Categories of AI frameworks and libraries
Categories
Browse libraries by field. A library may appear in more than one category when it serves more than one field.
Scientific Computing & Data Processing Libraries
Scientific Computing & Data Processing
The libraries most AI projects are built on: arrays and numerical computing, cleaning and analysing tables, charts, and big data processing.
Machine Learning Frameworks & Libraries
Machine Learning
Classical machine learning algorithms for tabular data, gradient boosting trees, class imbalance handling, hyperparameter tuning and automated machine learning.
Deep Learning Frameworks
Deep Learning Frameworks
The frameworks used to build and train neural networks: tensors, automatic differentiation, GPU and distributed training, and exporting models for deployment.
Computer Vision & Image Processing
Computer Vision & Image Processing
Image processing and data augmentation, ready-made classification, detection and segmentation models, and libraries specialised in 3D medical images.
Natural Language Processing & Language Models
NLP & LLMs
Transformers and language models, tokenization, classical NLP, Arabic in particular, semantic embeddings, and frameworks for building RAG applications and agents.
Generative AI
Generative AI
Diffusion models for image generation, parameter-efficient fine-tuning of language models, preference training, and running and serving models locally and on servers.
Graph Neural Networks
Graph AI
Libraries for graph analysis and for building graph neural networks (GNN) to classify nodes, links and whole graphs.
Reinforcement Learning
Reinforcement Learning
Standard reinforcement learning environments, reliable implementations of PPO, DQN, SAC and other algorithms, and multi-agent and distributed reinforcement learning.
Quantum Machine Learning
Quantum Machine Learning
Differentiable quantum circuit frameworks, quantum neural networks, and circuit simulation on classical processors and GPUs.
Model Deployment & MLOps
MLOps & Deployment
Experiment tracking, managing models, data and their versions, pipelines, serving models in production, and distributed computing.
Interpretability & Explainable AI
Explainable AI (XAI)
Tools that explain model decisions: Shapley values, local explanations, attribution methods for neural networks, and inherently interpretable models.
Parallel Computing & AI Acceleration
AI Acceleration & GPU Computing
GPU programming platforms, GPU arrays, compiling Python to fast code, writing kernels, and deep learning model compilers.
