deepbox 0.1.0 → 0.2.0

package/dist/{index-D9Loo1_A.d.cts → index-CNj2Mxwf.d.cts}

@@ -1,14 +1,15 @@
-import { T as Tensor } from './Tensor-BQLk1ltW.cjs';
+import { T as Tensor } from './Tensor-fxBg-TFZ.cjs';
+import { S as Shape, c as ScalarDType } from './tensor-IlVTF0bz.cjs';
 
 /**
  * Base type for all estimators (models) in Deepbox.
  *
- * This follows the scikit-learn estimator type design.
+ * Base estimator type for all ML models.
  *
  * @template FitParams - Type of parameters passed to fit method
  *
  * References:
- * - scikit-learn API: https://scikit-learn.org/stable/developers/develop.html
+ * - Deepbox ML: https://deepbox.dev/docs/ml-linear
  */
 type Estimator<FitParams = void> = {
     /**
@@ -238,7 +239,7 @@ type OutlierDetector = Estimator<void> & {
  * // labels: [0, 0, 0, 1, 1, -1]  (-1 = noise)
  * ```
  *
- * @see {@link https://scikit-learn.org/stable/modules/generated/sklearn.cluster.DBSCAN.html | scikit-learn DBSCAN}
+ * @see {@link https://deepbox.dev/docs/ml-clustering | Deepbox Clustering}
  */
 declare class DBSCAN implements Clusterer {
     private eps;
@@ -295,6 +296,13 @@ declare class DBSCAN implements Clusterer {
      * @throws {NotFittedError} If the model has not been fitted
      */
     get labels(): Tensor;
+    /**
+     * Number of clusters found (excluding noise).
+     *
+     * @returns Number of distinct clusters (labels >= 0)
+     * @throws {NotFittedError} If the model has not been fitted
+     */
+    get nClusters(): number;
     /**
      * Get indices of core samples discovered during fitting.
      *
@@ -352,8 +360,8 @@ declare class DBSCAN implements Clusterer {
  * console.log('Centroids:', kmeans.clusterCenters);
  * ```
  *
- * @see {@link https://en.wikipedia.org/wiki/K-means_clustering | Wikipedia: K-means}
- * @see {@link https://scikit-learn.org/stable/modules/generated/sklearn.cluster.KMeans.html | scikit-learn KMeans}
+ * @see {@link https://deepbox.dev/docs/ml-clustering | Deepbox Clustering}
+ * @see {@link https://deepbox.dev/docs/ml-clustering | Deepbox Clustering}
  */
 declare class KMeans implements Clusterer {
     private nClusters;
@@ -482,8 +490,8 @@ declare class KMeans implements Clusterer {
  * console.log('Explained variance ratio:', pca.explainedVarianceRatio);
  * ```
  *
- * @see {@link https://en.wikipedia.org/wiki/Principal_component_analysis | Wikipedia: PCA}
- * @see {@link https://scikit-learn.org/stable/modules/generated/sklearn.decomposition.PCA.html | scikit-learn PCA}
+ * @see {@link https://deepbox.dev/docs/ml-decomposition | Deepbox Dimensionality Reduction}
+ * @see {@link https://deepbox.dev/docs/ml-decomposition | Deepbox Dimensionality Reduction}
  */
 declare class PCA implements Transformer {
     private readonly nComponents?;
@@ -590,7 +598,7 @@ declare class PCA implements Transformer {
  * const predictions = gbr.predict(X);
  * ```
  *
- * @see {@link https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.GradientBoostingRegressor.html | scikit-learn GradientBoostingRegressor}
+ * @see {@link https://deepbox.dev/docs/ml-ensemble | Deepbox Ensemble Methods}
  */
 declare class GradientBoostingRegressor implements Regressor {
     /** Number of boosting stages (trees) */
@@ -669,8 +677,10 @@ declare class GradientBoostingRegressor implements Regressor {
 /**
  * Gradient Boosting Classifier.
  *
- * Uses gradient boosting with shallow regression trees for binary classification.
- * Optimizes log loss (cross-entropy) using sigmoid function.
+ * Uses gradient boosting with shallow regression trees for classification.
+ * Supports both binary and multiclass classification.
+ * - Binary: optimizes log loss using sigmoid function.
