dataeval 0.73.0__tar.gz → 0.74.0__tar.gz

{dataeval-0.73.0 → dataeval-0.74.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: dataeval
-Version: 0.73.0
+Version: 0.74.0
 Summary: DataEval provides a simple interface to characterize image data and its impact on model performance across classification and object-detection tasks
 Home-page: https://dataeval.ai/
 License: MIT
@@ -23,7 +23,6 @@ Classifier: Topic :: Scientific/Engineering
 Provides-Extra: all
 Provides-Extra: tensorflow
 Provides-Extra: torch
-Requires-Dist: hdbscan (>=0.8.36)
 Requires-Dist: markupsafe (<3.0.2) ; extra == "tensorflow" or extra == "all"
 Requires-Dist: matplotlib ; extra == "torch" or extra == "all"
 Requires-Dist: numpy (>1.24.3)

{dataeval-0.73.0 → dataeval-0.74.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "dataeval"
-version = "0.73.0" # dynamic
+version = "0.74.0" # dynamic
 description = "DataEval provides a simple interface to characterize image data and its impact on model performance across classification and object-detection tasks"
 license = "MIT"
 readme = "README.md"
@@ -42,7 +42,6 @@ packages = [
 [tool.poetry.dependencies]
 # required
 python = ">=3.9,<3.13"
-hdbscan = {version = ">=0.8.36"}
 numpy = {version = ">1.24.3"}
 pillow = {version = ">=10.3.0"}
 scipy = {version = ">=1.10"}
@@ -69,8 +68,7 @@ all = ["matplotlib", "markupsafe", "tensorflow", "tensorflow_probability", "tf-k
 optional = true
 
 [tool.poetry.group.dev.dependencies]
-tox = {version = "*"}
-tox-uv = {version = "*"}
+nox = {version = "*", extras = ["uv"]}
 uv = {version = "*"}
 poetry = {version = "*"}
 poetry-lock-groups-plugin = {version = "*"}
@@ -122,7 +120,6 @@ files = ["src/dataeval/__init__.py"]
 name = "dataeval"
 
 [tool.poetry2conda.dependencies]
-nvidia-cudnn-cu11 = { name = "cudnn" }
 tensorflow_probability = { name = "tensorflow-probability" }
 torch = { name = "pytorch" }
 xxhash = { name = "python-xxhash" }
@@ -145,6 +142,9 @@ parallel = true
 exclude_also = [
   "raise NotImplementedError",
   "if TYPE_CHECKING:",
+  "if _IS_TENSORFLOW_AVAILABLE",
+  "if _IS_TORCH_AVAILABLE",
+  "if _IS_TORCHVISION_AVAILABLE",
 ]
 include = ["*/src/dataeval/*"]
 omit = [
@@ -164,6 +164,7 @@ exclude = [
   "*env*",
   "output",
   "_build",
+  ".nox",
   ".tox",
   "prototype",
 ]

{dataeval-0.73.0 → dataeval-0.74.0}/src/dataeval/__init__.py

@@ -1,4 +1,4 @@
-__version__ = "0.73.0"
+__version__ = "0.74.0"
 
 from importlib.util import find_spec
 
@@ -12,12 +12,12 @@ from dataeval import detectors, metrics  # noqa: E402
 
 __all__ = ["detectors", "metrics"]
 
-if _IS_TORCH_AVAILABLE:  # pragma: no cover
+if _IS_TORCH_AVAILABLE:
     from dataeval import workflows
 
     __all__ += ["workflows"]
 
-if _IS_TENSORFLOW_AVAILABLE or _IS_TORCH_AVAILABLE:  # pragma: no cover
+if _IS_TENSORFLOW_AVAILABLE or _IS_TORCH_AVAILABLE:
     from dataeval import utils
 
     __all__ += ["utils"]

{dataeval-0.73.0 → dataeval-0.74.0}/src/dataeval/detectors/__init__.py

@@ -7,7 +7,7 @@ from dataeval.detectors import drift, linters
 
 __all__ = ["drift", "linters"]
 
-if _IS_TENSORFLOW_AVAILABLE:  # pragma: no cover
+if _IS_TENSORFLOW_AVAILABLE:
     from dataeval.detectors import ood
 
