[allenporter]

Summary

• Check in baseline OCR benchmark results and per-page JSON transcripts under evals/ocr/results/ (apple.json for apple:vision-accurate / apple:vision-fast, and device_rtr.json for embedded Supernote RTR) alongside a combined comparison report in evals/ocr/RESULTS.md.
• Add merge_eval_reports() in supernote/server/ai/eval.py and --from-json in script/eval_ocr.py so new OCR providers (gemini, openai/Ollama/llama.cpp, etc.) can be evaluated incrementally on separate machines/environments and merged into evals/ocr/RESULTS.md without re-running existing providers.
• Include Median CER in format_markdown_report() and validate in tests/server/ai/test_eval.py that checked-in evals/ocr/results/*.json files parse cleanly and remain in sync with evals/ocr/RESULTS.md.

Baseline Summary (supernote-ocr-suite-v1, 20 pages)

Target	Model	Pages	Mean CER	Median CER	Content CER	Word Recall	BuJo Symbol Recall	Median Latency	P95 Latency	Errors
apple:vision-accurate	vision-accurate	20	15.9%	4.2%	15.6%	77.0%	67.7%	365.5 ms	499.2 ms	0
apple:vision-fast	vision-fast	20	65.3%	81.7%	65.0%	24.1%	12.2%	113.2 ms	160.0 ms	0
device-rtr	supernote-rtr-embedded	3	4.3%	1.9%	4.3%	98.8%	—	0.0 ms	0.0 ms	0

Test Plan

• ./script/lint
• pytest tests/test_init.py tests/server/ai/test_eval.py (100% statement coverage on supernote/server/ai/eval.py)

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 94.57%. Comparing base (1061932) to head (7406e09).
⚠️ Report is 2 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #550      +/-   ##
==========================================
- Coverage   94.59%   94.57%   -0.02%     
==========================================
  Files         154      155       +1     
  Lines       18531    18667     +136     
==========================================
+ Hits        17529    17655     +126     
- Misses       1002     1012      +10     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.