+ * - Multiclass: uses One-vs-Rest (OvR) strategy, training one binary model per class.
  *
  * @example
  * ```ts
@@ -685,7 +695,7 @@ declare class GradientBoostingRegressor implements Regressor {
  * const predictions = gbc.predict(X);
  * ```
  *
- * @see {@link https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.GradientBoostingClassifier.html | scikit-learn GradientBoostingClassifier}
+ * @see {@link https://deepbox.dev/docs/ml-ensemble | Deepbox Ensemble Methods}
  */
 declare class GradientBoostingClassifier implements Classifier {
     /** Number of boosting stages */
@@ -696,10 +706,10 @@ declare class GradientBoostingClassifier implements Classifier {
     private maxDepth;
     /** Minimum samples to split */
     private minSamplesSplit;
-    /** Array of weak learners */
-    private estimators;
-    /** Initial log-odds prediction */
-    private initPrediction;
+    /** Per-class arrays of weak learners (OvR for multiclass, single for binary) */
+    private estimatorsPerClass;
+    /** Per-class initial log-odds predictions */
+    private initPredictions;
     /** Number of features */
     private nFeatures;
     /** Unique class labels */
@@ -712,19 +722,29 @@ declare class GradientBoostingClassifier implements Classifier {
         readonly maxDepth?: number;
         readonly minSamplesSplit?: number;
     });
+    /**
+     * Fit a single binary boosting ensemble.
+     * Trains nEstimators regression trees to optimize log loss for a binary target.
+     */
+    private fitBinary;
+    /**
+     * Compute raw scores for a single binary ensemble.
+     */
+    private predictRawBinary;
     /**
      * Fit the gradient boosting classifier on training data.
      *
      * Builds an additive model by sequentially fitting regression trees
      * to the pseudo-residuals (gradient of log loss).
+     * Supports binary (2 classes) and multiclass (>2 classes via OvR).
      *
      * @param X - Training data of shape (n_samples, n_features)
-     * @param y - Target class labels of shape (n_samples,). Must contain exactly 2 classes.
+     * @param y - Target class labels of shape (n_samples,). Must contain at least 2 classes.
      * @returns this - The fitted estimator
      * @throws {ShapeError} If X is not 2D or y is not 1D
      * @throws {ShapeError} If X and y have different number of samples
      * @throws {DataValidationError} If X or y contain NaN/Inf values
-     * @throws {InvalidParameterError} If y does not contain exactly 2 classes
+     * @throws {InvalidParameterError} If y does not contain at least 2 classes
      */
     fit(X: Tensor, y: Tensor): this;
     /**
@@ -740,11 +760,10 @@ declare class GradientBoostingClassifier implements Classifier {
     /**
      * Predict class probabilities for samples in X.
      *
-     * Returns a matrix of shape (n_samples, 2) where columns are
-     * [P(class_0), P(class_1)].
+     * Returns a matrix of shape (n_samples, n_classes).
      *
      * @param X - Samples of shape (n_samples, n_features)
-     * @returns Class probability matrix of shape (n_samples, 2)
+     * @returns Class probability matrix of shape (n_samples, n_classes)
      * @throws {NotFittedError} If the model has not been fitted
      * @throws {ShapeError} If X has wrong dimensions or feature count
      * @throws {DataValidationError} If X contains NaN/Inf values
@@ -1041,7 +1060,7 @@ declare class LinearRegression implements Regressor {
      * @throws {ShapeError} If X has wrong dimensions or feature count
      * @throws {DataValidationError} If X contains NaN/Inf values
      */
-    predict(X: Tensor): Tensor;
+    predict(X: Tensor): Tensor<Shape, ScalarDType>;
     /**
      * Return the coefficient of determination R^2 of the prediction.
      *
@@ -1434,7 +1453,7 @@ declare class Ridge implements Regressor {
  * const embedding = tsne.fitTransform(X);
  * ```
  *
- * @see {@link https://scikit-learn.org/stable/modules/generated/sklearn.manifold.TSNE.html | scikit-learn TSNE}
+ * @see {@link https://deepbox.dev/docs/ml-manifold | Deepbox Manifold Learning}
  * @see van der Maaten, L.J.P.; Hinton, G.E. (2008). "Visualizing High-Dimensional Data Using t-SNE"
  */
 declare class TSNE {
@@ -1543,6 +1562,14 @@ declare class TSNE {
      * Fit the model (same as fitTransform for t-SNE).
