Packages
Runs small and large language models and multi-modal models on-device and in the cloud using ONNX Runtime as the execution backend.
Install only if you already have or plan to convert models to ONNX format; it is not a model hub or download tool.
Chatlas provides a unified Python interface for building LLM chat applications across multiple model providers with tool registration and message management.
Install it if you're building multi-provider LLM chat applications or want to avoid vendor lock-in.
Loads and manipulates SMPL-X, a unified 3D body model with 10,475 vertices and 54 joints that jointly represents the human body, face, and hands using PyTorch.
However, it has received no updates since 2021-05-26, so expect no support for newer PyTorch or Python versions, no bug fixes, and no maintenance.
Compel lets you re-weight different parts of text prompts to control how much influence each part has on the embedding tensor produced by diffusion models, using an intuitive syntax like `word++` or `word--`.
Install it if you need to adjust the relative importance of different parts of your prompts; skip it if you're happy with flat prompt strings.
Agent Lightning optimizes AI agents using reinforcement learning, prompt optimization, and fine-tuning without requiring code rewrites to your existing agent framework.
However, the unclear license, aging maintenance (233 days since last release), and 20 heavy dependencies make this a research-stage tool best suited for teams with…
Semble is a code search library that indexes and searches codebases using embeddings, returning relevant code snippets in response to natural-language queries with minimal token overhead.
Install it if you work with coding agents or need fast, token-efficient code search; the MCP integration and CLI are both straightforward to set up.
A plugin for the llm command-line tool that provides access to GitHub Models, a free service offering many AI language models via the Azure AI Inference SDK.
Install it if you want to experiment with or integrate GitHub Models into command-line workflows or GitHub Actions without cost.
LangDiff streams structured LLM outputs to frontends with granular, type-safe parsing events and automatic JSON Patch generation for efficient state synchronization.
However, the project is very new (first release August 2025) with aging maintenance status—verify that it is actively maintained and test compatibility with your LLM…
Unified installer that bundles policy enforcement, agent identity and trust infrastructure, execution supervision, and reliability monitoring into a single governance stack for production AI agents.
However, the package is in public preview with potential API changes; evaluate your tolerance for breaking changes before general availability.
ms-swift is a framework for training, evaluating, quantizing, and deploying large language models and multimodal models at scale, supporting 600+ text models and 400+ multimodal models with distributed training and inference acceleration.
Tracks machine learning experiments, trials, and trial components in AWS SageMaker training jobs, processing jobs, and notebooks.
Integrates Optuna hyperparameter optimization into Hydra's configuration framework, enabling sweeps over application parameters using Optuna's sampling and pruning strategies.
However, verify that this version is compatible with your target Hydra release (1.3 stable or 1.4 development) before committing, given the 2022 release date.
Integrates Redis with LangChain to provide vector storage, semantic caching, and chat history management for AI applications.
Install it if you are already running Redis and need vector storage, semantic caching, or chat history for a LangChain application.
MACE trains and evaluates machine learning interatomic potentials using equivariant message passing neural networks, designed for accurate molecular dynamics and materials simulations.
Install it if you need to train or deploy machine learning interatomic potentials for molecular dynamics or materials simulations.
TSDB downloads and loads 173 open-source time-series datasets with minimal setup, handling caching and basic data cleaning automatically so you can focus on analysis.
Install it if you regularly work with benchmark time-series data.
Extracts subgraphs from ONNX model files by specifying input and output operation names, generating smaller ONNX files without the overhead of standard extraction methods.
Install it if you work with ONNX models and need to extract subgraphs or partition files by operation boundaries.
Computes multiple object tracker (MOT) evaluation metrics including MOTA, MOTP, and ID measures by matching ground truth objects to predictions across video frames.
Install it if you need to evaluate or compare object trackers; skip it if your use case does not involve multi-object tracking metrics.
Evaluates language models against standardized benchmarks (MMLU, IFEval, MATH, etc.) by sending requests to model API endpoints and collecting results.
Implements motion planning algorithms including path planners for N-D grids, trajectory optimizers, path-tracking controllers, and visualization tools for robotics applications.
