haifengl/smile

Features

• JVM Machine Learning Libraries - Provides a comprehensive suite of machine learning algorithms implemented natively for Java, Scala, and Kotlin.
• GPU Acceleration - Accelerates neural network training and inference using native LibTorch bindings for GPU computation.
• Inference Engines - Executes text generation and chat completions using BPE tokenization and streaming REST APIs.
• Natural Language Processing - Implements core linguistic analysis tasks including tokenization, stemming, part-of-speech tagging, and keyword extraction.
• Natural Language Processing Libraries - Offers a full NLP library for tokenization, stemming, part-of-speech tagging, and keyword extraction.
• Neural Network Training Toolkits - Develops deep learning models using GPU-accelerated tensors and standard neural network layer types.
• Deep Learning Acceleration - Utilizes LibTorch bindings to accelerate tensor operations and neural network training on GPUs.
• Linear Algebra - Provides high-performance mathematical routines for vector and matrix operations via BLAS and LAPACK interfaces.
• Numerical Computing - Provides a mathematical toolkit for linear algebra, numerical optimization, and statistical computing on the JVM.
• Statistical Analysis Libraries - Provides a toolkit for statistical analysis, hypothesis testing, and probability distributions on the JVM.
• Native-Free Implementations - Implements statistical and machine learning logic directly in Java and Scala to eliminate external native dependencies.
• Association Rule Learning - Implements algorithms for discovering frequent itemsets and association rules within large datasets.
• Spectral Clustering - Implements spectral clustering using the eigenvalues of a similarity matrix for dimensionality reduction and grouping.
• Dimensionality Reduction Techniques - Provides algorithms for reducing the number of variables in a dataset while preserving essential structural information.
• Genetic Algorithms - Applies genetic algorithms to evolve candidate solutions for complex optimization problems.
• Keyword and Phrase Extraction - Identifies significant terms and multi-word phrases from text using TF-IDF and BM25.
• Clustering Algorithms - Groups similar data points into clusters based on shared characteristics using various unsupervised algorithms.
• Kernel Methods - Implements kernel functions for mapping data into higher-dimensional spaces, specifically for support vector machines.
• Large Language Model Serving - Runs text generation and chat completions using BPE tokenization and an OpenAI-compatible REST API.
• Linear Regression - Implements statistical methods for modeling relationships between variables using linear equations.
• Machine Learning Classification - Trains models to categorize data into predefined classes using a variety of supervised classification algorithms.
• ONNX Model Exporters - Supports exporting and importing machine learning models using the standardized ONNX format for cross-framework compatibility.
• Tensor Memory Management - Automatically allocates and reclaims GPU tensor memory to prevent leaks during deep learning computation.
• Byte Pair Encodings - Implements Byte-Pair Encoding for subword tokenization, ensuring compatibility with large language model vocabularies.
• Manifold Learning Algorithms - Implements algorithms that model the underlying manifold of high-dimensional data to identify topological structures.
• Hyperparameter Optimization - Uses genetic algorithms to automate the search and selection of optimal configuration parameters for models.
• Model Performance Evaluators - Quantifies model accuracy and reliability using metrics like AUC and RMSE via cross-validation.
• Model Serialization - Saves trained machine learning models to disk for deployment and pipeline integration.
• OpenAI-Compatible Model Servers - Hosts trained models via an OpenAI-compatible REST API with server-sent event streaming.
• Word Stemming - Reduces words to their root forms using standard stemming and lemmatization algorithms.
• Non-Linear Regression - Trains decision trees or neural networks to model complex non-linear relationships for continuous prediction.
• Part-of-Speech Taggers - Assigns grammatical tags to words in sentences using Hidden Markov Model taggers.
• Sequence Model Training - Fits Hidden Markov Models or Conditional Random Fields to sequential data for tagging tasks.
• Sequential Learning - Provides methods for training models on ordered data sequences, including HMMs and CRFs for sequence labeling.
• Byte-Level Encoders - Encodes text into tokens using byte-level Byte-Pair Encoding for language model pipelines.
• Time Series Forecasting - Implements models and architectures for predicting future values in temporal data sequences via autocorrelation analysis.
• Text Preprocessing - Extracts meaning from text through sentence splitting, tokenization, stemming, and tagging.
• Swing-Based Renderers - Renders interactive 2D and 3D plots using the Java Swing framework for desktop data exploration.
• Interactive Data Science Environments - Offers a desktop environment with an interactive shell for exploratory data analysis and model training.
• Interactive REPLs - Provides an interactive Java and Scala REPL with pre-imported packages for rapid experimentation.
• LLaMA Model Support - Generates text responses from LLaMA-3 models with support for chat and streaming API serving.
• Interactive Scientific Plot Constructors - Generates interactive 2D and 3D scatter plots and histograms for scientific data exploration.
• Idiomatic API Wrappers - Provides concise, idiomatic Scala, Kotlin, and Java wrappers for accessing machine learning algorithms.
• Random Number Generation - Provides high-quality pseudorandom number generation as an alternative to standard library generators.
• Distance Metrics - Implements algorithms for calculating distances between elements, including Euclidean, Mahalanobis, Hamming, and edit distances.
• Nearest Neighbor Searches - Provides efficient algorithms for finding the closest points in multi-dimensional datasets using spatial structures like k-d trees.
• Numerical Function Optimization - Provides numerical function optimization using BFGS and L-BFGS algorithms.
• Probability Distributions - Implements various probability distributions such as Normal, Poisson, and Beta for statistical analysis.
• Numerical Problem Solving - Provides random number generators and unconstrained optimization for mathematical functions.
• Hypothesis Testing - Conducts statistical hypothesis tests including t-tests, chi-squared, and ANOVA to validate assumptions.
• Data Plotting Components - Renders interactive 2D and 3D statistical plots using the Java Swing framework.
• Deep Learning Frameworks - Functions as a deep learning framework featuring GPU acceleration via LibTorch and ONNX model interchange support.
• Machine Learning - Statistical Machine Intelligence & Learning Engine.
• Machine Learning Libraries - Comprehensive set of pure Java libraries for statistical machine learning.
• Science and Data Analysis - Statistical machine intelligence and learning engine.

Star history