      */
     fit(X: Tensor): this;
+    /**
+     * Return the fitted embedding. For t-SNE, transform is equivalent to
+     * returning the already-computed embedding (t-SNE is non-parametric).
+     *
+     * @param _X - Ignored, present for API compatibility
+     * @returns Low-dimensional embedding of shape (n_samples, n_components)
+     */
+    transform(_X?: Tensor): Tensor;
     /**
      * Get the embedding.
      */
@@ -1585,8 +1612,8 @@ declare class TSNE {
  * const predictions = nb.predict(tensor([[2.5, 3.5]]));
  * ```
  *
- * @see {@link https://en.wikipedia.org/wiki/Naive_Bayes_classifier | Wikipedia: Naive Bayes}
- * @see {@link https://scikit-learn.org/stable/modules/generated/sklearn.naive_bayes.GaussianNB.html | scikit-learn GaussianNB}
+ * @see {@link https://deepbox.dev/docs/ml-naive-bayes | Deepbox Naive Bayes}
+ * @see {@link https://deepbox.dev/docs/ml-naive-bayes | Deepbox Naive Bayes}
  */
 declare class GaussianNB implements Classifier {
     private readonly varSmoothing;
@@ -1737,8 +1764,8 @@ declare abstract class KNeighborsBase {
  * const predictions = knn.predict(tensor([[1.5, 1.5]]));
  * ```
  *
- * @see {@link https://en.wikipedia.org/wiki/K-nearest_neighbors_algorithm | Wikipedia: KNN}
- * @see {@link https://scikit-learn.org/stable/modules/generated/sklearn.neighbors.KNeighborsClassifier.html | scikit-learn KNeighborsClassifier}
+ * @see {@link https://deepbox.dev/docs/ml-neighbors | Deepbox Nearest Neighbors}
+ * @see {@link https://deepbox.dev/docs/ml-neighbors | Deepbox Nearest Neighbors}
  */
 declare class KNeighborsClassifier extends KNeighborsBase implements Classifier {
     /**
@@ -1805,7 +1832,7 @@ declare class KNeighborsClassifier extends KNeighborsBase implements Classifier
  * const predictions = knn.predict(tensor([[1.5]]));
  * ```
  *
- * @see {@link https://scikit-learn.org/stable/modules/generated/sklearn.neighbors.KNeighborsRegressor.html | scikit-learn KNeighborsRegressor}
+ * @see {@link https://deepbox.dev/docs/ml-neighbors | Deepbox Nearest Neighbors}
  */
 declare class KNeighborsRegressor extends KNeighborsBase implements Regressor {
     /**
@@ -1868,7 +1895,7 @@ declare class KNeighborsRegressor extends KNeighborsBase implements Regressor {
  * const predictions = svm.predict(X);
  * ```
  *
- * @see {@link https://scikit-learn.org/stable/modules/generated/sklearn.svm.LinearSVC.html | scikit-learn LinearSVC}
+ * @see {@link https://deepbox.dev/docs/ml-svm | Deepbox SVM}
  */
 declare class LinearSVC implements Classifier {
     /** Regularization parameter (inverse of regularization strength) */
@@ -1877,13 +1904,13 @@ declare class LinearSVC implements Classifier {
     private readonly maxIter;
     /** Tolerance for stopping criterion */
     private readonly tol;
-    /** Weight vector of shape (n_features,) */
-    private weights;
-    /** Bias term */
-    private bias;
+    /** Per-class weight vectors (OvR for multiclass, single for binary) */
+    private weightsPerClass;
+    /** Per-class bias terms */
+    private biasPerClass;
     /** Number of features seen during fit */
     private nFeatures;
-    /** Unique class labels [0, 1] mapped from original labels */
+    /** Unique class labels */
     private classLabels;
     /** Whether the model has been fitted */
     private fitted;
@@ -1900,19 +1927,27 @@ declare class LinearSVC implements Classifier {
         readonly maxIter?: number;
         readonly tol?: number;
     });
+    /**
+     * Fit a single binary SVM using sub-gradient descent on hinge loss.