     __all__ += ["ood"]

{dataeval-0.73.0 → dataeval-0.74.0}/src/dataeval/detectors/drift/__init__.py

@@ -10,7 +10,7 @@ from dataeval.detectors.drift.ks import DriftKS
 
 __all__ = ["DriftCVM", "DriftKS", "DriftOutput", "updates"]
 
-if _IS_TORCH_AVAILABLE:  # pragma: no cover
+if _IS_TORCH_AVAILABLE:
     from dataeval.detectors.drift.mmd import DriftMMD, DriftMMDOutput
     from dataeval.detectors.drift.torch import preprocess_drift
     from dataeval.detectors.drift.uncertainty import DriftUncertainty

{dataeval-0.73.0 → dataeval-0.74.0}/src/dataeval/detectors/drift/base.py

@@ -18,7 +18,7 @@ from typing import Any, Callable, Literal, TypeVar
 import numpy as np
 from numpy.typing import ArrayLike, NDArray
 
-from dataeval.interop import as_numpy, to_numpy
+from dataeval.interop import as_numpy
 from dataeval.output import OutputMetadata, set_metadata
 
 R = TypeVar("R")
@@ -196,7 +196,7 @@ class BaseDrift:
         if correction not in ["bonferroni", "fdr"]:
             raise ValueError("`correction` must be `bonferroni` or `fdr`.")
 
-        self._x_ref = to_numpy(x_ref)
+        self._x_ref = as_numpy(x_ref)
         self.x_ref_preprocessed: bool = x_ref_preprocessed
 
         # Other attributes

{dataeval-0.73.0 → dataeval-0.74.0}/src/dataeval/detectors/drift/torch.py

@@ -10,7 +10,6 @@ from __future__ import annotations
 
 __all__ = []
 
-from functools import partial
 from typing import Any, Callable
 
 import numpy as np
@@ -18,30 +17,7 @@ import torch
 import torch.nn as nn
 from numpy.typing import NDArray
 
-
-def get_device(device: str | torch.device | None = None) -> torch.device:
-    """
-    Instantiates a PyTorch device object.
-
-    Parameters
-    ----------
-    device : str | torch.device | None, default None
-        Either ``None``, a str ('gpu' or 'cpu') indicating the device to choose, or an
-        already instantiated device object. If ``None``, the GPU is selected if it is
-        detected, otherwise the CPU is used as a fallback.
-
-    Returns
-    -------
-    The instantiated device object.
-    """
-    if isinstance(device, torch.device):  # Already a torch device
-        return device
-    else:  # Instantiate device
-        if device is None or device.lower() in ["gpu", "cuda"]:
-            torch_device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
-        else:
-            torch_device = torch.device("cpu")
-    return torch_device
+from dataeval.utils.torch.utils import get_device, predict_batch
 
 
 def _mmd2_from_kernel_matrix(
@@ -79,82 +55,6 @@ def _mmd2_from_kernel_matrix(
     return mmd2
 
