DeepTutor
AI Personalized Tutoring Private Teacher Tool
Loading…
Description
#AI Learning Assistant #Personalized Tutoring #Open Source Project #Intelligent Agent #Knowledge Management #RAG #Local Model #Ollama #Docker #PDF Q&A #Collaborative Writing #Windows #macOS #Linux
DeepTutor is an open-source AI personalized learning and tutoring platform developed by a team from the University of Hong Kong, integrating chat Q&A, automatic question generation, problem analysis, material research, knowledge management, and learning path planning into a single intelligent agent workspace.
It is not just a simple AI chatbot; it aims to provide a more coherent personalized tutoring experience by continuously understanding the user's learning content, ability changes, and goals through long-term memory, knowledge bases, and multiple learning modes.
DeepTutor offers six modes: Chat, Quiz, Research, Solve, Visualize, and Mastery Path, all sharing the same Agent engine, allowing for flexible switching between Q&A, quizzes, research, and knowledge mastery.
The project supports installation via pip or deployment via Docker and can connect to Ollama local large models. Once deployed, you can open the local interface directly in your browser to use natural language to arrange learning tasks, allowing the AI to automatically plan and execute them.
* Chat Learning: Engage in continuous dialogue with AI around courses, materials, or specific questions to receive personalized explanations and learning suggestions.
* Quiz Intelligent Testing: Automatically generates questions based on learning content, checking knowledge mastery through practice and feedback.
* Research Mode: Assists in organizing materials, breaking down research questions, and summarizing information, suitable for paper reading and thematic studies.
* Solve Problem-Solving Mode: Analyzes, reasons, and explains problems step-by-step, helping to understand problem-solving approaches rather than just providing answers.
* Visualize: Transforms complex knowledge, structural relationships, or reasoning processes into more intuitive content.
* Mastery Path Learning Path: Plans learning progress based on goals and mastery levels, gradually filling in weak knowledge points.
* Multi-layer Memory System: Retains and associates learning contexts through a three-layer memory and Memory Graph, allowing historical content to be traced back.
* Partners and My Agents: Supports creating exclusive Agents with independent Personas, skills, and memory spaces, adapting to different learning tasks.
* Local Model Access: Can connect to local models like Ollama, suitable for users concerned about privacy, offline use, or wanting to control model costs.
* Knowledge Center: Supports importing learning materials such as PDFs and utilizes multi-engine RAG for retrieval, Q&A, and content organization.
* Active Book Compilation: Allows notes and materials to be organized into knowledge books that can continue to be conversed with and updated, facilitating long-term learning.
* Co-Writer Collaborative Writing: Provides intelligent editing, content annotation, and collaborative writing capabilities, supporting Markdown export.
* TTS Voice Function: Can convert some content into speech, making it convenient to review knowledge through listening.
* Multiple Deployment Methods: Supports pip installation and Docker deployment, allowing access to the local operation interface via a browser.
* Natural Language Execution: Users only need to describe learning goals, and the AI can automatically break down tasks, plan steps, and invoke corresponding capabilities to complete them.
DeepTutor is an open-source AI personalized learning and tutoring platform developed by a team from the University of Hong Kong, integrating chat Q&A, automatic question generation, problem analysis, material research, knowledge management, and learning path planning into a single intelligent agent workspace.
It is not just a simple AI chatbot; it aims to provide a more coherent personalized tutoring experience by continuously understanding the user's learning content, ability changes, and goals through long-term memory, knowledge bases, and multiple learning modes.
DeepTutor offers six modes: Chat, Quiz, Research, Solve, Visualize, and Mastery Path, all sharing the same Agent engine, allowing for flexible switching between Q&A, quizzes, research, and knowledge mastery.
The project supports installation via pip or deployment via Docker and can connect to Ollama local large models. Once deployed, you can open the local interface directly in your browser to use natural language to arrange learning tasks, allowing the AI to automatically plan and execute them.
Software Features
* Chat Learning: Engage in continuous dialogue with AI around courses, materials, or specific questions to receive personalized explanations and learning suggestions.
* Quiz Intelligent Testing: Automatically generates questions based on learning content, checking knowledge mastery through practice and feedback.
* Research Mode: Assists in organizing materials, breaking down research questions, and summarizing information, suitable for paper reading and thematic studies.
* Solve Problem-Solving Mode: Analyzes, reasons, and explains problems step-by-step, helping to understand problem-solving approaches rather than just providing answers.
* Visualize: Transforms complex knowledge, structural relationships, or reasoning processes into more intuitive content.
* Mastery Path Learning Path: Plans learning progress based on goals and mastery levels, gradually filling in weak knowledge points.
* Multi-layer Memory System: Retains and associates learning contexts through a three-layer memory and Memory Graph, allowing historical content to be traced back.
* Partners and My Agents: Supports creating exclusive Agents with independent Personas, skills, and memory spaces, adapting to different learning tasks.
* Local Model Access: Can connect to local models like Ollama, suitable for users concerned about privacy, offline use, or wanting to control model costs.
* Knowledge Center: Supports importing learning materials such as PDFs and utilizes multi-engine RAG for retrieval, Q&A, and content organization.
* Active Book Compilation: Allows notes and materials to be organized into knowledge books that can continue to be conversed with and updated, facilitating long-term learning.
* Co-Writer Collaborative Writing: Provides intelligent editing, content annotation, and collaborative writing capabilities, supporting Markdown export.
* TTS Voice Function: Can convert some content into speech, making it convenient to review knowledge through listening.
* Multiple Deployment Methods: Supports pip installation and Docker deployment, allowing access to the local operation interface via a browser.
* Natural Language Execution: Users only need to describe learning goals, and the AI can automatically break down tasks, plan steps, and invoke corresponding capabilities to complete them.
