From 277163871264e63c59bc21a19451a8fca8e06b0e Mon Sep 17 00:00:00 2001
From: Jeffrey Lim <Jeffrey-Lim@outlook.com>
Date: Fri, 8 Mar 2024 09:48:27 +0100
Subject: [PATCH 1/9] Fix pyproject.toml

---
 pyproject.toml   | 12 ++++++------
 requirements.txt |  4 ++--
 2 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/pyproject.toml b/pyproject.toml
index 6e54b23..8f59573 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -17,20 +17,20 @@ classifiers = [
 ]
 dependencies= [
     # Machine Learning Libraries
-    "numpy>=1.24.4",
+    "numpy~=1.26.4",
 
     # Parallel Processing Libraries
-    "dask>=2023.12.0",
+    "dask~=2023.12.0",
 
     # Data Processing Libraries
-    "pandas>=1.3.3",
+    "pandas~=1.3.3",
 
     # Parquet
-    "pyarrow>=6.0.0",
+    "pyarrow~=6.0.0",
 
     # PyTorch
-    "torch>=2.1.2+cu118",
+    "torch~=2.2.1",
 
     # Agogos
-    "agogos>=0.2.0",
+    "agogos~=0.2.0",
 ]
diff --git a/requirements.txt b/requirements.txt
index bfea61f..b7cbcd5 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,4 +1,4 @@
-agogos>=0.2.0
+agogos==0.2.0
 annotated-types==0.6.0
 certifi==2024.2.2
 charset-normalizer==3.3.2
@@ -23,7 +23,7 @@ pandas==2.2.1
 partd==1.4.1
 pillow==10.2.0
 pluggy==1.4.0
-pyarrow>=6.0.0
+pyarrow==6.0.0
 pytest==8.0.2
 pytest-cov==4.1.0
 PyYAML==6.0.1

From 50d29f8a6476e1dced56ef7af22381c1f0f99282 Mon Sep 17 00:00:00 2001
From: schobbejak <jmvanselm@gmail.com>
Date: Fri, 8 Mar 2024 10:29:23 +0100
Subject: [PATCH 2/9] Add log_to_terminal for loading cache

---
 epochalyst/pipeline/model/training/training.py                 | 3 +++
 epochalyst/pipeline/model/training/training_block.py           | 3 +++
 epochalyst/pipeline/model/transformation/transformation.py     | 3 +++
 .../pipeline/model/transformation/transformation_block.py      | 3 +++
 4 files changed, 12 insertions(+)

diff --git a/epochalyst/pipeline/model/training/training.py b/epochalyst/pipeline/model/training/training.py
index 8c625fd..bc79946 100644
--- a/epochalyst/pipeline/model/training/training.py
+++ b/epochalyst/pipeline/model/training/training.py
@@ -25,6 +25,9 @@ def train(
             and self._cache_exists(name=self.get_hash() + "x", cache_args=cache_args)
             and self._cache_exists(name=self.get_hash() + "y", cache_args=cache_args)
         ):
+            self.log_to_terminal(
+                f"Cache exists for training pipeline with hash: {self.get_hash()}. Using the cache."
+            )
             x = self._get_cache(name=self.get_hash() + "x", cache_args=cache_args)
             y = self._get_cache(name=self.get_hash() + "y", cache_args=cache_args)
             return x, y
diff --git a/epochalyst/pipeline/model/training/training_block.py b/epochalyst/pipeline/model/training/training_block.py
index 3252c8e..bd4872c 100644
--- a/epochalyst/pipeline/model/training/training_block.py
+++ b/epochalyst/pipeline/model/training/training_block.py
@@ -70,6 +70,9 @@ def train(
             and self._cache_exists(name=self.get_hash() + "x", cache_args=cache_args)
             and self._cache_exists(name=self.get_hash() + "y", cache_args=cache_args)
         ):
+            self.log_to_terminal(
+                f"Cache exists for {self.__class__} with hash: {self.get_hash()}. Using the cache."
+            )
             x = self._get_cache(name=self.get_hash() + "x", cache_args=cache_args)
             y = self._get_cache(name=self.get_hash() + "y", cache_args=cache_args)
             return x, y
diff --git a/epochalyst/pipeline/model/transformation/transformation.py b/epochalyst/pipeline/model/transformation/transformation.py
index 77dcddb..689eed8 100644
--- a/epochalyst/pipeline/model/transformation/transformation.py
+++ b/epochalyst/pipeline/model/transformation/transformation.py
@@ -26,6 +26,9 @@ def transform(
         """
 