 
-def predict_batch(
-    x: NDArray[Any] | torch.Tensor,
-    model: Callable | nn.Module | nn.Sequential,
-    device: torch.device | None = None,
-    batch_size: int = int(1e10),
-    preprocess_fn: Callable | None = None,
-    dtype: type[np.generic] | torch.dtype = np.float32,
-) -> NDArray[Any] | torch.Tensor | tuple[Any, ...]:
-    """
-    Make batch predictions on a model.
-
-    Parameters
-    ----------
-    x : np.ndarray | torch.Tensor
-        Batch of instances.
-    model : Callable | nn.Module | nn.Sequential
-        PyTorch model.
-    device : torch.device | None, default None
-        Device type used. The default None tries to use the GPU and falls back on CPU.
-        Can be specified by passing either torch.device('cuda') or torch.device('cpu').
-    batch_size : int, default 1e10
-        Batch size used during prediction.
-    preprocess_fn : Callable | None, default None
-        Optional preprocessing function for each batch.
-    dtype : np.dtype | torch.dtype, default np.float32
-        Model output type, either a :term:`NumPy` or torch dtype, e.g. np.float32 or torch.float32.
-
-    Returns
-    -------
-    NDArray | torch.Tensor | tuple
-        Numpy array, torch tensor or tuples of those with model outputs.
-    """
-    device = get_device(device)
-    if isinstance(x, np.ndarray):
-        x = torch.from_numpy(x)
-    n = len(x)
-    n_minibatch = int(np.ceil(n / batch_size))
-    return_np = not isinstance(dtype, torch.dtype)
-    preds = []
-    with torch.no_grad():
-        for i in range(n_minibatch):
-            istart, istop = i * batch_size, min((i + 1) * batch_size, n)
-            x_batch = x[istart:istop]
-            if isinstance(preprocess_fn, Callable):
-                x_batch = preprocess_fn(x_batch)
-            preds_tmp = model(x_batch.to(device))
-            if isinstance(preds_tmp, (list, tuple)):
-                if len(preds) == 0:  # init tuple with lists to store predictions
-                    preds = tuple([] for _ in range(len(preds_tmp)))
-                for j, p in enumerate(preds_tmp):
-                    if isinstance(p, torch.Tensor):
-                        p = p.cpu()
-                    preds[j].append(p if not return_np or isinstance(p, np.ndarray) else p.numpy())
-            elif isinstance(preds_tmp, (np.ndarray, torch.Tensor)):
-                if isinstance(preds_tmp, torch.Tensor):
-                    preds_tmp = preds_tmp.cpu()
-                if isinstance(preds, tuple):
-                    preds = list(preds)
-                preds.append(
-                    preds_tmp
-                    if not return_np or isinstance(preds_tmp, np.ndarray)  # type: ignore
-                    else preds_tmp.numpy()
-                )
-            else:
-                raise TypeError(
-                    f"Model output type {type(preds_tmp)} not supported. The model \
-                    output type needs to be one of list, tuple, NDArray or \
-                    torch.Tensor."
-                )
-    concat = partial(np.concatenate, axis=0) if return_np else partial(torch.cat, dim=0)
-    out: tuple | np.ndarray | torch.Tensor = (
-        tuple(concat(p) for p in preds) if isinstance(preds, tuple) else concat(preds)  # type: ignore
-    )
-    return out
-
-
 def preprocess_drift(
     x: NDArray[Any],
     model: nn.Module,

{dataeval-0.73.0 → dataeval-0.74.0}/src/dataeval/detectors/linters/clusterer.py

@@ -480,7 +480,7 @@ class Clusterer:
             samples = self.clusters[level][cluster_id].samples
             if len(samples) >= self._min_num_samples_per_cluster:
                 duplicates_std.append(self.clusters[level][cluster_id].dist_std)
-        diag_mask = np.ones_like(self._sqdmat, dtype=bool)
+        diag_mask = np.ones_like(self._sqdmat, dtype=np.bool_)
         np.fill_diagonal(diag_mask, 0)
         diag_mask = np.triu(diag_mask)
 

dataeval-0.74.0/src/dataeval/detectors/ood/__init__.py

@@ -0,0 +1,22 @@
+"""
+Out-of-distribution (OOD)` detectors identify data that is different from the data used to train a particular model.
+"""
+
+from dataeval import _IS_TENSORFLOW_AVAILABLE, _IS_TORCH_AVAILABLE
+from dataeval.detectors.ood.base import OODOutput, OODScoreOutput
+
+__all__ = ["OODOutput", "OODScoreOutput"]
+
+if _IS_TENSORFLOW_AVAILABLE:
+    from dataeval.detectors.ood.ae import OOD_AE
+    from dataeval.detectors.ood.aegmm import OOD_AEGMM
+    from dataeval.detectors.ood.llr import OOD_LLR
+    from dataeval.detectors.ood.vae import OOD_VAE
+    from dataeval.detectors.ood.vaegmm import OOD_VAEGMM
+
+    __all__ += ["OOD_AE", "OOD_AEGMM", "OOD_LLR", "OOD_VAE", "OOD_VAEGMM"]
+
+elif _IS_TORCH_AVAILABLE:
+    from dataeval.detectors.ood.ae_torch import OOD_AE
+
+    __all__ += ["OOD_AE", "OODOutput"]

