Packages
Ingest Spark DataFrames into Amazon SageMaker FeatureStore's online and offline stores, and load feature definitions from schema.
tfx_bsl provides shared libraries and utilities used internally by TensorFlow eXtended (TFX) components and standalone TFX libraries like TFDV, TFMA, and TFT.
Lion is a PyTorch optimizer that implements an evolved sign momentum algorithm discovered by Google Brain as an alternative to AdamW for training neural networks.
Install it if you are training models in domains the paper evaluated (language modeling, vision transformers, text-to-image) and are willing to invest in…
TensorFlow CPU for AWS is a machine learning framework optimized for numerical computation on CPU-based AWS infrastructure, supporting deployment across diverse platforms and devices.
However, verify that this AWS-specific variant offers advantages over the standard TensorFlow CPU distribution, and check whether version 2.15.1 (released 2024-03-14)…
Provides utility functions and helpers for the Lightly self-supervised learning framework, supporting computer vision model pretraining with numpy and Pillow.
However, verify that this utility package is still the recommended approach for your Lightly version, as the last release was in 2021.
Guardrails is a Python framework that validates and constrains LLM inputs and outputs by running configurable guards, and helps generate structured data from language models.
Mooncake Transfer Engine provides high-performance, zero-copy RDMA-based data transfer for distributed KV cache management in large-scale LLM inference and training systems.
A gRPC client library for the Clarifai AI platform, enabling Python developers to call computer vision and natural language processing models via the Clarifai API.
Clingo is a Python binding to the clingo Answer Set Programming solver, allowing you to model and solve combinatorial logic problems by writing logic programs and computing their answer sets.
Provides a unified Python interface to communicate with multiple LLM providers (OpenAI, Anthropic, Mistral, Ollama, and others) using a single API, letting you switch providers by changing a string parameter.
Install it if you need to support multiple LLM providers or want the flexibility to switch between them without code changes.
fast-agent-mcp is a CLI-first framework for building and running LLM agents with Model Context Protocol (MCP) server integration, supporting multiple LLM providers and interactive or programmatic execution.
Install only if you need agent orchestration; it is not suitable for simple LLM API calls.
Integrates MCP (Model Context Protocol) servers with NVIDIA NeMo Agent Toolkit, allowing agent workflows to discover and call external MCP tools as functions.
Analyzes PyTorch Profiler traces to identify performance bottlenecks in distributed training workloads by breaking down GPU time, kernel execution, communication overlap, and idle periods.
Install it if you profile distributed training workloads and need structured analysis of PyTorch Profiler traces.
nnU-Net is a semantic segmentation framework that automatically configures U-Net variants based on dataset characteristics and provides end-to-end workflows for preprocessing, training, model selection, and inference on 2D and 3D images.
Provides pretrained PyTorch models for face detection using MTCNN and face recognition using Inception ResNet V1, with automatic model downloading and caching.
Install it if you need pretrained face detection and recognition models and can accept that bug fixes or compatibility updates may lag behind new PyTorch releases.
Teradata ModelOps Client provides a CLI and SDK for managing machine learning model lifecycle—training, evaluation, deployment, and versioning—within Teradata's data platform.
However, the proprietary license restricts use to internal purposes tied to a Teradata database license, and Windows users must manually install OpenSSL.
ColPali-engine provides training and inference code for vision-language document retrieval models that create multi-vector embeddings from document images using visual transformers, enabling efficient document search without OCR.
Aigie is the Python SDK for Kytte, a runtime reliability platform that detects and autonomously remediates failures in agentic AI systems during execution.
Provides a Model Context Protocol (MCP) server that connects AI assistants like Claude to a Nextcloud instance, enabling them to read and write notes, calendar events, contacts, files, and other Nextcloud app data through a unified API.
Builds multi-agent swarm systems where specialized agents hand off control to each other dynamically, with built-in support for memory and streaming via LangGraph.
However, the aging maintenance status (253 days since last release) means you should verify compatibility with your LangGraph version and be prepared for slower bug…
Computes Fréchet Inception Distance (FID), a metric for measuring similarity between two image datasets using PyTorch and the Inception network.
