The TypeScript Toolkit for AI & Numerical Computing

Comprehensive, type-safe framework unifying tensors, DataFrames, neural networks, and classical ML into a single modular package.

npm install deepboxNode ≥ 24.13
quickstart.ts
1import { tensor, add, parameter } from "deepbox/ndarray"2import { DataFrame } from "deepbox/dataframe"3import { LinearRegression } from "deepbox/ml"45// Tensor operations6const a = tensor([[1, 2], [3, 4]])7const b = tensor([[5, 6], [7, 8]])8const c = a.add(b) // [[6, 8], [10, 12]]910// Automatic differentiation11const x = parameter([2, 3])12const y = x.mul(x).sum()13y.backward() // x.grad -> tensor([4, 6])1415// DataFrames16const df = new DataFrame({17  name: ["Alice", "Bob"],18  score: [85, 90]19})2021// Machine learning22const model = new LinearRegression()23model.fit(XTrain, yTrain)
4,344+
Tests passing
Zero
Dependencies
100%
Type-safe

Complete ML Toolkit

From tensor operations to model training — everything in one package.

90+ ops

N-Dimensional Arrays

90+ tensor operations with full broadcasting and multiple dtypes support.

Autograd

Automatic Differentiation

GradTensor with full backward pass support for neural network training.

50+ ops

DataFrames & Series

Tabular API with 50+ operations for data manipulation and analysis.

Deep Learning

Neural Networks

Linear, Conv, RNN/LSTM/GRU, Attention layers with optimizers and schedulers.

Classical ML

Machine Learning

Classical ML models including Trees, SVM, KNN, Naive Bayes, and Ensembles.

40+ metrics

Statistics & Metrics

40+ ML metrics, hypothesis tests, correlations, and statistical analysis.

13 Modules, One Package

deepbox/ndarrayN-D tensors, autograd, broadcasting, 90+ ops, sparse matrices
deepbox/linalgSVD, QR, LU, Cholesky, eigendecomposition, solvers, norms
deepbox/dataframeDataFrames & Series with 50+ operations, CSV I/O
deepbox/statsDescriptive stats, correlations, hypothesis tests
deepbox/mlLinear, Ridge, Trees, SVM, KNN, ensembles, clustering, PCA
deepbox/nnLinear, Conv, RNN/LSTM/GRU, Attention, normalization, losses
deepbox/optimSGD, Adam, AdamW, RMSprop + LR schedulers
deepbox/metrics40+ classification, regression, and clustering metrics
deepbox/preprocessScalers, encoders, normalizers, CV splits
deepbox/randomDistributions (normal, uniform, gamma, etc.) + sampling
deepbox/datasetsIris, Digits, Breast Cancer + synthetic generators
deepbox/plotSVG/PNG scatter, line, bar, hist, heatmap, ML plots

Ready to get started?

Read the docs and start building with Deepbox today.

Read the Docs