{dataeval-0.73.0 → dataeval-0.74.0}/src/dataeval/detectors/ood/ae.py

@@ -15,7 +15,8 @@ from typing import TYPE_CHECKING, Callable
 import numpy as np
 from numpy.typing import ArrayLike
 
-from dataeval.detectors.ood.base import OODBase, OODScoreOutput
+from dataeval.detectors.ood.base import OODScoreOutput
+from dataeval.detectors.ood.base_tf import OODBase
 from dataeval.interop import as_numpy
 from dataeval.utils.lazy import lazyload
 from dataeval.utils.tensorflow._internal.utils import predict_batch

dataeval-0.74.0/src/dataeval/detectors/ood/ae_torch.py

@@ -0,0 +1,70 @@
+"""
+Adapted for Pytorch from
+
+Source code derived from Alibi-Detect 0.11.4
+https://github.com/SeldonIO/alibi-detect/tree/v0.11.4
+
+Original code Copyright (c) 2023 Seldon Technologies Ltd
+Licensed under Apache Software License (Apache 2.0)
+"""
+
+from __future__ import annotations
+
+from typing import Callable
+
+import numpy as np
+import torch
+from numpy.typing import ArrayLike
+
+from dataeval.detectors.ood.base import OODScoreOutput
+from dataeval.detectors.ood.base_torch import OODBase
+from dataeval.interop import as_numpy
+from dataeval.utils.torch.utils import predict_batch
+
+
+class OOD_AE(OODBase):
+    """
+    Autoencoder based out-of-distribution detector.
+
+    Parameters
+    ----------
+    model : AriaAutoencoder
+        An Autoencoder model.
+    """
+
+    def __init__(self, model: torch.nn.Module, device: str | torch.device | None = None) -> None:
+        super().__init__(model, device)
+
+    def fit(
+        self,
+        x_ref: ArrayLike,
+        threshold_perc: float,
+        loss_fn: Callable[..., torch.nn.Module] | None = None,
+        optimizer: torch.optim.Optimizer | None = None,
+        epochs: int = 20,
+        batch_size: int = 64,
+        verbose: bool = False,
+    ) -> None:
+        if loss_fn is None:
+            loss_fn = torch.nn.MSELoss()
+
+        if optimizer is None:
+            optimizer = torch.optim.Adam(self.model.parameters(), lr=0.001)
+
+        super().fit(x_ref, threshold_perc, loss_fn, optimizer, epochs, batch_size, verbose)
+
+    def _score(self, X: ArrayLike, batch_size: int = int(1e10)) -> OODScoreOutput:
+        self._validate(X := as_numpy(X))
+
+        # reconstruct instances
+        X_recon = predict_batch(X, self.model, batch_size=batch_size)
+
+        # compute feature and instance level scores
+        fscore = np.power(X - X_recon, 2)
+        fscore_flat = fscore.reshape(fscore.shape[0], -1).copy()
+        n_score_features = int(np.ceil(fscore_flat.shape[1]))
+        sorted_fscore = np.sort(fscore_flat, axis=1)
+        sorted_fscore_perc = sorted_fscore[:, -n_score_features:]
+        iscore = np.mean(sorted_fscore_perc, axis=1)
+
+        return OODScoreOutput(iscore, fscore)

{dataeval-0.73.0 → dataeval-0.74.0}/src/dataeval/detectors/ood/aegmm.py

@@ -14,7 +14,8 @@ from typing import TYPE_CHECKING, Callable
 
 from numpy.typing import ArrayLike
 
-from dataeval.detectors.ood.base import OODGMMBase, OODScoreOutput
+from dataeval.detectors.ood.base import OODScoreOutput
+from dataeval.detectors.ood.base_tf import OODBaseGMM
 from dataeval.interop import to_numpy
 from dataeval.utils.lazy import lazyload
 from dataeval.utils.tensorflow._internal.gmm import gmm_energy
@@ -32,7 +33,7 @@ else:
     tf_models = lazyload("dataeval.utils.tensorflow._internal.models")
 
 
-class OOD_AEGMM(OODGMMBase):
+class OOD_AEGMM(OODBaseGMM):
     """
     AE with Gaussian Mixture Model based outlier detector.
 
