Steven Chong 056f970881 test(eval): measure Grok image recall and the factsheet contract ...

Grok was imaging on the Responses path with no readability data. Add
eval/grok-density: density ladder, climb receipts, and a live
image-vs-factsheet comparison on grok-4.5.

Findings locked into the receipts:
- Production 5x8 image-only confabulates exact IDs (0/4).
- Image + factsheet clears 4/4 exact, 0 confab, gist+guard ok (~70%
  fixture savings).
- Looser pure-image packing can clear exact recall but only ~30%
  savings — not worth the default.

Supports keeping Grok opt-in with dense 5x8 + factsheet when enabled.

2026-07-09 12:44:14 -04:00

..

atlas-gray-jbmono8.ts

test(eval): measure Grok image recall and the factsheet contract

2026-07-09 12:44:14 -04:00

atlas-gray-jbmono10.ts

test(eval): measure Grok image recall and the factsheet contract

2026-07-09 12:44:14 -04:00

atlas-gray-jbmono12.ts

test(eval): measure Grok image recall and the factsheet contract

2026-07-09 12:44:14 -04:00

atlas-gray-spleen5x8.ts

test(eval): measure Grok image recall and the factsheet contract

2026-07-09 12:44:14 -04:00

atlas-gray-unifont8.ts

test(eval): measure Grok image recall and the factsheet contract

2026-07-09 12:44:14 -04:00

atlas-gray-unifont10.ts

test(eval): measure Grok image recall and the factsheet contract

2026-07-09 12:44:14 -04:00

atlas-jbmono8.ts

test(eval): measure Grok image recall and the factsheet contract

2026-07-09 12:44:14 -04:00

atlas-jbmono10.ts

test(eval): measure Grok image recall and the factsheet contract

2026-07-09 12:44:14 -04:00

atlas-jbmono12.ts

test(eval): measure Grok image recall and the factsheet contract

2026-07-09 12:44:14 -04:00

atlas-spleen5x8.ts

test(eval): measure Grok image recall and the factsheet contract

2026-07-09 12:44:14 -04:00

atlas-unifont8.ts

test(eval): measure Grok image recall and the factsheet contract

2026-07-09 12:44:14 -04:00

atlas-unifont10.ts

test(eval): measure Grok image recall and the factsheet contract

2026-07-09 12:44:14 -04:00