+     * Maps labels to {-1, +1} and returns learned weights + bias.
+     */
+    private fitBinary;
+    /**
+     * Compute decision value for a single binary classifier.
+     */
+    private decisionBinary;
     /**
      * Fit the SVM classifier using sub-gradient descent.
      *
-     * Uses a simplified hinge loss optimization with L2 regularization.
-     * Objective: minimize (1/2)||w||² + C * Σmax(0, 1 - y_i(w · x_i + b))
+     * Supports both binary and multiclass classification (via OvR).
      *
      * @param X - Training data of shape (n_samples, n_features)
-     * @param y - Target labels of shape (n_samples,). Must contain exactly 2 classes.
+     * @param y - Target labels of shape (n_samples,). Must contain at least 2 classes.
      * @returns this - The fitted estimator
      * @throws {ShapeError} If X is not 2D or y is not 1D
      * @throws {ShapeError} If X and y have different number of samples
      * @throws {DataValidationError} If X or y contain NaN/Inf values
-     * @throws {InvalidParameterError} If y does not contain exactly 2 classes
+     * @throws {InvalidParameterError} If y does not contain at least 2 classes
      */
     fit(X: Tensor, y: Tensor): this;
     /**
@@ -1954,12 +1989,12 @@ declare class LinearSVC implements Classifier {
      */
     get coef(): Tensor;
     /**
-     * Get the bias term.
+     * Get the bias terms.
      *
-     * @returns Bias value
+     * @returns Bias values as tensor
      * @throws {NotFittedError} If the model has not been fitted
      */
-    get intercept(): number;
+    get intercept(): Tensor;
     /**
      * Get hyperparameters for this estimator.
      *
@@ -1992,7 +2027,7 @@ declare class LinearSVC implements Classifier {
  * const predictions = svr.predict(X);
  * ```
  *
- * @see {@link https://scikit-learn.org/stable/modules/generated/sklearn.svm.LinearSVR.html | scikit-learn LinearSVR}
+ * @see {@link https://deepbox.dev/docs/ml-svm | Deepbox SVM}
  */
 declare class LinearSVR implements Regressor {
     /** Regularization parameter */
@@ -2088,7 +2123,7 @@ declare class LinearSVR implements Regressor {
  * const predictions = clf.predict(X);
  * ```
  *
- * @see {@link https://scikit-learn.org/stable/modules/generated/sklearn.tree.DecisionTreeClassifier.html | scikit-learn DecisionTreeClassifier}
+ * @see {@link https://deepbox.dev/docs/ml-tree | Deepbox Decision Trees}
  */
 declare class DecisionTreeClassifier implements Classifier {
     private maxDepth;
@@ -2182,7 +2217,7 @@ declare class DecisionTreeClassifier implements Classifier {
  *
  * Uses MSE reduction to find optimal splits for regression tasks.
  *
- * @see {@link https://scikit-learn.org/stable/modules/generated/sklearn.tree.DecisionTreeRegressor.html | scikit-learn DecisionTreeRegressor}
+ * @see {@link https://deepbox.dev/docs/ml-tree | Deepbox Decision Trees}
  */
 declare class DecisionTreeRegressor implements Regressor {
     private maxDepth;
@@ -2275,7 +2310,7 @@ declare class DecisionTreeRegressor implements Regressor {
  * const predictions = clf.predict(X_test);
  * ```
  *
- * @see {@link https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestClassifier.html | scikit-learn RandomForestClassifier}
+ * @see {@link https://deepbox.dev/docs/ml-tree | Deepbox Decision Trees}
  */
 declare class RandomForestClassifier implements Classifier {
     private readonly nEstimators;
@@ -2389,7 +2424,7 @@ declare class RandomForestClassifier implements Classifier {
  * const predictions = reg.predict(X_test);
  * ```
  *
- * @see {@link https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestRegressor.html | scikit-learn RandomForestRegressor}
+ * @see {@link https://deepbox.dev/docs/ml-tree | Deepbox Decision Trees}
  */
 declare class RandomForestRegressor implements Regressor {
     private readonly nEstimators;

package/dist/{index-DmEg_LCm.d.cts → index-CYlxeNW1.d.cts}

@@ -1,5 +1,5 @@
-import { A as AnyTensor } from './index-eJgeni9c.cjs';
-import { T as Tensor } from './Tensor-BQLk1ltW.cjs';
+import { A as AnyTensor } from './index-DLdiQzf0.cjs';
+import { T as Tensor } from './Tensor-fxBg-TFZ.cjs';
 
 /**
  * @internal
@@ -676,7 +676,9 @@ declare function barh(y: Tensor, width: Tensor, options?: PlotOptions): void;
 /**
  * Plot a histogram on the current axes.