         if cache_args and self._cache_exists(self.get_hash(), cache_args):
+            self.log_to_terminal(
+                f"Cache exists for {self.title} with hash: {self.get_hash()}. Using the cache."
+            )
             return self._get_cache(self.get_hash(), cache_args)
 
         if self.steps:
diff --git a/epochalyst/pipeline/model/transformation/transformation_block.py b/epochalyst/pipeline/model/transformation/transformation_block.py
index bb11dfd..0254fe2 100644
--- a/epochalyst/pipeline/model/transformation/transformation_block.py
+++ b/epochalyst/pipeline/model/transformation/transformation_block.py
@@ -79,6 +79,9 @@ def transform(
         if cache_args and self._cache_exists(
             name=self.get_hash(), cache_args=cache_args
         ):
+            self.log_to_terminal(
+                f"Cache exists for {self.__class__} with hash: {self.get_hash()}. Using the cache."
+            )
             return self._get_cache(name=self.get_hash(), cache_args=cache_args)
 
         data = self.custom_transform(data, **transform_args)

From 4d55f7f1e9c039486e817764c1de5d35f6329820 Mon Sep 17 00:00:00 2001
From: schobbejak <jmvanselm@gmail.com>
Date: Fri, 8 Mar 2024 10:41:46 +0100
Subject: [PATCH 3/9] Add log_to_terminal method to test classes

---
 .gitignore                                                     | 2 +-
 tests/pipeline/model/training/test_training.py                 | 3 +++
 tests/pipeline/model/training/test_training_block.py           | 3 +++
 tests/pipeline/model/transformation/test_transformation.py     | 3 +++
 .../pipeline/model/transformation/test_transformation_block.py | 3 +++
 5 files changed, 13 insertions(+), 1 deletion(-)

diff --git a/.gitignore b/.gitignore
index 1aa502a..4510c5e 100644
--- a/.gitignore
+++ b/.gitignore
@@ -216,4 +216,4 @@ plots/images/train
 coverage_re/
 
 # Ignore vscode settings
-.vscode/
\ No newline at end of file
+.vscode/
diff --git a/tests/pipeline/model/training/test_training.py b/tests/pipeline/model/training/test_training.py
index feee1e7..42d264a 100644
--- a/tests/pipeline/model/training/test_training.py
+++ b/tests/pipeline/model/training/test_training.py
@@ -43,6 +43,9 @@ class CustomTrainingPipeline(TrainingPipeline):
             def log_to_debug(self, message: str) -> None:
                 return None
 
+            def log_to_terminal(self, message: str) -> None:
+                return None
+
         t1 = TestTrainingBlock()
         t2 = TestTrainingBlock()
 
diff --git a/tests/pipeline/model/training/test_training_block.py b/tests/pipeline/model/training/test_training_block.py
index e4ab074..4d42be2 100644
--- a/tests/pipeline/model/training/test_training_block.py
+++ b/tests/pipeline/model/training/test_training_block.py
@@ -43,6 +43,9 @@ def custom_predict(self, x: int) -> int:
             def log_to_debug(self, message: str) -> None:
                 return None
 
+            def log_to_terminal(self, message: str) -> None:
+                return None
+
         tb = TestTrainingBlockImpl()
         cache_args = {
             "output_data_type": "numpy_array",
diff --git a/tests/pipeline/model/transformation/test_transformation.py b/tests/pipeline/model/transformation/test_transformation.py
index b46fc19..7b9a9bf 100644
--- a/tests/pipeline/model/transformation/test_transformation.py
+++ b/tests/pipeline/model/transformation/test_transformation.py
@@ -36,6 +36,9 @@ class CustomTransformationPipeline(TransformationPipeline):
             def log_to_debug(self, message: str) -> None:
                 return None
 
+            def log_to_terminal(self, message: str) -> None:
+                return None
+
         t1 = TestTransformationBlock()
         t2 = TestTransformationBlock()
 
diff --git a/tests/pipeline/model/transformation/test_transformation_block.py b/tests/pipeline/model/transformation/test_transformation_block.py
index fa580b3..7a3cb2e 100644
--- a/tests/pipeline/model/transformation/test_transformation_block.py
+++ b/tests/pipeline/model/transformation/test_transformation_block.py
@@ -63,6 +63,9 @@ def custom_transform(self, data: np.ndarray[int], **transform_args) -> int:
             def log_to_debug(self, message: str) -> None:
                 return None
 
