Tag

Keras

Builds and trains deep learning models from one Python API across JAX, TensorFlow, PyTorch, and OpenVINO inference. Its real value is portability: model code, custom layers, and data pipelines can move across backends instead of locking into one stack.

scikit-learn: machine learning in Python — scikit-learn 1.8.0 documentation

Provides a consistent Python API for classical machine learning, covering preprocessing, model selection, supervised and unsupervised estimators, and pipelines. Best for tabular, text, and medium-scale in-memory workflows.

NumPy

The N-dimensional array (ndarray) underpinning Python's scientific stack — pandas, scikit-learn, and SciPy build directly on it. Vectorized math, broadcasting, and a C/Fortran bridge move numeric work out of Python loops into compiled code.

pgmpy

Provides APIs to build, learn, and run Bayesian and dynamic Bayesian networks, perform probabilistic inference, and compute interventional/counterfactual queries. Ships example notebooks, tutorials, and PyPI/conda packages. ([github.com](https://github.com/pgmpy/pgmpy))

TensorFlow

Builds and deploys machine learning models across research, production, web, mobile, and edge environments. Its ecosystem spans Keras, TFX, LiteRT, TensorFlow.js, datasets, model hubs, and visualization tools.

ONNX

Defines a portable model format and operator set for moving trained machine learning models across frameworks, runtimes, and hardware targets without locking the model to one toolchain.

Ray (by Anyscale)

Scales any Python or ML workload across CPUs and GPUs with a few decorators, instead of rewriting code for Spark or MPI. Bundles libraries for distributed training, hyperparameter tuning, RL, batch inference, and online model serving on one cluster.

OpenVINO

Converts, quantizes, and runs deep learning models from PyTorch, TensorFlow, ONNX, and PaddlePaddle across Intel CPUs, GPUs, and NPUs without the training framework. Adds a GenAI pipeline for LLMs plus Hugging Face, vLLM, and LangChain integrations.

JAX

Provides composable function transformations for Python+NumPy—automatic differentiation, JIT compilation, and vectorization—while using XLA to run and scale numerical programs on GPUs/TPUs for ML and scientific computing.

PennyLane

Differentiable programming framework for quantum computers: build variational circuits, compute their gradients alongside PyTorch, TensorFlow, or JAX, and run identical code on simulators or real hardware via IBM, AWS Braket, and Google plugins.

PyTorch Lightning

Turns raw PyTorch training loops into structured modules that scale from a laptop to multi-node GPUs without rewriting model logic. It handles precision, checkpointing, logging, and distributed execution while preserving PyTorch control.

NVIDIA NeMo

Build, fine-tune, and deploy speech AI on NVIDIA GPUs: ASR, text-to-speech, and speech LLMs in one PyTorch stack. Ships pretrained Parakeet/Canary recognition and Magpie TTS checkpoints; broader LLM/multimodal training now lives in v2.7.0.