Merge pull request #206 from TeamEpochGithub/v0.3.7

V0.3.7

.github/workflows/branch-name-check.yml

@@ -0,0 +1,18 @@
+name: Branch Name Check
+
+on:
+  pull_request:
+    branches:
+      - main
+
+jobs:
+  check-branch-name:
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Check if the branch name follows the version pattern
+        run: |
+          if [[ ! ${{ github.head_ref }} =~ ^v[0-9]+\.[0-9]+\.[0-9]+$ ]]; then
+            echo "Error: Branch name ${{ github.head_ref }} does not follow the required version pattern vX.Y.Z"
+            exit 1
+          fi

.github/workflows/main-branch-testing.yml

@@ -56,7 +56,7 @@ jobs:
       run: |
         venv/bin/python -m pip install --upgrade pip
         venv/bin/python -m pip install pytest
-        venv/bin/python -m pip install -r requirements.txt
+        venv/bin/python -m pip install -r requirements-dev.lock
         venv/bin/python -m pip install pytest-cov coverage
 
     - name: Test with pytest

.github/workflows/version-branch-testing.yml

@@ -53,7 +53,7 @@ jobs:
       run: |
         venv/bin/python -m pip install --upgrade pip
         venv/bin/python -m pip install pytest
-        venv/bin/python -m pip install -r requirements.txt
+        venv/bin/python -m pip install -r requirements-dev.lock
         venv/bin/python -m pip install pytest-cov coverage
 
     - name: Test with pytest

.pre-commit-config.yaml

@@ -1,23 +1,44 @@
-exclude: ^(venv/|.venv/|tests/)
+exclude: ^(external/|venv/|.venv/|tests/|.cache)
 repos:
+  - repo: local  # Remove this when a new version of pre-commit-hooks (>4.6.0) is released
+    hooks:
+      - id: check-illegal-windows-names
+        name: check illegal windows names
+        entry: Illegal Windows filenames detected
+        language: fail
+        files: '(?i)((^|/)(CON|PRN|AUX|NUL|COM[\d¹²³]|LPT[\d¹²³])(\.|/|$)|[<>:\"\\|?*\x00-\x1F]|/[^/]*[\.\s]/|[^/]*[\.\s]$)'
   - repo: https://github.com/pre-commit/pre-commit-hooks
-    rev: v4.5.0
+    rev: v4.6.0
     hooks:
-      - id: check-ast
+      - id: check-added-large-files
+      - id: check-builtin-literals
+      - id: check-executables-have-shebangs
+      # - id: check-illegal-windows-names  # Uncomment this when a new version of pre-commit-hooks (>4.6.0) is released
       - id: check-json
       - id: pretty-format-json
-        args: [--autofix, --no-sort-keys]
+        args: [ "--autofix", "--no-sort-keys" ]
+      - id: check-merge-conflict
+      - id: check-shebang-scripts-are-executable
       - id: check-symlinks
       - id: check-toml
       - id: check-xml
       - id: check-yaml
+      - id: detect-private-key
+      - id: destroyed-symlinks
       - id: mixed-line-ending
       - id: end-of-file-fixer
+      - id: fix-byte-order-marker
+      - id: name-tests-test
+        args: [ "--pytest-test-first" ]
       - id: trailing-whitespace
   - repo: https://github.com/abravalheri/validate-pyproject
-    rev: v0.16
+    rev: v0.18
     hooks:
       - id: validate-pyproject
+  - repo: https://github.com/citation-file-format/cffconvert
+    rev: b6045d78aac9e02b039703b030588d54d53262ac
+    hooks:
+      - id: validate-cff
   - repo: https://github.com/pre-commit/pygrep-hooks
     rev: v1.10.0
     hooks:
@@ -32,23 +53,31 @@ repos:
       - id: sphinx-lint
         args: [ --enable=default-role ]
   - repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: v0.1.9
+    rev: v0.4.10
     hooks:
-      - id: ruff
       - id: ruff-format
+      - id: ruff
+  # Pydoclint is useful, but gives too many false positives
+  #  - repo: https://github.com/jsh9/pydoclint
+  #    rev: 0.4.1
+  #    hooks:
+  #      - id: pydoclint
+  #        args: [ --quiet ]
   - repo: https://github.com/pre-commit/mirrors-mypy
-    rev: v1.8.0
+    rev: v1.10.0
     hooks:
       - id: mypy
         additional_dependencies:
-          - agogos
-          - annotated-types
-          - dask
-          - matplotlib
-          - numpy
-          - polars
-          - torch
-          - traitlets
-          - timm
-          - kornia
-        args: [ --disallow-any-generics, --disallow-untyped-defs, --disable-error-code=import-untyped]
+          - numpy==1.26.4
+          - pandas-stubs>=2.2.2.240514
+          - matplotlib==3.8.4
+          - torch==2.3.1
+          - dask==2024.6.2
+          - typing_extensions==4.9.0
+          - annotated-types==0.7.0
+          - polars==0.20.31
+          - kornia==0.7.2
+          - timm==1.0.7
+        args:
+          - "--fast-module-lookup"
+          - "--disallow-any-generics"