+            def log_to_terminal(self, message: str) -> None:
+                return None
+
         tb = TestTransformationBlockImpl()
         cache_args = {
             "output_data_type": "numpy_array",

From f65d0adc10117f5760a6a1f3e742792e0cda60a6 Mon Sep 17 00:00:00 2001
From: schobbejak <jmvanselm@gmail.com>
Date: Fri, 8 Mar 2024 10:43:57 +0100
Subject: [PATCH 4/9] Upgrade protocol of pickle dump

---
 epochalyst/_core/_caching/_cacher.py | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/epochalyst/_core/_caching/_cacher.py b/epochalyst/_core/_caching/_cacher.py
index a772ea1..2d38313 100644
--- a/epochalyst/_core/_caching/_cacher.py
+++ b/epochalyst/_core/_caching/_cacher.py
@@ -287,7 +287,11 @@ def _store_cache(
         elif storage_type == ".pkl":
             # Store the pickle file
             self.log_to_debug(f"Storing pickle file to {storage_path + name + '.pkl'}")
-            pickle.dump(data, open(storage_path + name + ".pkl", "wb"))
+            pickle.dump(
+                data,
+                open(storage_path + name + ".pkl", "wb"),
+                protocol=pickle.HIGHEST_PROTOCOL,
+            )
         else:
             self.log_to_debug(f"Invalid storage type: {storage_type}")
             raise ValueError(

From e08f1c3f573c4546457d4b5eeb0ec57f4b8e2b5c Mon Sep 17 00:00:00 2001
From: schobbejak <jmvanselm@gmail.com>
Date: Fri, 8 Mar 2024 11:54:46 +0100
Subject: [PATCH 5/9] Update documentation for TorchTrainer and
 TransformationPipeline

---
 .../pipeline/model/training/torch_trainer.py  | 90 ++++++++++++++-----
 .../model/transformation/transformation.py    | 47 +++++++++-
 2 files changed, 114 insertions(+), 23 deletions(-)

diff --git a/epochalyst/pipeline/model/training/torch_trainer.py b/epochalyst/pipeline/model/training/torch_trainer.py
index 34d510a..356e51b 100644
--- a/epochalyst/pipeline/model/training/torch_trainer.py
+++ b/epochalyst/pipeline/model/training/torch_trainer.py
@@ -24,30 +24,49 @@
 class TorchTrainer(TrainingBlock):
     """Abstract class for torch trainers, override necessary functions for custom implementation.
 
-    To use this block, you must inherit from it and implement the following methods:
-    - `log_to_terminal`
-    - `log_to_debug`
-    - `log_to_warning`
-    - `log_to_external`
-    - `external_define_metric`
-
-    :param model: The model to train.
-    :param optimizer: Optimizer to use for training.
-    :param criterion: Criterion to use for training.
-    :param scheduler: Scheduler to use for training.
-    :param epochs: Number of epochs
-    :param batch_size: Batch size
-    :param patience: Patience for early stopping
-    :param test_size: Relative size of the test set
-
+    ### Parameters:
+    - `model` (nn.Module): The model to train.
+    - `optimizer` (functools.partial[Optimizer]): Optimizer to use for training.
+    - `criterion` (nn.Module): Criterion to use for training.
+    - `scheduler` (Callable[[Optimizer], LRScheduler] | None): Scheduler to use for training.
+    - `epochs` (int): Number of epochs
+    - `batch_size` (int): Batch size
+    - `patience` (int): Patience for early stopping
+    - `test_size` (float): Relative size of the test set
+
+    ### Methods:
     ```python
-    ------------------------------
+    @abstractmethod
+    def log_to_terminal(self, message: str) -> None: 
+        # Logs to terminal if implemented
+
+    @abstractmethod
+    def log_to_debug(self, message: str) -> None: 
+        # Logs to debugger if implemented
+
+    @abstractmethod
+    def log_to_warning(self, message: str) -> None: 
+        # Logs to warning if implemented
+
+    @abstractmethod
+    def log_to_external(self, message: dict[str, Any], **kwargs: Any) -> None: 
+        # Logs to external site
+
+    @abstractmethod
+    def external_define_metric(self, metric: str, metric_type: str) -> None: 
+        # Defines an external metric
 
