Packages
Triton is a language and compiler for writing custom deep-learning operations that compile to efficient GPU and CPU code, offering higher productivity than CUDA with greater flexibility than other domain-specific languages.
Not recommended if you only use standard framework operations or lack GPU hardware for testing.
A fast, Rust-backed tokenizer library that encodes and decodes text using BPE, WordPiece, and Unigram algorithms, compatible with HuggingFace tokenizer models.
Adds Theory of Mind capabilities to software engineering agents, enabling them to understand and adapt to individual user preferences, working styles, and intent through LLM-powered user modeling and consultation.
Tombi is a TOML formatter, linter, and language server that provides command-line tools and IDE integration for working with TOML files.
Install it if you work regularly with TOML files and want a unified tool for formatting and validation.
Speeds up Stable Diffusion image generation by merging redundant tokens in the transformer, reducing computation and memory without requiring model retraining.
Parses and generates TOML configuration files, converting between TOML-formatted text and Python dictionaries.
Install it if you need to read or write TOML files in Python.
Command-line tool for reading, writing, and querying TOML files without opening an editor, using JMESPath for complex lookups.
Provides shared Python utilities and logic for TOML formatting tools within the toml-fmt project ecosystem.
Install only if you are developing a formatter within the toml-fmt project ecosystem.
A high-performance TOML parser and serializer for Python, written in Rust, with drop-in compatibility for most tomllib use cases and support for TOML v1.0.0 and v1.1.0.
Command-line utility and Python library to sort and format TOML files while preserving comments and standardizing whitespace.
Tomli parses TOML configuration files and strings into Python dictionaries, with optional support for custom float handling and compatibility with Python's standard library tomllib module.
Install it if you need to read TOML on Python versions before 3.11, or if you want mypyc-compiled performance on any version.
Tomli-W serializes Python dictionaries to TOML format strings or files, providing a write-only counterpart to TOML parsing.
Install it if you need to write TOML files from Python.
Parses and edits TOML files while preserving formatting, comments, and structure, then serializes them back with layout intact.
Install it if you're building tools that touch TOML files and user readability of the source matters.
Reads and writes TOML files while preserving formatting, comments, whitespace, and string styles exactly as they appear in the original document.
Low-level Python library for interacting with the TON blockchain, including wallet creation, message construction, and token transfers.
However, maintenance is dormant (766 days since last release), so verify compatibility with your target TON network version and be prepared to handle the acknowledged…
Parses and formats RFC 3339 date-time strings according to the Internet Date/Time Format specification, providing a UTC tzinfo implementation and conversion routines for datetime objects.
A compact modular framework for building WSGI web applications, console interfaces, and configuration management, last updated in 2011.
Enables Google ADK agents to natively use tools from the MCP Toolbox through a Python SDK integration layer.
Install it if you are building Google ADK agents and need access to MCP Toolbox tools.
A Python SDK for connecting to and invoking tools from a running MCP Toolbox service, loading tool definitions as callable Python functions for use in GenAI applications.
Generates deterministic Python code that enforces business policy constraints on AI agent tool invocations before they execute, preventing policy violations at the tool level.
A collection of Python utility functions for common programming tasks, including pickle serialization, data manipulation, and optional API-specific tools for libraries like scikit-learn and PyTorch.
ToolUniverse provides a standardized interface for AI agents and LLMs to discover, compose, and execute over 1000 scientific tools, datasets, APIs, and ML models through a unified protocol.
However, the license treatment is unclear—verify the actual license before use in commercial projects.
Toolz provides utility functions for functional-style programming with iterators, functions, and dictionaries—operations like groupby, unique, memoize, curry, compose, and merge.
The aging maintenance status is not a blocker—the team commits to critical fixes—but it signals the library is feature-complete and not actively developed; install it…
Encodes and decodes data structures using TOON (Token-Oriented Object Notation), a compact format designed to reduce token usage when passing structured data to language models.
toons is a Rust-backed Python parser and serializer for the TOON (Token Oriented Object Notation) format, a token-efficient data serialization designed for LLM contexts, with an API mirroring the standard json module.
Tooz provides a coordination API for distributed systems, centralizing common primitives like group membership, lock services, and leader election.
Install it if you're building distributed applications that need coordination primitives.
topojson encodes spatial data into TopoJSON format, a compressed topology representation that reduces file size by eliminating coordinate redundancy and applying fixed-precision integer encoding and arc simplification.
Implements topological sorting of directed acyclic graphs, ordering nodes so dependencies are processed before dependents.
Computes time-optimal path parameterization for robots subject to kinematic and dynamic constraints, converting a geometric path into the fastest feasible trajectory.
However, note the deprecation warning: the project is transitioning to a C++ version with bindings, so for new long-term projects, evaluate whether that alternative…
PyTorch provides GPU-accelerated tensor computation and automatic differentiation for building and training deep neural networks in Python.
Provides PyTorch-native audio data augmentation transforms that run on CPU or GPU, designed to integrate directly into neural network models as differentiable modules.
However, be aware that maintenance is aging (last release 576 days ago), multiprocessing and multi-GPU setups have known limitations, and some transforms have edge…
Provides an ahead-of-time compiled module for faster DLPack v1.2 conversion with torch, avoiding JIT compilation overhead and compiler toolchain requirements.
However, maintenance is aging (last release 214 days ago), so verify compatibility with your specific torch and library versions before relying on it in production.
Provides a Python class wrapping real and imaginary PyTorch tensors to simulate complex tensor operations until native PyTorch support arrives.
Install only if you are locked into an old PyTorch version that lacks complex tensor support and cannot upgrade.
Enables PyTorch to use DirectML for GPU acceleration on DirectX 12-compatible hardware from AMD, Intel, NVIDIA, and Qualcomm.
However, operator coverage is still developing—verify that your specific PyTorch operations are supported via the operator roadmap before committing to production use.
Provides utility functions for PyTorch and einops to accelerate tensor manipulation in ML workflows.
Computes exponential moving averages of PyTorch model parameters during training, allowing you to evaluate models using smoothed weights without modifying the original training state.
However, maintenance is dormant (last release 2021-11-17); test compatibility with your PyTorch version before production use, and monitor the repository for any…
PyTorch Geometric is a library for building and training Graph Neural Networks (GNNs) on structured data, providing pre-built GNN layers, datasets, data loaders, and utilities for geometric deep learning.
It is the standard library for GNN work in PyTorch—install it if you need to build or train graph neural networks.
Torch Model Archiver creates .mar archive files from trained neural network models for deployment with TorchServe inference serving.
However, approach with caution for new projects: the repository is archived and maintenance is abandoned, so security patches and compatibility updates with future…
Enables PyTorch to run computations on Huawei Ascend NPU hardware, bridging PyTorch's tensor operations to Ascend AI Processors.
Provides a collection of alternative optimization algorithms for PyTorch that are compatible with the standard optim module interface.