math-llm-demo / README.md
Joash2024's picture
fix: use simple pipeline with our fine-tuned model and A100
c5de890
|
raw
history blame
1 kB
metadata
title: Mathematics Problem Solver
emoji: 🧮
colorFrom: blue
colorTo: green
sdk: gradio
sdk_version: 4.8.0
app_file: app.py
pinned: false
hardware:
  accelerator: a100
  gpu: true
python_packages:
  - torch>=2.0.0
  - transformers>=4.30.0
  - accelerate>=0.20.0
  - peft==0.5.0
  - numpy>=1.21.0

Mathematics Problem Solver

This Space demonstrates our fine-tuned math model for solving various mathematical problems, with a focus on derivatives. Compare solutions between:

  1. Base Model: HuggingFaceTB/SmolLM2-1.7B-Instruct
  2. Our Fine-tuned Model: Joash2024/Math-SmolLM2-1.7B

Features

  • Side-by-side comparison of base and fine-tuned models
  • Performance monitoring:
    • Response times
    • Success rates
    • Problem type distribution
  • Support for various problems:
    • Derivatives
    • Addition
    • Roots
    • Custom problems

Technical Details

  • A100 GPU acceleration
  • Float16 precision for efficient inference
  • LaTeX notation support
  • Real-time performance tracking