  */
-declare function hist(x: Tensor, bins?: number, options?: PlotOptions): void;
+declare function hist(x: Tensor, bins?: number | (PlotOptions & {
+    bins?: number;
+}), options?: PlotOptions): void;
 /**
  * Plot a box-and-whisker summary on the current axes.
  */

package/dist/{index-D61yaSMY.d.cts → index-CiTd61a5.d.ts}

@@ -1,4 +1,4 @@
-import { T as Tensor } from './Tensor-BQLk1ltW.cjs';
+import { T as Tensor } from './Tensor-BORFp_zt.js';
 
 type DataLoaderOptions = {
     batchSize?: number;
@@ -9,7 +9,7 @@ type DataLoaderOptions = {
 /**
  * Data loader for batching and shuffling datasets.
  *
- * Similar to PyTorch's DataLoader. Provides efficient iteration over datasets with support for
+ * Provides efficient iteration over datasets with support for
  * batching, shuffling, and deterministic reproducibility.
  *
  * @remarks
@@ -58,7 +58,7 @@ type DataLoaderOptions = {
  * }
  * ```
  *
- * @see {@link https://pytorch.org/docs/stable/data.html#torch.utils.data.DataLoader | PyTorch DataLoader}
+ * @see {@link https://deepbox.dev/docs/datasets-dataloader | Deepbox DataLoader}
  */
 declare class DataLoader<TTarget extends Tensor | undefined = undefined> {
     private X;
@@ -75,8 +75,7 @@ declare class DataLoader<TTarget extends Tensor | undefined = undefined> {
      * Number of batches in the data loader.
      */
     get length(): number;
-    [Symbol.iterator](this: DataLoader<Tensor>): IterableIterator<[Tensor, Tensor]>;
-    [Symbol.iterator](this: DataLoader<undefined>): IterableIterator<[Tensor]>;
+    [Symbol.iterator](): IterableIterator<TTarget extends Tensor ? [Tensor, Tensor] : [Tensor]>;
     private prepareIteration;
     private iterateX;
     private iterateXY;
@@ -98,7 +97,7 @@ declare class DataLoader<TTarget extends Tensor | undefined = undefined> {
  * @param options.randomState - Seed for reproducibility.
  * @returns A tuple `[X, y]` where X has shape `[nSamples, nFeatures]` and y has shape `[nSamples]` with dtype `int32`.
  *
- * @see {@link https://scikit-learn.org/stable/modules/generated/sklearn.datasets.make_classification.html | sklearn.datasets.make_classification}
+ * @see {@link https://deepbox.dev/docs/datasets-synthetic | Deepbox Synthetic Datasets}
  */
 declare function makeClassification(options?: {
     nSamples?: number;
@@ -106,6 +105,7 @@ declare function makeClassification(options?: {
     nInformative?: number;
     nRedundant?: number;
     nClasses?: number;
+    flipY?: number;
     randomState?: number;
 }): [Tensor, Tensor];
 /**
@@ -121,7 +121,7 @@ declare function makeClassification(options?: {
  * @param options.randomState - Seed for reproducibility.
  * @returns A tuple `[X, y]` where X has shape `[nSamples, nFeatures]` and y has shape `[nSamples]`.
  *
- * @see {@link https://scikit-learn.org/stable/modules/generated/sklearn.datasets.make_regression.html | sklearn.datasets.make_regression}
+ * @see {@link https://deepbox.dev/docs/datasets-synthetic | Deepbox Synthetic Datasets}
  */
 declare function makeRegression(options?: {
     nSamples?: number;
@@ -144,7 +144,7 @@ declare function makeRegression(options?: {
  * @param options.randomState - Seed for reproducibility.