@@ -62,5 +63,5 @@ class OOD_AEGMM(OODGMMBase):
     def _score(self, X: ArrayLike, batch_size: int = int(1e10)) -> OODScoreOutput:
         self._validate(X := to_numpy(X))
         _, z, _ = predict_batch(X, self.model, batch_size=batch_size)
-        energy, _ = gmm_energy(z, self.gmm_params, return_mean=False)
+        energy, _ = gmm_energy(z, self._gmm_params, return_mean=False)
         return OODScoreOutput(energy.numpy())  # type: ignore

{dataeval-0.73.0 → dataeval-0.74.0}/src/dataeval/detectors/ood/base.py

@@ -12,23 +12,14 @@ __all__ = ["OODOutput", "OODScoreOutput"]
 
 from abc import ABC, abstractmethod
 from dataclasses import dataclass
-from typing import TYPE_CHECKING, Callable, Literal, cast
+from typing import Callable, Generic, Literal, TypeVar
 
 import numpy as np
 from numpy.typing import ArrayLike, NDArray
 
 from dataeval.interop import to_numpy
 from dataeval.output import OutputMetadata, set_metadata
-from dataeval.utils.lazy import lazyload
-from dataeval.utils.tensorflow._internal.gmm import GaussianMixtureModelParams, gmm_params
-from dataeval.utils.tensorflow._internal.trainer import trainer
-
-if TYPE_CHECKING:
-    import tensorflow as tf
-    import tf_keras as keras
-else:
-    tf = lazyload("tensorflow")
-    keras = lazyload("tf_keras")
+from dataeval.utils.gmm import GaussianMixtureModelParams
 
 
 @dataclass(frozen=True)
@@ -85,16 +76,62 @@ class OODScoreOutput(OutputMetadata):
         return self.instance_score if ood_type == "instance" or self.feature_score is None else self.feature_score
 
 
-class OODBase(ABC):
-    def __init__(self, model: keras.Model) -> None:
-        self.model = model
+TGMMData = TypeVar("TGMMData")
+
+
+class OODGMMMixin(Generic[TGMMData]):
+    _gmm_params: GaussianMixtureModelParams[TGMMData]
+
+
+TModel = TypeVar("TModel", bound=Callable)
+TLossFn = TypeVar("TLossFn", bound=Callable)
+TOptimizer = TypeVar("TOptimizer")
+
+
+class OODFitMixin(Generic[TLossFn, TOptimizer], ABC):
+    @abstractmethod
+    def fit(
+        self,
+        x_ref: ArrayLike,
+        threshold_perc: float,
+        loss_fn: TLossFn | None,
+        optimizer: TOptimizer | None,
+        epochs: int,
+        batch_size: int,
+        verbose: bool,
+    ) -> None:
+        """
+        Train the model and infer the threshold value.
+
+        Parameters
+        ----------
+        x_ref : ArrayLike
+            Training data.
+        threshold_perc : float, default 100.0
+            Percentage of reference data that is normal.
+        loss_fn : TLossFn
+            Loss function used for training.
+        optimizer : TOptimizer
+            Optimizer used for training.
+        epochs : int, default 20
+            Number of training epochs.
+        batch_size : int, default 64
+            Batch size used for training.
+        verbose : bool, default True
+            Whether to print training progress.
+        """
 
-        self._ref_score: OODScoreOutput
-        self._threshold_perc: float
-        self._data_info: tuple[tuple, type] | None = None
 
-        if not isinstance(model, keras.Model):
-            raise TypeError("Model should be of type 'keras.Model'.")
+class OODBaseMixin(Generic[TModel], ABC):
+    _ref_score: OODScoreOutput
+    _threshold_perc: float
+    _data_info: tuple[tuple, type] | None = None
+
+    def __init__(
+        self,
+        model: TModel,
+    ) -> None:
+        self.model = model
 
     def _get_data_info(self, X: NDArray) -> tuple[tuple, type]:
         if not isinstance(X, np.ndarray):
@@ -107,9 +144,8 @@ class OODBase(ABC):
             raise RuntimeError(f"Expect data of type: {self._data_info[1]} and shape: {self._data_info[0]}. \
                                Provided data is type: {check_data_info[1]} and shape: {check_data_info[0]}.")
 