-    def train(x: npt.NDArray[np.float32], y: npt.NDArray[np.float32], train_indices: list[int], test_indices: list[int], cache_size: int = -1, save_model: bool = True) -> tuple[npt.NDArray[np.float32], npt.NDArray[np.float32]]:
-        # Train the model.
+    def train(self, x: Any, y: Any, cache_args: dict[str, Any] = {}, **train_args: Any) -> tuple[Any, Any]: 
+        # Applies caching and calls custom_train, overridding removes caching
 
-    def predict(x: npt.NDArray[np.float32]) -> npt.NDArray[np.float32]:
-        # Predict on the test data.
+    def predict(self, x: Any, cache_args: dict[str, Any] = {}, **pred_args: Any) -> Any: 
+        # Applies caching and calls custom_predict, overridding removes caching
+
+    def custom_train(self, x: Any, y: Any, **train_args: Any) -> tuple[Any, Any]: 
+        # Implements torch training. If you are going to override this method and not use any other functionality, inherit from TrainingBlock. 
+    
+    def custom_predict(self, x: Any, **pred_args: Any) -> Any: 
+        # Implements torch prediction. If you are going to override this method and not use any other functionality, inherit from TrainingBlock.
 
     def predict_on_loader(loader: DataLoader[tuple[Tensor, ...]]) -> npt.NDArray[np.float32]:
         # Predict using a dataloader.
@@ -64,6 +83,35 @@ def create_dataloaders(train_dataset: Dataset[tuple[Tensor, ...]], test_dataset:
     def update_model_directory(model_directory: str) -> None:
         # Update the model directory for caching (default: tm).
     ```
+
+    ### Usage:
+    ```python
+    from epochalyst.pipeline.model.training.torch_trainer import TorchTrainer
+    from torch import nn
+    from torch.optim import Adam
+    from torch.optim.lr_scheduler import StepLR
+    from torch.nn import MSELoss
+
+    class MyTorchTrainer(TorchTrainer):
+    
+        def log_to_terminal(self, message: str) -> None:
+        
+        ....
+
+    model = nn.Sequential(nn.Linear(1, 1))
+    optimizer = functools.partial(Adam, lr=0.01)
+    criterion = MSELoss()
+    scheduler = functools.partial(StepLR, step_size=1, gamma=0.1)
+    epochs = 10
+    batch_size = 32
+    patience = 5
+    test_size = 0.2
+
+    trainer = MyTorchTrainer(model=model, optimizer=optimizer, criterion=criterion, scheduler=scheduler, epochs=epochs, batch_size=batch_size, patience=patience, test_size=test_size)
+
+    x, y = trainer.train(x, y)
+    x = trainer.predict(x)
+    ```
     """
 
     model: nn.Module
diff --git a/epochalyst/pipeline/model/transformation/transformation.py b/epochalyst/pipeline/model/transformation/transformation.py
index 77dcddb..c03e608 100644
--- a/epochalyst/pipeline/model/transformation/transformation.py
+++ b/epochalyst/pipeline/model/transformation/transformation.py
@@ -10,8 +10,51 @@
 class TransformationPipeline(TransformingSystem, _Cacher, _Logger):
     """TransformationPipeline is the class used to create the pipeline for the transformation of the data.
 
-    :param steps: The steps to transform the data. Can be a list of Transformers, TransformationPipelines, or a combination of both.
-    :param title: The title of the pipeline. (Default: "Transformation Pipeline")
+    ### Parameters:
+    - `steps` (List[Union[Transformer, TransformationPipeline]]): The steps to transform the data. Can be a list of Transformers, TransformationPipelines, or a combination of both.
+    - `title` (str): The title of the pipeline. (Default: "Transformation Pipeline")
+
+    ### Methods:
+    ```python
+    @abstractmethod
+    def log_to_terminal(self, message: str) -> None: # Log the message to the terminal.
+    
+    @abstractmethod
+    def log_to_debug(self, message: str) -> None: # Log the message to the debug file.
+    
+    @abstractmethod
+    def log_to_warning(self, message: str) -> None: # Log the message to the warning file.
+    
+    @abstractmethod
+    def log_to_external(self, message: str) -> None: # Log the message to an external file.
+    
+    @abstractmethod
+    def log_section_separator(self, title: str) -> None: # Log a section separator to the terminal.
+    
+    @abstractmethod
+    def external_define_metric(self, metric: str, metric_type: str) -> None: # Define a metric to be logged to an external file.
+    
+    def transform(self, data: Any, cache_args: dict[str, Any] = {}, **transform_args: Any) -> Any: # Transform the input data.
+    
+    def get_hash(self) -> str: # Get the hash of the pipeline.
+    ```
+
+    ### Usage:  
+    ```python
+    from epochalyst.pipeline.model.transformation import TransformationPipeline
+
+    class MyTransformationPipeline(TransformationPipeline):
+        def log_to_terminal(self, message: str) -> None:
+            print(message)
+        
+        ....
+
+    step1 = MyTransformer1()
+    step2 = MyTransformer2()
+    pipeline = MyTransformationPipeline(steps=[step1, step2])
+
+    data = pipeline.transform(data)
+    ```
     """
 
     title: str = "Transformation Pipeline"  # The title of the pipeline since transformation pipeline can be used for multiple purposes. (Feature, Label, etc.)