.python-version

@@ -0,0 +1 @@
+3.10.14

CITATION.cff

@@ -17,7 +17,7 @@ authors:
 - family-names: "Kopar"
   given-names: "Cahit Tolga"
   affiliation: "TU Delft Dream Team Epoch"
-  email: "https://github.com/tolgakopar"
+  email: "cahittolgakopar@gmail.com"
 - family-names: "Selm"
   name-particle: "van"
   given-names: "Jasper"

epochalyst/_core/_caching/_cacher.py

@@ -23,13 +23,13 @@
 try:
     import polars as pl
 except ImportError:
-    """User doen't require these packages"""
+    """User doesn't require these packages"""
 
 from epochalyst._core._logging._logger import _Logger
 
-if sys.version_info < (3, 11):
+if sys.version_info < (3, 11):  # pragma: no cover (<py311)
     from typing_extensions import NotRequired
-else:
+else:  # pragma: no cover (py311+)
     from typing import NotRequired
 
 
@@ -197,7 +197,7 @@ def _get_cache(self, name: str, cache_args: CacheArgs | None = None) -> Any:  #
             if output_data_type == "polars_dataframe":
                 return pl.read_parquet(storage_path + name + ".parquet", **read_args)
 
-            self.log_to_debug(
+            self.log_to_debug(  # type: ignore[unreachable]
                 f"Invalid output data type: {output_data_type}, for loading .parquet file.",
             )
             raise ValueError(
@@ -239,7 +239,7 @@ def _get_cache(self, name: str, cache_args: CacheArgs | None = None) -> Any:  #
             with open(storage_path + name + ".pkl", "rb") as file:
                 return pickle.load(file, **read_args)  # noqa: S301
 
-        self.log_to_debug(f"Invalid storage type: {storage_type}")
+        self.log_to_debug(f"Invalid storage type: {storage_type}")  # type: ignore[unreachable]
         raise ValueError(
             "storage_type must be .npy, .parquet, .csv, or .npy_stack, other types not supported yet",
         )
@@ -348,7 +348,7 @@ def _store_cache(self, name: str, data: Any, cache_args: CacheArgs | None = None
                     **({"protocol": pickle.HIGHEST_PROTOCOL} | store_args),
                 )
         else:
-            self.log_to_debug(f"Invalid storage type: {storage_type}")
+            self.log_to_debug(f"Invalid storage type: {storage_type}")  # type: ignore[unreachable]
             raise ValueError(
                 "storage_type must be .npy, .parquet, .csv or .npy_stack, other types not supported yet",
             )

epochalyst/_core/_logging/_logger.py

@@ -1,4 +1,5 @@
 """_Logger add abstract logging functionality to other classes."""
+
 from abc import abstractmethod
 from typing import Any
 

epochalyst/pipeline/ensemble.py

@@ -1,4 +1,5 @@
 """EnsemblePipeline for ensebling multiple ModelPipelines."""
+
 from typing import Any
 