  * @returns A tuple `[X, y]` where X has shape `[nSamples, nFeatures]` and y has shape `[nSamples]` with dtype `int32`.
  *
- * @see {@link https://scikit-learn.org/stable/modules/generated/sklearn.datasets.make_blobs.html | sklearn.datasets.make_blobs}
+ * @see {@link https://deepbox.dev/docs/datasets-synthetic | Deepbox Synthetic Datasets}
  */
 declare function makeBlobs(options?: {
     nSamples?: number;
@@ -166,7 +166,7 @@ declare function makeBlobs(options?: {
  * @param options.randomState - Seed for reproducibility.
  * @returns A tuple `[X, y]` where X has shape `[nSamples, 2]` and y has shape `[nSamples]` with dtype `int32`.
  *
- * @see {@link https://scikit-learn.org/stable/modules/generated/sklearn.datasets.make_moons.html | sklearn.datasets.make_moons}
+ * @see {@link https://deepbox.dev/docs/datasets-synthetic | Deepbox Synthetic Datasets}
  */
 declare function makeMoons(options?: {
     nSamples?: number;
@@ -187,7 +187,7 @@ declare function makeMoons(options?: {
  * @param options.randomState - Seed for reproducibility.
  * @returns A tuple `[X, y]` where X has shape `[nSamples, 2]` and y has shape `[nSamples]` with dtype `int32`.
  *
- * @see {@link https://scikit-learn.org/stable/modules/generated/sklearn.datasets.make_circles.html | sklearn.datasets.make_circles}
+ * @see {@link https://deepbox.dev/docs/datasets-synthetic | Deepbox Synthetic Datasets}
  */
 declare function makeCircles(options?: {
     nSamples?: number;
@@ -209,7 +209,7 @@ declare function makeCircles(options?: {
  * @param options.randomState - Seed for reproducibility.
  * @returns A tuple `[X, y]` where X has shape `[nSamples, nFeatures]` and y has shape `[nSamples]` with dtype `int32`.
  *
- * @see {@link https://scikit-learn.org/stable/modules/generated/sklearn.datasets.make_gaussian_quantiles.html | sklearn.datasets.make_gaussian_quantiles}
+ * @see {@link https://deepbox.dev/docs/datasets-synthetic | Deepbox Synthetic Datasets}
  */
 declare function makeGaussianQuantiles(options?: {
     nSamples?: number;
@@ -224,6 +224,7 @@ type Dataset = {
     featureNames: string[];
     targetNames?: string[];
     description: string;
+    images?: Tensor;
 };
 /**
  * Load the synthetic Iris dataset.

package/dist/{index-BndMbqsM.d.ts → index-Cjnn0KeN.d.cts}

@@ -1,7 +1,7 @@
-import { A as AnyTensor } from './index-tk4lSYod.js';
-import { D as DType, a as Device, A as Axis } from './tensor-B96jjJLQ.js';
-import { G as GradTensor } from './index-DIp_RrRt.js';
-import { T as Tensor } from './Tensor-g8mUClel.js';
+import { A as AnyTensor } from './index-DLdiQzf0.cjs';
+import { D as DType, a as Device, A as Axis } from './tensor-IlVTF0bz.cjs';
+import { G as GradTensor } from './index-GUHYEhxs.cjs';
+import { T as Tensor } from './Tensor-fxBg-TFZ.cjs';
 
 type StateEntry = {
     data: Array<number | string | bigint>;
@@ -31,7 +31,7 @@ type ForwardHook = (module: Module, inputs: AnyTensor[], output: AnyTensor) => A
  * All models should subclass this class. Modules can contain other modules,
  * allowing to nest them in a tree structure.
  *
- * This is analogous to PyTorch's nn.Module.
