Packages
Runs OpenAI's Whisper speech recognition models on Apple silicon using MLX, with support for models ranging from 39 million to 1.5 billion parameters.
APLR builds interpretable regression and classification models using automatic piecewise linear regression, offering smoother predictions than tree-based methods while maintaining competitive accuracy.
Integrates Qdrant vector database with LlamaIndex for storing and retrieving embeddings in RAG and semantic search applications.
Provides Python bindings to call Microsoft Azure's Computer Vision API for image analysis tasks like object detection, OCR, image tagging, and description generation.
Install only if maintaining legacy code that cannot yet be refactored.
Manages cryptographic keys and wallets for the Bittensor network, providing Python bindings to a Rust-based wallet implementation for creating and accessing blockchain credentials.
Install only if you are maintaining legacy code that explicitly depends on bittensor-wallet.
Feature-engine provides transformers for engineering, selecting, and preprocessing features in machine learning pipelines, following scikit-learn's fit/transform interface.
torchtune is a PyTorch library for fine-tuning, post-training, and experimenting with large language models using recipes for SFT, knowledge distillation, DPO, PPO, GRPO, and quantization-aware training.
Integrates Hugging Face embedding models with LlamaIndex, allowing you to use transformer-based embeddings from Hugging Face within LlamaIndex applications.
Integrates BM25 full-text search retrieval into LlamaIndex applications, enabling keyword-based document ranking and retrieval without external dependencies.
Install only if BM25 ranking fits your retrieval needs; for semantic or dense vector search, you'd want a different retriever.
Integrates Weaviate vector database with LangChain, enabling LangChain applications to store and retrieve embeddings in Weaviate.
Resemblyzer generates a 256-value embedding that summarizes voice characteristics from audio, enabling speaker verification, diarization, and voice similarity comparisons through a pretrained deep learning model.
A Gradio custom component that adds a dual-handle range slider to select minimum and maximum values within a specified interval.
However, verify that your Gradio version is compatible and accept that no active support or updates are forthcoming—test thoroughly before production use, especially…
EnCodec is a neural audio codec that compresses audio to low bitrates while preserving high fidelity, with separate models for 24 kHz mono and 48 kHz stereo audio.
Inspect SWE provides a suite of software engineering agents built on Inspect AI, enabling automated evaluation and testing of code generation and engineering tasks.
Accelerates large language model inference on Intel hardware (GPU, NPU, CPU) with low-bit quantization and integration into popular LLM frameworks.
Implements multiple residual streams in neural networks as proposed in the Hyper-Connections paper, allowing you to split, process, and recombine residual pathways within PyTorch models.
Imports datasets from the UC Irvine Machine Learning Repository directly into Python scripts and notebooks, returning data as pandas dataframes along with structured metadata.
Integrates training scripts into Docker containers for use with Amazon SageMaker, handling environment setup, hyperparameter passing, and entry point execution within containerized training jobs.
Computes edit distance, alignment, and word error rate (WER) between sequences using Kaldi's original algorithms, with support for compound word matching and statistical confidence intervals.
Adds task planning and tracking tools to Pydantic AI agents, enabling them to create, manage, and complete hierarchical tasks with subtasks, dependencies, and persistent storage.
NeuralForecast provides a collection of neural network models for time series forecasting, including RNNs, LSTMs, Transformers, and specialized architectures like NBEATS and NHITS, with sklearn-like fit/predict syntax.
Removes backgrounds from images and videos using a deep learning model (InSPyReNet), producing transparent or custom-background output via command-line, Python API, or GUI.
Generates high-quality synthetic datasets from scratch or seed data, with control over field relationships, statistical distributions, and built-in validation and quality scoring.
Provides a model serving stack for deploying machine learning models in Docker containers on Amazon SageMaker, built on Multi Model Server.
However, the abandoned repository status (last commit 2023-11-20) means you should verify compatibility with your target SageMaker version and Python runtime before…
Automatically discovers and trains the best machine learning model and hyperparameters for your dataset, handling model selection and tuning without manual experimentation.
tslearn provides machine learning algorithms optimized for time series data, including clustering, classification, regression, and distance metrics like Dynamic Time Warping, built on scikit-learn's API.
Provides a configuration API for building synthetic data generation pipelines in the NeMo Data Designer framework, allowing you to define data sources, LLM models, and column generation rules.
Execution engine for the NeMo Data Designer synthetic data generation framework, handling data transformation, generation, and LLM integration for creating synthetic datasets.
Implements local windowed attention for transformer models, allowing efficient attention computation over fixed-size windows rather than full sequences.
However, the 394-day gap since the last release and aging maintenance status suggest it is stable but not actively developed—suitable for production use in…
PettingZoo provides multi-agent reinforcement learning environments with a Gymnasium-like API, supporting both sequential (AEC) and parallel action modes across Atari, classic games, and custom environments.
Install it if you need a multi-agent environment interface; use base install for API exploration and add extras only for specific environment families.
Integrates Azure AI services, particularly Azure Speech, into LiveKit Agents for real-time voice and audio processing in agent applications.
pyAgrum is a Python library for creating, learning, and performing inference on Bayesian Networks and other Probabilistic Graphical Models, with a C++ backend and high-level Python interface.
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…
Automatically generates and assigns operation names to unnamed operators in ONNX model files, converting old-format ONNX graphs to have properly named operations.
AutoGluon automates machine learning model training and deployment across tabular, time series, image, text, and multimodal data with minimal code.
Install it if you need to train accurate ML models quickly across tabular, time series, or multimodal data without manual tuning.
ProbLog is a probabilistic logic programming toolbox that combines logic programming with probability annotations to perform inference tasks like computing marginals given evidence and learning from interpretations.
Install it if you need to combine logic programming with probabilistic reasoning; skip it if you're looking for general-purpose Bayesian inference or constraint solving.
Extends scikit-learn with additional estimators and preprocessing tools designed to support SageMaker Autopilot, including dimension reduction, feature extraction, imputation, and encoding transformers.
SeqIO builds scalable data pipelines for sequence models using TensorFlow's tf.data.Dataset, with support for preprocessing, tokenization, and evaluation metrics, while remaining compatible with JAX, PyTorch, and other frameworks.
However, the 11 runtime dependencies—particularly TensorFlow, JAX, and TensorFlow Text—make it heavy for lightweight use cases; consider it only if you need…
Computes fast, differentiable SSIM and MS-SSIM metrics for PyTorch tensors using separable Gaussian kernels, enabling efficient image quality assessment and loss functions for neural networks.
AutoRound quantizes large language models and vision-language models to 2–4 bits with minimal accuracy loss, using sign-gradient descent and supporting multiple export formats and inference backends.