __all__ = [
@@ -55,6 +56,7 @@ class MLTrainer:
         self.kind = kind
         self.criterion = criterion
         self.optimizer = optimizer
+        self.scheduler = None
         self.device = self._validate_device(device)
         self.dataloader_workers = dataloader_workers
         
@@ -70,6 +72,7 @@ class MLTrainer:
         self.history = {}
         self.epoch = 0
         self.epochs = 0 # Total epochs for the fit run
+        self.start_epoch = 1
         self.stop_training = False
 
     def _validate_device(self, device: str) -> torch.device:
@@ -109,8 +112,66 @@ class MLTrainer:
             num_workers=loader_workers, 
             pin_memory=("cuda" in self.device.type)
         )
+        
+    def _load_checkpoint(self, path: Union[str, Path]):
+        """Loads a training checkpoint to resume training."""
+        p = make_fullpath(path, enforce="file")
+        _LOGGER.info(f"Loading checkpoint from '{p.name}' to resume training...")
+        
+        try:
+            checkpoint = torch.load(p, map_location=self.device)
+            
+            if PyTorchCheckpointKeys.MODEL_STATE not in checkpoint or PyTorchCheckpointKeys.OPTIMIZER_STATE not in checkpoint:
+                _LOGGER.error(f"Checkpoint file '{p.name}' is invalid. Missing 'model_state_dict' or 'optimizer_state_dict'.")
+                raise KeyError()
 
-    def fit(self, epochs: int = 10, batch_size: int = 10, shuffle: bool = True):
+            self.model.load_state_dict(checkpoint[PyTorchCheckpointKeys.MODEL_STATE])
+            self.optimizer.load_state_dict(checkpoint[PyTorchCheckpointKeys.OPTIMIZER_STATE])
+            self.start_epoch = checkpoint.get(PyTorchCheckpointKeys.EPOCH, 0) + 1 # Resume on the *next* epoch
+            
+            # --- Scheduler State Loading Logic ---
+            scheduler_state_exists = PyTorchCheckpointKeys.SCHEDULER_STATE in checkpoint
+            scheduler_object_exists = self.scheduler is not None
+
+            if scheduler_object_exists and scheduler_state_exists:
+                # Case 1: Both exist. Attempt to load.
+                try:
+                    self.scheduler.load_state_dict(checkpoint[PyTorchCheckpointKeys.SCHEDULER_STATE]) # type: ignore
+                    scheduler_name = self.scheduler.__class__.__name__
+                    _LOGGER.info(f"Restored LR scheduler state for: {scheduler_name}")
+                except Exception as e:
+                    # Loading failed, likely a mismatch
+                    scheduler_name = self.scheduler.__class__.__name__
+                    _LOGGER.error(f"Failed to load scheduler state for '{scheduler_name}'. A different scheduler type might have been used.")
+                    raise e
+
+            elif scheduler_object_exists and not scheduler_state_exists:
+                # Case 2: Scheduler provided, but no state in checkpoint.
+                scheduler_name = self.scheduler.__class__.__name__
+                _LOGGER.warning(f"'{scheduler_name}' was provided, but no scheduler state was found in the checkpoint. The scheduler will start from its initial state.")
+            
+            elif not scheduler_object_exists and scheduler_state_exists:
+                # Case 3: State in checkpoint, but no scheduler provided.
+                _LOGGER.error("Checkpoint contains an LR scheduler state, but no LRScheduler callback was provided.")
+                raise ValueError()
+            
+            # Restore callback states
+            for cb in self.callbacks:
+                if isinstance(cb, ModelCheckpoint) and PyTorchCheckpointKeys.BEST_SCORE in checkpoint:
+                    cb.best = checkpoint[PyTorchCheckpointKeys.BEST_SCORE]
+                    _LOGGER.info(f"Restored {cb.__class__.__name__} 'best' score to: {cb.best:.4f}")
+            
+            _LOGGER.info(f"Checkpoint loaded. Resuming training from epoch {self.start_epoch}.")
+            
+        except Exception as e:
+            _LOGGER.error(f"Failed to load checkpoint from '{p}': {e}")
+            raise
+
+    def fit(self, 
+            epochs: int = 10, 
+            batch_size: int = 10, 
+            shuffle: bool = True,
+            resume_from_checkpoint: Optional[Union[str, Path]] = None):
         """
         Starts the training-validation process of the model.
         