From aa806a5258bf1a653e2d8b24d17ea8880e57b6c7 Mon Sep 17 00:00:00 2001
From: schobbejak <jmvanselm@gmail.com>
Date: Fri, 8 Mar 2024 11:55:07 +0100
Subject: [PATCH 6/9] Pre-commit fixes

---
 .gitignore                                    |  2 +-
 .../pipeline/model/training/torch_trainer.py  | 26 +++++++++----------
 .../model/transformation/transformation.py    | 18 ++++++-------
 3 files changed, 23 insertions(+), 23 deletions(-)

diff --git a/.gitignore b/.gitignore
index 1aa502a..4510c5e 100644
--- a/.gitignore
+++ b/.gitignore
@@ -216,4 +216,4 @@ plots/images/train
 coverage_re/
 
 # Ignore vscode settings
-.vscode/
\ No newline at end of file
+.vscode/
diff --git a/epochalyst/pipeline/model/training/torch_trainer.py b/epochalyst/pipeline/model/training/torch_trainer.py
index 356e51b..1268048 100644
--- a/epochalyst/pipeline/model/training/torch_trainer.py
+++ b/epochalyst/pipeline/model/training/torch_trainer.py
@@ -37,35 +37,35 @@ class TorchTrainer(TrainingBlock):
     ### Methods:
     ```python
     @abstractmethod
-    def log_to_terminal(self, message: str) -> None: 
+    def log_to_terminal(self, message: str) -> None:
         # Logs to terminal if implemented
 
     @abstractmethod
-    def log_to_debug(self, message: str) -> None: 
+    def log_to_debug(self, message: str) -> None:
         # Logs to debugger if implemented
 
     @abstractmethod
-    def log_to_warning(self, message: str) -> None: 
+    def log_to_warning(self, message: str) -> None:
         # Logs to warning if implemented
 
     @abstractmethod
-    def log_to_external(self, message: dict[str, Any], **kwargs: Any) -> None: 
+    def log_to_external(self, message: dict[str, Any], **kwargs: Any) -> None:
         # Logs to external site
 
     @abstractmethod
-    def external_define_metric(self, metric: str, metric_type: str) -> None: 
+    def external_define_metric(self, metric: str, metric_type: str) -> None:
         # Defines an external metric
 
-    def train(self, x: Any, y: Any, cache_args: dict[str, Any] = {}, **train_args: Any) -> tuple[Any, Any]: 
+    def train(self, x: Any, y: Any, cache_args: dict[str, Any] = {}, **train_args: Any) -> tuple[Any, Any]:
         # Applies caching and calls custom_train, overridding removes caching
 
-    def predict(self, x: Any, cache_args: dict[str, Any] = {}, **pred_args: Any) -> Any: 
+    def predict(self, x: Any, cache_args: dict[str, Any] = {}, **pred_args: Any) -> Any:
         # Applies caching and calls custom_predict, overridding removes caching
 
-    def custom_train(self, x: Any, y: Any, **train_args: Any) -> tuple[Any, Any]: 
-        # Implements torch training. If you are going to override this method and not use any other functionality, inherit from TrainingBlock. 
-    
-    def custom_predict(self, x: Any, **pred_args: Any) -> Any: 
+    def custom_train(self, x: Any, y: Any, **train_args: Any) -> tuple[Any, Any]:
+        # Implements torch training. If you are going to override this method and not use any other functionality, inherit from TrainingBlock.
+
+    def custom_predict(self, x: Any, **pred_args: Any) -> Any:
         # Implements torch prediction. If you are going to override this method and not use any other functionality, inherit from TrainingBlock.
 
     def predict_on_loader(loader: DataLoader[tuple[Tensor, ...]]) -> npt.NDArray[np.float32]:
@@ -93,9 +93,9 @@ def update_model_directory(model_directory: str) -> None:
     from torch.nn import MSELoss
 
     class MyTorchTrainer(TorchTrainer):
-    
+
         def log_to_terminal(self, message: str) -> None:
-        
+
         ....
 