-    def _validate_state(self, X: NDArray, additional_attrs: list[str] | None = None) -> None:
-        attrs = ["_data_info", "_threshold_perc", "_ref_score"]
-        attrs = attrs if additional_attrs is None else attrs + additional_attrs
+    def _validate_state(self, X: NDArray) -> None:
+        attrs = [k for c in self.__class__.mro()[:-1][::-1] if hasattr(c, "__annotations__") for k in c.__annotations__]
         if not all(hasattr(self, attr) for attr in attrs) or any(getattr(self, attr) for attr in attrs) is None:
             raise RuntimeError("Metric needs to be `fit` before method call.")
         self._validate(X)
@@ -140,52 +176,6 @@ class OODBase(ABC):
     def _threshold_score(self, ood_type: Literal["feature", "instance"] = "instance") -> np.floating:
         return np.percentile(self._ref_score.get(ood_type), self._threshold_perc)
 
-    def fit(
-        self,
-        x_ref: ArrayLike,
-        threshold_perc: float,
-        loss_fn: Callable[..., tf.Tensor],
-        optimizer: keras.optimizers.Optimizer,
-        epochs: int,
-        batch_size: int,
-        verbose: bool,
-    ) -> None:
-        """
-        Train the model and infer the threshold value.
-
-        Parameters
-        ----------
-        x_ref : ArrayLike
-            Training data.
-        threshold_perc : float, default 100.0
-            Percentage of reference data that is normal.
-        loss_fn : Callable | None, default None
-            Loss function used for training.
-        optimizer : Optimizer, default keras.optimizers.Adam
-            Optimizer used for training.
-        epochs : int, default 20
-            Number of training epochs.
-        batch_size : int, default 64
-            Batch size used for training.
-        verbose : bool, default True
-            Whether to print training progress.
-        """
-
-        # Train the model
-        trainer(
-            model=self.model,
-            loss_fn=loss_fn,
-            x_train=to_numpy(x_ref),
-            optimizer=optimizer,
-            epochs=epochs,
-            batch_size=batch_size,
-            verbose=verbose,
-        )
-
-        # Infer the threshold values
-        self._ref_score = self.score(x_ref, batch_size)
-        self._threshold_perc = threshold_perc
-
     @set_metadata()
     def predict(
         self,
@@ -215,43 +205,3 @@ class OODBase(ABC):
         score = self.score(X, batch_size=batch_size)
         ood_pred = score.get(ood_type) > self._threshold_score(ood_type)
         return OODOutput(is_ood=ood_pred, **score.dict())
-
-
-class OODGMMBase(OODBase):
-    def __init__(self, model: keras.Model) -> None:
-        super().__init__(model)
-        self.gmm_params: GaussianMixtureModelParams
-
-    def _validate_state(self, X: NDArray, additional_attrs: list[str] | None = None) -> None:
-        if additional_attrs is None:
-            additional_attrs = ["gmm_params"]
-        super()._validate_state(X, additional_attrs)
-
-    def fit(
-        self,
-        x_ref: ArrayLike,
-        threshold_perc: float,
-        loss_fn: Callable[..., tf.Tensor],
-        optimizer: keras.optimizers.Optimizer,
-        epochs: int,
-        batch_size: int,
-        verbose: bool,
-    ) -> None:
-        # Train the model
-        trainer(
-            model=self.model,
-            loss_fn=loss_fn,
-            x_train=to_numpy(x_ref),
-            optimizer=optimizer,
-            epochs=epochs,
-            batch_size=batch_size,
-            verbose=verbose,
-        )
-
-        # Calculate the GMM parameters
-        _, z, gamma = cast(tuple[tf.Tensor, tf.Tensor, tf.Tensor], self.model(x_ref))
-        self.gmm_params = gmm_params(z, gamma)
-
-        # Infer the threshold values
-        self._ref_score = self.score(x_ref, batch_size)
-        self._threshold_perc = threshold_perc