← Back to Token Labs

🔍 Baseline Comparison

Model Accuracy vs. Baseline

Auto-updated via CI/CD
Loading comparison results...

📐 About Baseline Comparison

What is this?
This page shows how the current model's accuracy compares to a baseline reference model. The comparison helps ensure that quantized or optimized models maintain acceptable accuracy levels.

Baseline Model: Loading...
This is the reference implementation that establishes expected accuracy levels.

Comparison Thresholds

  • ✅ PASS: Model accuracy is within ±5% of baseline (acceptable quality)
  • 🎉 IMPROVED: Model accuracy exceeds baseline by more than 5%
  • ❌ FAIL: Model accuracy is more than 5% below baseline (significant degradation)
📄 View Raw Comparison Data (JSON) 🎯 View Accuracy Results 📊 View Performance Benchmarks