scaffold_ai - vv2.0.1

GPU-optimized ONNX + retrieval config (TOPKFINAL=5, TOPKINITIAL=50, context 800/3000).

- Python
Published by kevinmastascusa 7 months ago

scaffold_ai - v2.0.0

GPU-optimized ONNX + retrieval config updates; includes rungpuoptimized.bat and Cloudflare launcher.

- Python
Published by kevinmastascusa 7 months ago

scaffold_ai - ScaffoldAI Enhanced UI v0.1.0

ScaffoldAI Enhanced UI v0.1.0

For Professors - Complete Step-by-Step Guide

Step 1: Download the EXE

1. Go to this GitHub release page
2. Scroll down to "Assets" section
3. Click on ScaffoldAI-EnhancedUI.exe (it will download automatically)
4. Create a new folder on your Desktop called ScaffoldAI
5. Move the downloaded EXE into the ScaffoldAI folder

Step 2: Get Your Hugging Face Token

IMPORTANT: You need your own Hugging Face token to access the TinyLlama model.

1. Go to: https://huggingface.co/
2. Click "Sign Up" (if you don't have an account) or "Sign In"
3. Create a free account (use your university email)
4. After logging in, go to: https://huggingface.co/settings/tokens
5. Click "New token"
6. Give it a name like "ScaffoldAI"
7. Select "Read" permissions
8. Click "Generate token"
9. Copy the token (it starts with "hf_") - save it somewhere safe

Step 3: Create the Environment File

1. Open Notepad (or any text editor)
2. Copy and paste this exact text (replace YOURTOKENHERE with your actual token): HUGGINGFACE_TOKEN=YOUR_TOKEN_HERE SC_LLM_KEY=tinyllama-onnx
3. Replace YOUR_TOKEN_HERE with the token you copied from Hugging Face
4. Save the file as .env (exactly this name, with the dot)
   • In Notepad: File → Save As
   • Navigate to your ScaffoldAI folder
   • In "File name" field, type: .env
   • In "Save as type" dropdown, select "All Files (.)"
   • Click Save
5. Verify: You should now have both ScaffoldAI-EnhancedUI.exe and .env in the same folder

Step 4: Run the Application

1. Double-click ScaffoldAI-EnhancedUI.exe
2. Wait for the startup messages (may take 1-2 minutes first time)
3. Look for this message: "🎉 All checks passed! Starting the Enhanced UI..."
4. Open your web browser (Chrome, Firefox, Edge, etc.)
5. Go to: http://localhost:5002
6. You should see the ScaffoldAI interface

Step 5: Upload Your Syllabus

1. In the web interface, look for the sidebar on the left
2. Scroll down in the sidebar to find "Upload Syllabus PDF"
3. Click "Choose a PDF" button
4. Select your course syllabus (must be a PDF file)
5. Wait for upload and processing (may take 30-60 seconds)
6. You'll see: "Syllabus uploaded and analyzed successfully"

Step 6: Ask Questions

1. In the main chat area, type your question
2. Example questions to try:
   • "How can I integrate sustainability into my fluid mechanics course?"
   • "What are some hands-on sustainability projects for thermodynamics?"
   • "Show me examples of sustainable design in materials engineering"
   • "How do I assess student learning of sustainability principles?"
3. Press Enter or click the send button
4. Wait for the AI response (may take 10-30 seconds)
5. Scroll down to see the sources that inform the answer

Step 7: View Sources and Continue Chat

1. After each answer, look for the "Sources" section
2. Click on source citations to see the research papers
3. Ask follow-up questions - the system remembers your syllabus
4. Use the sidebar buttons:
   • "Clear conversation" - starts fresh
   • "Clear memory" - keeps chat but resets context

Troubleshooting - Exact Solutions

Problem: "Hugging Face token not found" or "Model access denied"

Solution: 1. Make sure you created a Hugging Face account 2. Verify your token starts with "hf_" 3. Check that you have "Read" permissions on the token 4. Try accessing https://huggingface.co/TinyLlama/TinyLlama-1.1B-Chat-v1.0 in your browser to ensure you can access the model

Problem: EXE won't start

Solution: 1. Check that .env file is in the same folder as the EXE 2. Right-click the EXE → "Run as administrator" 3. Check Windows Defender: Windows Security → Virus & threat protection → Manage settings → Add or remove exclusions → Add folder → Select your ScaffoldAI folder

Problem: "Port 5002 is already in use"

Solution: 1. Open the .env file in Notepad 2. Add this line: SC_PORT=5003 3. Save the file 4. Restart the EXE 5. Go to http://localhost:5003 instead

Problem: "Model download failed"

Solution: 1. Check your internet connection 2. Verify the token in .env file is correct and starts with "hf_" 3. Make sure you're logged into Hugging Face in your browser 4. Try running the EXE again (it will retry downloading)

Problem: Browser shows "This site can't be reached"

Solution: 1. Make sure the EXE is still running (check the command window) 2. Try http://127.0.0.1:5002 instead of localhost:5002 3. Check if your antivirus is blocking the connection

System Requirements - Exact Specifications

• Operating System: Windows 10 (version 1903 or later) or Windows 11
• RAM: At least 4GB available (8GB recommended)
• Storage: 2GB free space
• Internet: Required for initial model download (about 500MB)
• Browser: Chrome, Firefox, Edge, or Safari
• Hugging Face Account: Free account required for model access

What You Can Do With This Tool

1. Upload any engineering course syllabus (PDF format)
2. Ask specific questions about integrating sustainability
3. Get research-backed answers with citations
4. View source documents that inform the responses
5. Continue conversations across multiple sessions
6. Get personalized suggestions based on your course content

Example Questions for Different Courses

• Fluid Mechanics: "How can I teach sustainable water management in fluid mechanics?"
• Thermodynamics: "What are some renewable energy examples for thermodynamics labs?"
• Materials: "How do I incorporate life cycle assessment in materials science?"
• Design: "What sustainable design principles can I add to my engineering design course?"
• Mechanics: "How can I include environmental impact analysis in statics and dynamics?"

Getting Help

• Check the command window where you ran the EXE for error messages
• Press F12 in your browser to see any web errors
• Ensure your Hugging Face token is valid and has Read permissions
• The system works best with specific, course-related questions
• If you can't access TinyLlama, try visiting the model page in your browser first

---

Important Notes: - This is a research prototype designed for educational use - All responses are based on sustainability education research literature - The system requires an internet connection for the AI models - Your syllabus and conversations are stored locally on your computer - For production use in classrooms, consider hosting on a server - You must have your own Hugging Face account and token to use this tool

- Python
Published by kevinmastascusa 7 months ago