However, maintenance is dormant (last release January 2023); verify current compatibility with your PyTorch version first.
A collection of command-line tools for transforming ONNX models: merging, splitting, deleting nodes, compressing size, modifying attributes and constants, changing opsets, converting channel formats, adjusting batch sizes, and converting to/from JSON.
Builds and orchestrates machine learning workflows on AWS Step Functions and SageMaker directly from Python code, without manual AWS service provisioning.
CartoBoost is a Rust-backed gradient boosting library for regression, classification, and forecasting that incorporates spatial, temporal, and structured relationship patterns into tree-based models.
Not recommended if place/time structure is irrelevant to your problem or if a simple interpretable model already answers your question.
Python SDK for building and running Vibe agents with stateful async/sync sessions, Pydantic-based tool authoring, and MCP server integration.
However, verify the license status before production use, and confirm Python 3.12+ is available in your environment.
Bridges LlamaIndex with LangChain LLM providers, allowing you to use any LangChain-compatible language model within LlamaIndex's indexing and retrieval framework.
Wraps XGBoost with weighted and focal loss functions to handle binary classification on imbalanced datasets, computing the required gradients automatically.
However, dormancy since 2021 means no active maintenance, so verify it works with your current environment before relying on it in production.
Integrates FAISS vector storage with LlamaIndex for efficient similarity search and retrieval in RAG applications.
Install it if you are building a LlamaIndex application and want to use FAISS for vector storage and retrieval.
Optimum Intel bridges Hugging Face Transformers and Diffusers models to OpenVINO, enabling model export, quantization, and accelerated inference on Intel CPUs, GPUs, and specialized accelerators.
Install it if you need to optimize and accelerate Transformers or Diffusers models on Intel CPUs, GPUs, or accelerators; skip it if you have no Intel hardware target…
Provides a PDF display component for Gradio applications, allowing you to render PDF files in web-based UIs.
batchgenerators provides data augmentation transforms for 2D and 3D images, including spatial transformations, color adjustments, noise injection, and cropping, designed for medical image analysis and deep learning pipelines.
Integrates Oracle Database with LangChain to enable vector search, document loading, text splitting, and embedding generation for building retrieval-augmented generation (RAG) pipelines.
Integrates Docling document conversion with LangChain, enabling you to load and process documents (PDFs, images, etc.) into LangChain-compatible formats for use in language model pipelines.
Install it if you're building a LangChain application that needs to ingest and process documents—either locally (with the `local` extra) or via a remote Docling…
Letta is a framework for building LLM agents with persistent memory that can learn and improve over time, accessible both as a local CLI tool and through a Python API for integration into applications.
However, the large dependency tree and requirement for an external API key (for the cloud API) or Node.js 18+ (for the CLI) may be friction points.
Provides inference components and utilities for generative AI applications, including model invocation, data processing, and integration with LLM frameworks.
However, verify the license terms first and confirm that the binary-only distribution meets your source-code requirements.
Provides debugging and instrumentation APIs for LLM training workflows using NVIDIA libraries (Transformer Engine, Megatron-LM, NeMo) and PyTorch models, with configurable layer-level feature selection and tensor statistics collection.
However, the limited repository activity (21 stars, last commit 2025-09-17) and aging status suggest this is not a heavily supported tool—verify compatibility with…
Provides a unified Python interface for making predictions using DataRobot's prediction methods, abstracting away implementation details to simplify model inference.
However, the aging maintenance status means you should verify that the library supports your specific prediction methods and be prepared for slower issue resolution…
Transformer Engine accelerates Transformer model training and inference on NVIDIA GPUs using low-precision formats including FP8, MXFP8, and NVFP4 to reduce memory use and improve performance.
A PyTorch model trainer framework that handles training loops, optimization, mixed precision, distributed training, and experiment logging with minimal boilerplate.
Optimizes GPU memory usage for running large generative models on consumer GPUs with limited VRAM by managing model loading, quantization, and offloading.
However, verify the unclear license status before use in proprietary work, and test your specific model and hardware combination first.