 from agogos.training import ParallelTrainingSystem

epochalyst/pipeline/model/model.py

@@ -1,4 +1,5 @@
 """ModelPipeline connects multiple transforming and training systems for extended training functionality."""
+
 from typing import Any
 
 from agogos.training import Pipeline

epochalyst/pipeline/model/training/models/timm.py

@@ -1,4 +1,5 @@
 """Timm model for 2D image classification."""
+
 import torch
 from torch import nn
 

epochalyst/pipeline/model/training/pretrain_block.py

@@ -1,4 +1,5 @@
 """PretrainBlock to implement modules such as scalers."""
+
 from abc import abstractmethod
 from dataclasses import dataclass
 from typing import Any

epochalyst/pipeline/model/training/torch_trainer.py

@@ -1,16 +1,18 @@
 """TorchTrainer is a module that allows for the training of Torch models."""
+
 import copy
 import functools
 import gc
 from collections.abc import Callable
 from dataclasses import dataclass, field
+from os import PathLike
 from pathlib import Path
-from typing import Annotated, Any, TypeVar
+from typing import Annotated, Any, Literal, TypeVar
 
 import numpy as np
 import numpy.typing as npt
 import torch
-from annotated_types import Gt, Interval
+from annotated_types import Ge, Gt, Interval
 from torch import Tensor, nn
 from torch.optim import Optimizer
 from torch.optim.lr_scheduler import LRScheduler
@@ -163,7 +165,7 @@ def log_to_terminal(self, message: str) -> None:
 
     # Training parameters
     epochs: Annotated[int, Gt(0)] = 10
-    patience: Annotated[int, Gt(0)] = 5  # Early stopping
+    patience: Annotated[int, Gt(0)] = -1  # Early stopping
     batch_size: Annotated[int, Gt(0)] = 32
     collate_fn: Callable[[tuple[Tensor, ...]], tuple[Tensor, ...]] = field(default=custom_collate, init=True, repr=False, compare=False)
 
@@ -174,18 +176,22 @@ def log_to_terminal(self, message: str) -> None:
 
     # Misc
     model_name: str | None = None  # No spaces allowed
-    trained_models_directory: Path = Path("tm")
-    to_predict: str = "validation"
+    trained_models_directory: PathLike[str] = field(default=Path("tm/"), repr=False, compare=False)
+    to_predict: Literal["validation", "all", "none"] = field(default="validation", repr=False, compare=False)
 
     # Parameters relevant for Hashing
-    n_folds: float = field(default=-1, init=True, repr=False, compare=False)
+    n_folds: Annotated[int, Ge(0)] = field(default=-1, init=True, repr=False, compare=False)
     _fold: int = field(default=-1, init=False, repr=False, compare=False)
     validation_size: Annotated[float, Interval(ge=0, le=1)] = 0.2
 
     # Types for tensors
     x_tensor_type: str = "float"
     y_tensor_type: str = "float"
 
+    # Prefix and postfix for logging to external
+    logging_prefix: str = field(default="", init=True, repr=False, compare=False)
+    logging_postfix: str = field(default="", init=True, repr=False, compare=False)
+
     def __post_init__(self) -> None:
         """Post init method for the TorchTrainer class."""
         # Make sure to_predict is either "validation" or "all" or "none"
@@ -570,8 +576,8 @@ def _training_loop(
         if fold > -1:
             fold_no = f"_{fold}"
 
-        self.external_define_metric(f"Training/Train Loss{fold_no}", "epoch")
-        self.external_define_metric(f"Validation/Validation Loss{fold_no}", "epoch")
+        self.external_define_metric(self.wrap_log(f"Training/Train Loss{fold_no}"), self.wrap_log("epoch"))
+        self.external_define_metric(self.wrap_log(f"Validation/Validation Loss{fold_no}"), self.wrap_log("epoch"))
 