+ *  { https://deepbox.dev/docs/nn-module | Deepbox Module & Sequential}
  *
  * @example
  * ```ts
@@ -63,7 +63,7 @@ type ForwardHook = (module: Module, inputs: AnyTensor[], output: AnyTensor) => A
  * ```
  *
  * References:
- * - PyTorch nn.Module: https://pytorch.org/docs/stable/generated/torch.nn.Module.html
+ * - Deepbox Module: https://deepbox.dev/docs/nn-module
  *
  * @category Neural Networks
  */
@@ -362,7 +362,7 @@ declare abstract class Module {
  * ```
  *
  * References:
- * - PyTorch Sequential: https://pytorch.org/docs/stable/generated/torch.nn.Sequential.html
+ * - Deepbox Sequential: https://deepbox.dev/docs/nn-module
  * - Keras Sequential: https://keras.io/guides/sequential_model/
  *
  * @category Neural Network Containers
@@ -580,7 +580,7 @@ declare class Mish extends Module {
  * const output = mha.forward(x, x, x);
  * ```
  *
- * @see {@link https://pytorch.org/docs/stable/generated/torch.nn.MultiheadAttention.html | PyTorch MultiheadAttention}
+ * @see {@link https://deepbox.dev/docs/nn-attention | Deepbox Attention}
  * @see Vaswani et al. (2017) "Attention Is All You Need"
  */
 declare class MultiheadAttention extends Module {
@@ -651,7 +651,7 @@ declare class MultiheadAttention extends Module {
  * const output = layer.forward(x);
  * ```
  *
- * @see {@link https://pytorch.org/docs/stable/generated/torch.nn.TransformerEncoderLayer.html | PyTorch TransformerEncoderLayer}
+ * @see {@link https://deepbox.dev/docs/nn-attention | Deepbox Attention}
  */
 declare class TransformerEncoderLayer extends Module {
     private readonly dModel;
@@ -666,7 +666,19 @@ declare class TransformerEncoderLayer extends Module {
     private readonly dropout1;
     private readonly dropout2;
     private readonly dropout3;
-    constructor(dModel: number, nHead: number, dFF: number, options?: {
+    constructor(dModelOrOpts: number | {
+        readonly dModel: number;
+        readonly nHead: number;
+        readonly dimFeedforward?: number;
+        readonly dFF?: number;
+        readonly dropout?: number;
+        readonly eps?: number;
+    }, nHead?: number, dFFOrOptions?: number | {
+        readonly dimFeedforward?: number;
+        readonly dFF?: number;
+        readonly dropout?: number;
+        readonly eps?: number;
+    }, options?: {
         readonly dropout?: number;
         readonly eps?: number;
     });
@@ -692,7 +704,7 @@ declare class TransformerEncoderLayer extends Module {
  * const conv = new Conv1d(16, 33, 3); // in_channels=16, out_channels=33, kernel_size=3
  * ```
  *
- * @see {@link https://pytorch.org/docs/stable/generated/torch.nn.Conv1d.html | PyTorch Conv1d}
+ * @see {@link https://deepbox.dev/docs/nn-layers | Deepbox Layers}
  */
 declare class Conv1d extends Module {
     private readonly inChannels;
@@ -724,7 +736,7 @@ declare class Conv1d extends Module {
  * const conv = new Conv2d(3, 64, 3); // RGB to 64 channels, 3x3 kernel
  * ```
  *
- * @see {@link https://pytorch.org/docs/stable/generated/torch.nn.Conv2d.html | PyTorch Conv2d}
+ * @see {@link https://deepbox.dev/docs/nn-layers | Deepbox Layers}
  */
 declare class Conv2d extends Module {
     private readonly inChannels;
@@ -756,7 +768,7 @@ declare class Conv2d extends Module {
  * const pool = new MaxPool2d(2); // 2x2 pooling
  * ```
  *
- * @see {@link https://pytorch.org/docs/stable/generated/torch.nn.MaxPool2d.html | PyTorch MaxPool2d}
+ * @see {@link https://deepbox.dev/docs/nn-layers | Deepbox Layers}
  */
 declare class MaxPool2d extends Module {
     private readonly kernelSizeValue;
@@ -780,7 +792,7 @@ declare class MaxPool2d extends Module {
  * const pool = new AvgPool2d(2); // 2x2 pooling
  * ```
  *
