benchmarks/parsebench/layout.md

# Evaluation Report

**Generated:** 2026-10-22T06:02:51.659641

## Summary

- **Total Examples:** 500
- **Successful:** 500
- **Failed:** 0
- **Skipped:** 0

## Latency Ms Statistics

- **Average:** 1349.9ms
- **Total:** 674945ms
- **Min:** 14.0ms
- **Max:** 23597.0ms
- **P50:** 615.0ms
- **P95:** 5501.0ms
- **P99:** 14153.0ms
- **Count:** 500

## Latency Ms Per Page Statistics

- **Average:** 1349.9ms/page
- **Total:** 674945ms/page
- **Min:** 14.0ms/page
- **Max:** 23597.0ms/page
- **P50:** 615.0ms/page
- **P95:** 5501.0ms/page
- **P99:** 14153.0ms/page
- **Count:** 500

## Aggregate Metrics

| Metric | Value |
|--------|-------|
| Ap50 | 0.3824 |
| Ap75 | 0.1683 |
| Af1 | 0.8520 |
| Content Faithfulness | 0.8317 |
| F1 Page-Footer | 0.3797 |
| F1 Page-Header | 0.3686 |
| F1 Picture | 0.3758 |
| F1 Section | 0.2745 |
| F1 Table | 0.2593 |
| F1 Text | 0.3157 |
| Lap | 0.8445 |
| Lar | 0.8817 |
| Layout Attribution Pass Rate | 0.5095 |
| Layout Classification Pass Rate | 0.3755 |
| Layout Element Rule Pass Rate | 0.2995 |
| Layout Localization Pass Rate | 0.4932 |
| Layout Reading Order Pass Rate | 0.6444 |
| Layout Rule Pass Rate | 0.4567 |
| Map@[.50:.95] | 0.2072 |
| Mean F1 | 0.3271 |
| Normalized Order | 0.8317 |
| Normalized Text Score | 0.8317 |
| Num Ground Truth | 32.6055 |
| Num Predictions | 24.1124 |
| Precision Page-Footer | 0.3868 |
| Precision Page-Header | 0.3933 |
| Precision Picture | 0.4535 |
| Precision Section | 0.3740 |
| Precision Table | 0.2275 |
| Precision Text | 0.3920 |
| Recall Page-Footer | 0.3917 |
| Recall Page-Header | 0.3652 |
| Recall Picture | 0.3659 |
| Recall Section | 0.2990 |
| Recall Table | 0.3723 |
| Recall Text | 0.3212 |
| Rule Order Pass Rate | 0.8317 |
| Rule Pass Rate | 0.5047 |
| Unmatched Gt Elements | 19.0505 |
| Unmatched Pred Elements | 11.3372 |