Apply for community grant: Personal project (gpu)
I'm building Mathetics AI—a step-by-step mathematics tutor that helps students understand concepts, not just get answers.
Core Features:
• Solves algebra, calculus, geometry, and competition math (AMC/AIME level)
• Provides detailed explanations with LaTeX formatting
• Uses multiple solution methods + answer verification
• Powered by Qwen2.5-Math-7B (outperforms GPT-4o on math benchmarks)
Current Challenges (Free Tier):
• 2 vCPU/16GB RAM causes 30s+ response times for complex problems
• Streaming LaTeX rendering lags on mobile devices
• Limited to ~3 concurrent users before 503 errors
GPU Grant Request:
A T4 Small GPU would enable:
• 3s response times (vs 30s+ current)
• Smooth LaTeX rendering for 15+ concurrent students
• Real-time dataset streaming from GSM8K + educational corpora
Community Impact:
• Free math tutoring for underserved students worldwide
• Open-source code + datasets for educational AI research
• Active development roadmap: Image OCR, progress tracking, PDF export
Technical Details:
• Qwen2.5-Math-7B-Instruct + Gradio + Hugging Face Spaces
• Live demo: https://khoi1234210-mathematicsai.hf.space/
• GitHub: [link if available]
The 183% Story: Loading the 7B model locally hit 183% CPU usage—kernel ascended to silicon nirvana. Now stable in the cloud, but GPU acceleration would unlock full potential.
Seeking T4 Small GPU for 3 months to scale from demo to classroom tool. Happy to share progress updates and contribute back to the community!
Try it now: Ask "derivative of x²sin(x)" → Gets full product rule explanation with LaTeX.
Feedback welcome! 🚀