4 Repos
Visual tools that display the intermediate inputs and outputs of logic nodes to inspect AI reasoning.
Distinct from Execution Trace Visualizers: Distinct from Execution Trace Visualizers: focuses on the logical state and data flow of AI nodes rather than performance flamegraphs.
Explore 4 awesome GitHub repositories matching user interface & experience · Reasoning Step Visualizers. Refine with filters or upvote what's useful.
Promptflow ist ein Entwicklungs-Framework und Orchestrator für den Aufbau von Anwendungen, die auf Large Language Models basieren. Es fungiert als Tool-Suite für das Design, die Orchestrierung und das Deployment von KI-Workflows, indem Prompts, benutzerdefinierter Python-Code und Sprachmodelle zu ausführbaren Sequenzen verknüpft werden. Das Projekt zeichnet sich durch einen visuellen KI-Workflow-Designer aus, der die Erstellung von gerichteten azyklischen Graphen (DAGs) aus Logik-Knoten ermöglicht. Es bietet eine dedizierte Prompt-Engineering-Umgebung für Versionierung und Vergleich von Templates sowie zustandsbehaftetes Execution-Tracing, um Funktionsaufrufe und Variablenwerte für schrittweises Debugging aufzuzeichnen. Die Plattform deckt ein breites Funktionsspektrum ab, einschließlich Retrieval Augmented Generation (RAG) via Vektordatenbank-Lookups und metrikgesteuerte Evaluierungspipelines für Batch-Tests und Qualitätssicherung. Sie deckt den gesamten Lebenszyklus von der Entwicklung bis zur Produktion ab, durch containerisiertes Deployment, Workflow-Endpoint-Serving und sicheres Verbindungsmanagement für API-Anmeldedaten. Ein Command-Line-Interface (CLI) und ein SDK für Workflow-Validierung und Integration in automatisierte CI/CD-Pipelines sind enthalten.
Generates visual snapshots of workflow steps to allow developers to inspect the reasoning process of the AI.
Tensorspace is a WebGL-based 3D visualization framework and renderer designed to map deep learning model architectures and tensor data into interactive three-dimensional spaces. It serves as a neural network architecture visualizer and model inspector, allowing users to render model topologies and analyze data flow within a web browser. The project distinguishes itself through its ability to convert pre-trained Keras and TensorFlow models into spatial representations. It integrates with TensorFlow.js to execute inference in the browser, enabling the real-time visualization of intermediate act
Displays internal activations and tensor states of hidden layers to visualize how outputs are generated.
ChatUI is a React conversational UI library and framework designed for building messaging interfaces. It provides a set of components for creating conversation flows, including message bubbles, input areas, and structured message hierarchies. The library distinguishes itself through specialized AI agent interface features, such as the visualization of an agent's reasoning process and simulated typing animations that render text character-by-character. It also includes a system of pre-designed conversational card templates for rendering banners, selection lists, and questionnaires within a cha
Visualizes the step-by-step reasoning process and internal logic of AI agents before delivering the final response.
This project is a comprehensive framework for constructing, managing, and evaluating knowledge graphs through multi-agent reasoning and deep search capabilities. It provides an end-to-end pipeline that ingests multi-format documents, extracts entities and relationships based on configurable schemas, and maintains structured knowledge bases to support evidence-based retrieval. The system distinguishes itself through its multi-agent orchestration, which decomposes complex queries into parallel research steps and synthesizes long-form reports. It leverages advanced graph-based techniques, includ
Visualizes the step-by-step reasoning paths and logic chains used by agents to generate answers.