Adaptive Reliability Layer is a Python project that sits between an inference pipeline and a monitoring layer. It detects distribution shift, decides whether adaptation is warranted, and learns from delayed revealed labels. It does not replace the model.
Directory
Search results
Published directory entries matching your search.
Airweave is an open-source context retrieval layer for AI agents. It aims to provide relevant context to support agent functionalities. The project is hosted on GitHub by the airweave-ai organization.
BGPT MCP is an evidence layer for scientific AI, giving models searchable access to research papers with claims linked to experiments, results, scope and provenance. It works with MCP-compatible clients and supports SSE and Streamable HTTP endpoints, plus a search API with the search_papers and lookup_paper tools.
deeplearn-rs is a Rust library for neural networks and deep learning, described by its author as mostly a proof of concept with a changing API. Implemented models include MNIST handwritten digit recognition and a char-rnn using LSTM, alongside layers such as fully connected, LSTM, softmax, and loss functions, with SGD and RMSProp optimizers. The repository was archived by its owner on Mar 8, 2024, and is read-only.
DSSTNE is an open source library from Amazon for training and deploying recommendation models with sparse inputs, fully connected hidden layers and sparse outputs. It is designed for production deployment emphasizing speed and scale, and supports multi-GPU training and prediction. The repository was archived and is read-only.
dstack is a unified orchestration layer for heterogeneous AI compute. It standardizes compute management and supports running training and inference workloads on GPU clouds, Kubernetes, virtual machines, or bare-metal clusters.
Metabase is an open source analytics and business intelligence platform. It grounds answers in a governed semantic layer of metrics and business logic so the query behind each answer can be inspected. It supports querying data in plain English with AI, interactive dashboards, documents, subscriptions and alerts, and embedded analytics. Users can self-host or use Metabase Cloud.
It implemented multi-perceptrons neural network (ニューラルネットワーク) based on Back Propagation Neural Networks (BPN) and designed unlimited-hidden-layers.
A parallel deep learning framework hosted on GitHub. The project provides a framework for building and running deep learning workflows with parallel processing support, and includes documentation for setup, available layers, and dependencies.
Neuroph is a lightweight, open source Java neural network framework released under the Apache 2.0 license. It provides a Java library and a GUI editor, Neuroph Studio, based on the NetBeans Platform, for creating, training and saving neural networks. It supports architectures including Perceptron, Multi Layer Perceptron, Hopfield, Kohonen and RBF networks, and both supervised and unsupervised learning rules.
nn_builder is a tool for building neural networks with less boilerplate code; the user specifies the network type and it builds it. It supports NN, CNN and RNN networks for PyTorch and TensorFlow 2.0, taking arguments such as layers_info, hidden_activations and dropout.
NN Visual is an interactive web-based neural network visualizer. It lets users configure hidden layers, select datasets such as auto_mpg, iris, xor, or mnist, and train a network while observing forward and backpropagation step by step. The tool displays training loss, mean absolute error, weight distributions, and node activation brightness. Written explanations accompany the visualizations to teach concepts like gradient descent, weights, biases, and activation functions without requiring code.
OpenMetadata is an open-source AI context layer powered by metadata, semantics, and memory that provides a shared, rich understanding of data for people and AI. It was built by the founders of Apache Hadoop, Apache Atlas, and Uber Databook. The unified platform supports data cataloging, discovery, quality, observability, governance, lineage, and collaboration.
PhotoDirector is a desktop photo editing application for PC and Mac with built-in generative AI functionality. It supports end-to-end photo workflows, including photo library organization, non-destructive RAW development, layer-based compositing, portrait retouching, and AI-powered content creation, with a chat-based editing option available.
Zetane Viewer is a free tool for visualizing machine learning and artificial neural network models, including architecture and internal data such as feature maps, weights, biases and layer output tensors. The viewer supports .ONNX and .ZTN files and is available for Windows, Linux and Mac. It can also be launched from existing scripts or notebooks via the Zetane Python API.