         # Set the scheduler to the correct epoch
         if self.initialized_scheduler is not None:
@@ -586,8 +592,8 @@ def _training_loop(
             # Log train loss
             self.log_to_external(
                 message={
-                    f"Training/Train Loss{fold_no}": train_losses[-1],
-                    "epoch": epoch,
+                    self.wrap_log(f"Training/Train Loss{fold_no}"): train_losses[-1],
+                    self.wrap_log("epoch"): epoch,
                 },
             )
 
@@ -616,8 +622,8 @@ def _training_loop(
                 # Log validation loss and plot train/val loss against each other
                 self.log_to_external(
                     message={
-                        f"Validation/Validation Loss{fold_no}": val_losses[-1],
-                        "epoch": epoch,
+                        self.wrap_log(f"Validation/Validation Loss{fold_no}"): val_losses[-1],
+                        self.wrap_log("epoch"): epoch,
                     },
                 )
 
@@ -631,21 +637,19 @@ def _training_loop(
                             ),  # Ensure it's a list, not a range object
                             "ys": [train_losses, val_losses],
                             "keys": [f"Train{fold_no}", f"Validation{fold_no}"],
-                            "title": f"Training/Loss{fold_no}",
+                            "title": self.wrap_log(f"Training/Loss{fold_no}"),
                             "xname": "Epoch",
                         },
                     },
                 )
 
                 # Early stopping
                 if self._early_stopping():
-                    self.log_to_external(
-                        message={f"Epochs{fold_no}": (epoch + 1) - self.patience},
-                    )
+                    self.log_to_external(message={self.wrap_log(f"Epochs{fold_no}"): (epoch + 1) - self.patience})
                     break
 
             # Log the trained epochs to wandb if we finished training
-            self.log_to_external(message={f"Epochs{fold_no}": epoch + 1})
+            self.log_to_external(message={self.wrap_log(f"Epochs{fold_no}"): epoch + 1})
 
     def _train_one_epoch(
         self,
@@ -815,15 +819,19 @@ def get_model_path(self) -> Path:
 
         :return: The model path.
         """
-        return Path(f"{self.trained_models_directory}/{self.get_hash()}.pt")
+        return Path(self.trained_models_directory) / f"{self.get_hash()}.pt"
 
     def get_model_checkpoint_path(self, epoch: int) -> Path:
         """Get the checkpoint path.
 
         :param epoch: The epoch number.
         :return: The checkpoint path.
         """
-        return Path(f"{self.trained_models_directory}/{self.get_hash()}_checkpoint_{epoch}.pt")
+        return Path(self.trained_models_directory) / f"{self.get_hash()}_checkpoint_{epoch}.pt"
+
+    def wrap_log(self, text: str) -> str:
+        """Add logging prefix and postfix to the message."""
+        return f"{self.logging_prefix}{text}{self.logging_postfix}"
 
 
 class TrainValidationDataset(Dataset[T_co]):

epochalyst/pipeline/model/training/training.py

@@ -1,4 +1,5 @@
 """TrainingPipeline for creating a sequential pipeline of TrainType classes."""
+
 from typing import Any
 
 from agogos.training import TrainingSystem, TrainType

epochalyst/pipeline/model/training/training_block.py

@@ -1,4 +1,5 @@
 """TrainingBlock that can be inherited from to make blocks for a training pipeline."""
+
 from abc import abstractmethod
 from typing import Any
 

epochalyst/pipeline/model/training/utils/tensor_functions.py

@@ -1,4 +1,5 @@
 """Module with tensor functions."""
+
 import torch
 from torch import Tensor
 

epochalyst/pipeline/model/transformation/transformation.py

@@ -1,4 +1,5 @@
 """TransformationPipeline that extends from TransformingSystem, _Cacher and _Logger."""
+
 from dataclasses import dataclass
 from typing import Any
 

epochalyst/pipeline/model/transformation/transformation_block.py

@@ -1,4 +1,5 @@
 """TransformationBlock module than can be extended by implementing the custom_transform method."""
+
 from abc import abstractmethod
 from typing import Any