However, be aware of the GPLv3 copyleft requirement if you plan to integrate this into proprietary code, and verify that the specific algorithms you need are present…
Authoring, managing, and submitting Azure Machine Learning components and pipelines through a Python SDK.
Skforecast is a Python library for time series forecasting that works with any scikit-learn compatible estimator, providing tools for feature engineering, model selection, hyperparameter tuning, and backtesting.
Install it if you need to forecast time series with scikit-learn compatible estimators.
Converts text to normalized form and back (inverse normalization) for speech recognition and synthesis pipelines, using weighted finite-state transducers and language models.
Manages run configurations for machine learning experiments by combining command-line arguments, INI/JSON/Gin config files, and Weights & Biases logging into a unified interface.
Install only if you can tolerate potential incompatibilities with newer versions of pytorch-lightning, wandb, or gradio, and if you are comfortable maintaining a fork…
Jupyter AI Magics provides a JupyterLab extension that integrates AI agents into computational notebooks through a chat interface, enabling agents to read files, run commands, and interact with notebooks via the Agent Client Protocol.
Python SDK for accessing TimeGPT, a foundation model for time series forecasting and anomaly detection via API.
Connects Python to SAP HANA to build machine learning models and query data directly in the database using PAL and APL algorithms without moving data to the client.
However, the SAP DEVELOPER LICENSE AGREEMENT is non-standard and requires review; one known vulnerability (PYSEC-2023-172) should be investigated before production use.
Reimplemented data augmentation transforms for deep learning, optimized for CPU performance and designed to work with torch and numpy on individual samples rather than batches.
However, verify that the currently implemented transforms cover your use case, since the package explicitly covers only a subset of the original batchgenerators.
Provides a CLI and Python interface to download, manage, and run Meta's Llama language models, handling authentication, model listing, verification, and local checkpoint management.
However, the aging maintenance status (no release in ~9 months, last commit in Feb 2026) suggests the project may be in maintenance mode; verify that it still works…
Unified Planning provides a Python library for formulating automated planning problems and invoking planners to solve them, supporting STRIPS and RDDL problem representations.
Adds fault tolerance, checkpointing, and performance monitoring to PyTorch distributed training, detecting hung ranks, straggler GPUs, and enabling automatic recovery without reallocating compute resources.
Not recommended for single-GPU or development-only use cases.
azureml-defaults is a metapackage that bundles Azure Machine Learning runtime dependencies for inference and dataset operations.
Provides reproducible random number generation for PyTorch by seeding Python, NumPy, and all CUDA RNGs together, plus context managers to isolate or save RNG state.
Computes Krippendorff's alpha, a statistical measure of inter-rater agreement, with optimized performance for reliability data matrices.
The GPL-3.0-or-later license requires careful review if you plan to use it in proprietary software; otherwise it is a straightforward choice for academic and…
Kubeflow SDK provides Python APIs to submit and manage distributed AI training jobs, hyperparameter optimization, and ML pipelines on Kubernetes without requiring direct Kubernetes knowledge.
Install it if you need to programmatically submit AI workloads to Kubeflow; skip it if you have no Kubernetes infrastructure or prefer direct kubectl/YAML workflows.
Enables visualization of TOSA neural network model files within the Model Explorer interface, acting as an adapter plugin for the ai-edge-model-explorer framework.
Norfair adds real-time multi-object tracking to any detector by matching detected coordinates across frames using configurable distance functions and Kalman filtering.
Install it if you need real-time multi-object tracking in video and already have a detector.
Parses TensorFlow Lite (*.tflite) model files and provides a Python API to inspect their structure, operators, and metadata.
Provides utility functions for PyTorch development, including random number generation control and visualization tools.
PyPOTS provides machine learning algorithms for time series with missing values, supporting imputation, classification, clustering, forecasting, and anomaly detection on partially-observed multivariate data.
CTGAN generates synthetic tabular data by training deep learning models (CTGAN and TVAE) on real datasets, learning patterns to produce high-fidelity synthetic records suitable for testing, privacy-preserving data sharing, or augmentation.
However, verify that the Business Source License aligns with your use case (commercial use may be restricted), and consider the SDV wrapper library if you need…