     model = nn.Sequential(nn.Linear(1, 1))
diff --git a/epochalyst/pipeline/model/transformation/transformation.py b/epochalyst/pipeline/model/transformation/transformation.py
index c03e608..c20c540 100644
--- a/epochalyst/pipeline/model/transformation/transformation.py
+++ b/epochalyst/pipeline/model/transformation/transformation.py
@@ -18,35 +18,35 @@ class TransformationPipeline(TransformingSystem, _Cacher, _Logger):
     ```python
     @abstractmethod
     def log_to_terminal(self, message: str) -> None: # Log the message to the terminal.
-    
+
     @abstractmethod
     def log_to_debug(self, message: str) -> None: # Log the message to the debug file.
-    
+
     @abstractmethod
     def log_to_warning(self, message: str) -> None: # Log the message to the warning file.
-    
+
     @abstractmethod
     def log_to_external(self, message: str) -> None: # Log the message to an external file.
-    
+
     @abstractmethod
     def log_section_separator(self, title: str) -> None: # Log a section separator to the terminal.
-    
+
     @abstractmethod
     def external_define_metric(self, metric: str, metric_type: str) -> None: # Define a metric to be logged to an external file.
-    
+
     def transform(self, data: Any, cache_args: dict[str, Any] = {}, **transform_args: Any) -> Any: # Transform the input data.
-    
+
     def get_hash(self) -> str: # Get the hash of the pipeline.
     ```
 
-    ### Usage:  
+    ### Usage:
     ```python
     from epochalyst.pipeline.model.transformation import TransformationPipeline
 
     class MyTransformationPipeline(TransformationPipeline):
         def log_to_terminal(self, message: str) -> None:
             print(message)
-        
+
         ....
 
     step1 = MyTransformer1()

From bde7400df6ecc9e637a9ad39c3539c12f7c04002 Mon Sep 17 00:00:00 2001
From: schobbejak <jmvanselm@gmail.com>
Date: Fri, 8 Mar 2024 12:22:22 +0100
Subject: [PATCH 7/9] Update version in pyproject.toml

---
 pyproject.toml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/pyproject.toml b/pyproject.toml
index f6ad3d5..8f59573 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "epochalyst"
-version = "0.1.2"
+version = "0.1.3"
 authors = [
     { name = "Jasper van Selm", email = "jmvanselm@gmail.com" },
     { name = "Ariel Ebersberger", email = "arielebersberger@gmail.com" },

From 75fb9788f0c3740cd69feeb8c5e54a25151f5fb7 Mon Sep 17 00:00:00 2001
From: Jeffrey Lim <Jeffrey-Lim@outlook.com>
Date: Fri, 8 Mar 2024 13:38:03 +0100
Subject: [PATCH 8/9] Fix pyproject

---
 pyproject.toml | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/pyproject.toml b/pyproject.toml
index 8f59573..c0dfead 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -17,20 +17,20 @@ classifiers = [
 ]
 dependencies= [
     # Machine Learning Libraries
-    "numpy~=1.26.4",
+    "numpy>=1.22.4",
 
     # Parallel Processing Libraries
-    "dask~=2023.12.0",
+    "dask>=2023.12.0",
 
     # Data Processing Libraries
-    "pandas~=1.3.3",
+    "pandas>=1.3.0",
 
     # Parquet
-    "pyarrow~=6.0.0",
+    "pyarrow>=6.0.0",
 
     # PyTorch
-    "torch~=2.2.1",
+    "torch>=2.1.0",
 
     # Agogos
-    "agogos~=0.2.0",
+    "agogos>=0.2.1",
 ]

From 7e6011cb95209c1f7381252fd297d0ddf0781208 Mon Sep 17 00:00:00 2001
From: schobbejak <jmvanselm@gmail.com>
Date: Fri, 8 Mar 2024 13:56:28 +0100
Subject: [PATCH 9/9] Update pyarrow version in requirements.txt

---
 requirements.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/requirements.txt b/requirements.txt
index b7cbcd5..923b87d 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -23,7 +23,7 @@ pandas==2.2.1
 partd==1.4.1
 pillow==10.2.0
 pluggy==1.4.0
-pyarrow==6.0.0
+pyarrow>=6.0.0
 pytest==8.0.2
 pytest-cov==4.1.0
 PyYAML==6.0.1