Machine Learning & Deep Learning Packages
Machine-learning and deep-learning frameworks and model libraries for training, fine-tuning, and serving models (PyTorch, scikit-learn, Transformers).
Packages in Machine Learning & Deep Learning
Kind
Sort:Used by Most Apps
Comfy Kitchen
A fast PyTorch kernel library for diffusion inference with multiple compute backends.
OpTree
Optimized C++-accelerated utilities for flattening, mapping, and transforming nested Python pytree structures.
Label Studio SDK
The official Python client for the Label Studio API, covering projects, tasks, annotations, ML backends, and enterprise workspace management.
wink-nlp-utils
A collection of composable NLP text-preparation utilities for tokenizing, stemming, and cleaning strings and tokens in Node.js.
transformers-stream-generator
A tiny add-on that makes Hugging Face Transformers' generate() stream tokens one at a time as a Python generator.
weasel
A small workflow system for orchestrating ML training, packaging, and serving pipelines
hannoy
A production-ready, key-value-backed HNSW vector search implementation in Rust using LMDB.
spacy-pkuseg
A multi-domain Chinese word segmentation toolkit, packaged for easy installation with spaCy
unidic-lite
A pip-installable, trimmed-down UniDic 2.1.2 Japanese dictionary for MeCab-based tokenizers.
regression-simple-linear
Ordinary least squares simple linear regression for JavaScript and TypeScript, with fit, predict, scoring, and JSON serialization built in.
regression-polynomial
Fit polynomial regression models to x/y data in JavaScript or TypeScript, with support for sparse power arrays and zero-intercept fitting.
tokenizers
Multi-arch Node.js bindings for HuggingFace's Rust tokenizers library, supporting more platforms and Node versions than the official package.
spacy-legacy
Backwards-compatible legacy architectures and registered functions for spaCy v3.x
PyMuPDF Layout
A CPU-only PDF layout analysis library that trains a graph neural network on PDF internals instead of rendered images, extracting Markdown, JSON, or text far faster than vision-based models.
turbovec
A Rust vector index with Python bindings that compresses embeddings to 2-4 bits and searches faster than FAISS.