- * @see {@link https://pytorch.org/docs/stable/generated/torch.nn.AvgPool2d.html | PyTorch AvgPool2d}
+ * @see {@link https://deepbox.dev/docs/nn-layers | Deepbox Layers}
  */
 declare class AvgPool2d extends Module {
     private readonly kernelSizeValue;
@@ -836,7 +848,7 @@ declare class AvgPool2d extends Module {
  *
  * References:
  * - Dropout paper: https://jmlr.org/papers/v15/srivastava14a.html
- * - PyTorch Dropout: https://pytorch.org/docs/stable/generated/torch.nn.Dropout.html
+ * - Deepbox Dropout: https://deepbox.dev/docs/nn-normalization
  *
  * @category Neural Network Layers
  */
@@ -923,7 +935,7 @@ declare class Dropout extends Module {
  * ```
  *
  * References:
- * - PyTorch Linear: https://pytorch.org/docs/stable/generated/torch.nn.Linear.html
+ * - Deepbox Linear: https://deepbox.dev/docs/nn-layers
  * - Xavier/Glorot initialization: http://proceedings.mlr.press/v9/glorot10a.html
  *
  * @category Neural Network Layers
@@ -1014,7 +1026,7 @@ declare class Linear extends Module {
  * const y = bn.forward(x);
  * ```
  *
- * @see {@link https://pytorch.org/docs/stable/generated/torch.nn.BatchNorm1d.html | PyTorch BatchNorm1d}
+ * @see {@link https://deepbox.dev/docs/nn-normalization | Deepbox Normalization & Dropout}
  */
 declare class BatchNorm1d extends Module {
     private readonly numFeatures;
@@ -1054,7 +1066,7 @@ declare class BatchNorm1d extends Module {
  * const y = ln.forward(x);
  * ```
  *
- * @see {@link https://pytorch.org/docs/stable/generated/torch.nn.LayerNorm.html | PyTorch LayerNorm}
+ * @see {@link https://deepbox.dev/docs/nn-normalization | Deepbox Normalization & Dropout}
  */
 declare class LayerNorm extends Module {
     private readonly normalizedShape;
@@ -1087,7 +1099,7 @@ declare class LayerNorm extends Module {
  * const output = rnn.forward(x);
  * ```
  *
- * @see {@link https://pytorch.org/docs/stable/generated/torch.nn.RNN.html | PyTorch RNN}
+ * @see {@link https://deepbox.dev/docs/nn-recurrent | Deepbox Recurrent Layers}
  */
 declare class RNN extends Module {
     private readonly inputSize;
@@ -1129,7 +1141,7 @@ declare class RNN extends Module {
  * - Cell state: c_t = f_t * c_{t-1} + i_t * g_t
  * - Hidden state: h_t = o_t * tanh(c_t)
  *
- * @see {@link https://pytorch.org/docs/stable/generated/torch.nn.LSTM.html | PyTorch LSTM}
+ * @see {@link https://deepbox.dev/docs/nn-recurrent | Deepbox Recurrent Layers}
  */
 declare class LSTM extends Module {
     private readonly inputSize;
@@ -1167,7 +1179,7 @@ declare class LSTM extends Module {
  * - New gate: n_t = tanh(W_in * x_t + b_in + r_t * (W_hn * h_{t-1} + b_hn))
  * - Hidden: h_t = (1 - z_t) * n_t + z_t * h_{t-1}
  *
- * @see {@link https://pytorch.org/docs/stable/generated/torch.nn.GRU.html | PyTorch GRU}
+ * @see {@link https://deepbox.dev/docs/nn-recurrent | Deepbox Recurrent Layers}
  */
 declare class GRU extends Module {
     private readonly inputSize;
@@ -1223,6 +1235,7 @@ declare class GRU extends Module {
  */
 declare function crossEntropyLoss(input: Tensor, target: Tensor): number;
 declare function crossEntropyLoss(input: GradTensor, target: AnyTensor): GradTensor;
+declare function crossEntropyLoss(input: AnyTensor, target: AnyTensor): number | GradTensor;
 /**
  * Binary Cross Entropy Loss with logits.
  *
@@ -1271,6 +1284,7 @@ declare function binaryCrossEntropyWithLogitsLoss(input: GradTensor, target: Any
  * @category Loss Functions
  */
 declare function mseLoss(predictions: Tensor, targets: Tensor, reduction?: "mean" | "sum" | "none"): Tensor;
+declare function mseLoss(predictions: GradTensor, targets: GradTensor, reduction?: "mean" | "sum" | "none"): GradTensor;
 /**
  * Mean Absolute Error (MAE) loss function, also known as L1 loss.
  *