@@ -120,6 +181,7 @@ class MLTrainer:
             epochs (int): The total number of epochs to train for.
             batch_size (int): The number of samples per batch.
             shuffle (bool): Whether to shuffle the training data at each epoch.
+            resume_from_checkpoint (str | Path | None): Optional path to a checkpoint to resume training.
             
         Note:
             For regression tasks using `nn.MSELoss` or `nn.L1Loss`, the trainer
@@ -132,15 +194,18 @@ class MLTrainer:
         self._create_dataloaders(batch_size, shuffle)
         self.model.to(self.device)
         
+        if resume_from_checkpoint:
+            self._load_checkpoint(resume_from_checkpoint)
+        
         # Reset stop_training flag on the trainer
         self.stop_training = False
 
-        self.callbacks_hook('on_train_begin')
+        self._callbacks_hook('on_train_begin')
         
-        for epoch in range(1, self.epochs + 1):
+        for epoch in range(self.start_epoch, self.epochs + 1):
             self.epoch = epoch
             epoch_logs = {}
-            self.callbacks_hook('on_epoch_begin', epoch, logs=epoch_logs)
+            self._callbacks_hook('on_epoch_begin', epoch, logs=epoch_logs)
 
             train_logs = self._train_step()
             epoch_logs.update(train_logs)
@@ -148,13 +213,13 @@ class MLTrainer:
             val_logs = self._validation_step()
             epoch_logs.update(val_logs)
             
-            self.callbacks_hook('on_epoch_end', epoch, logs=epoch_logs)
+            self._callbacks_hook('on_epoch_end', epoch, logs=epoch_logs)
             
             # Check the early stopping flag
             if self.stop_training:
                 break
 
-        self.callbacks_hook('on_train_end')
+        self._callbacks_hook('on_train_end')
         return self.history
     
     def _train_step(self):
@@ -166,7 +231,7 @@ class MLTrainer:
                 PyTorchLogKeys.BATCH_INDEX: batch_idx, 
                 PyTorchLogKeys.BATCH_SIZE: features.size(0)
             }
-            self.callbacks_hook('on_batch_begin', batch_idx, logs=batch_logs)
+            self._callbacks_hook('on_batch_begin', batch_idx, logs=batch_logs)
 
             features, target = features.to(self.device), target.to(self.device)
             self.optimizer.zero_grad()
@@ -188,7 +253,7 @@ class MLTrainer:
             
             # Add the batch loss to the logs and call the end-of-batch hook
             batch_logs[PyTorchLogKeys.BATCH_LOSS] = batch_loss
-            self.callbacks_hook('on_batch_end', batch_idx, logs=batch_logs)
+            self._callbacks_hook('on_batch_end', batch_idx, logs=batch_logs)
 
         return {PyTorchLogKeys.TRAIN_LOSS: running_loss / len(self.train_loader.dataset)} # type: ignore
 
@@ -538,11 +603,33 @@ class MLTrainer:
         else:
             _LOGGER.error("No attention weights were collected from the model.")
     
-    def callbacks_hook(self, method_name: str, *args, **kwargs):
+    def _callbacks_hook(self, method_name: str, *args, **kwargs):
         """Calls the specified method on all callbacks."""
         for callback in self.callbacks:
             method = getattr(callback, method_name)
             method(*args, **kwargs)
+            
+    def to_cpu(self):
+        """
+        Moves the model to the CPU and updates the trainer's device setting.
+        
+        This is useful for running operations that require the CPU.
+        """
+        self.device = torch.device('cpu')
+        self.model.to(self.device)
+        _LOGGER.info("Trainer and model moved to CPU.")
+    
+    def to_device(self, device: str):
+        """
+        Moves the model to the specified device and updates the trainer's device setting.
+
+        Args:
+            device (str): The target device (e.g., 'cuda', 'mps', 'cpu').
+        """
+        self.device = self._validate_device(device)
+        self.model.to(self.device)
+        _LOGGER.info(f"Trainer and model moved to {self.device}.")
+    
 
 def info():
     _script_info(__all__)

ml_tools/keys.py

@@ -68,6 +68,15 @@ class SHAPKeys:
     SAVENAME = "shap_summary"
 
 
+class PyTorchCheckpointKeys:
+    """Keys for saving/loading a training checkpoint dictionary."""
+    MODEL_STATE = "model_state_dict"
+    OPTIMIZER_STATE = "optimizer_state_dict"
+    SCHEDULER_STATE = "scheduler_state_dict"
+    EPOCH = "epoch"
+    BEST_SCORE = "best_score"
+
+
 class _OneHotOtherPlaceholder:
     """Used internally by GUI_tools."""
     OTHER_GUI = "OTHER"