CLAUDE.md

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

Project Overview

This is a Test Results Dashboard project (Tcid) that provides interactive visualization of AI model testing results. The project consists of two main applications:

1. Gradio Dashboard (app.py) - Python-based web dashboard using Gradio and Matplotlib
2. HTML Dashboard (index.html) - Standalone HTML dashboard with Chart.js visualization

Both dashboards display test results for AI models including metrics like passed, failed, skipped, and error counts.

Architecture

Core Components

• app.py: Main Gradio application with dark theme UI, sidebar navigation, and matplotlib pie charts
• model_stats.json: JSON data file containing test results for different AI models
• index.html: Self-contained HTML dashboard with device-specific performance comparison (NVIDIA vs AMD)
• requirements.txt: Python dependencies (currently only matplotlib>=3.8)

Data Structure

Model statistics follow this format:

{
    "model_name": {
        "passed": int,
        "failed": int, 
        "skipped": int,
        "error": int
    }
}

The HTML dashboard extends this with device-specific data for NVIDIA and AMD performance comparisons.

Development Commands

Environment Setup

# Activate virtual environment
source venv_tci/bin/activate

# Install dependencies
pip install -r requirements.txt

Running the Applications

Gradio Dashboard:

python app.py

HTML Dashboard: Open index.html directly in a web browser - no server required.

Python Environment

• Python 3.12.4
• Virtual environment located at venv_tci/
• Dependencies managed via requirements.txt

Key Implementation Details

Gradio Application (app.py)

• Uses MODELS dictionary for hardcoded test data (lines 8-12)
• plot_model_stats() function generates matplotlib pie charts with dark theme
• Custom CSS for dark theme styling (lines 77-133)
• Sidebar navigation with model selection buttons
• Real-time chart updates on model selection

Data Management

• Model data is currently hardcoded in app.py
• External JSON data file model_stats.json exists but is not integrated
• HTML dashboard has embedded JavaScript data

Styling

• Dark theme with black backgrounds (#000000)
• Custom color scheme: Green (passed), Red (failed), Orange (skipped), Purple (error)
• Responsive design with sidebar layout

Hugging Face Spaces Configuration

This project is configured as a Hugging Face Space:

• SDK: Gradio 5.38.0
• App file: app.py
• Space emoji: 👁
• Color theme: indigo to pink gradient