Document the multimodal vision config

README.md

@@ -370,7 +370,7 @@ The [exclusion evidence resource](quality/filtered-full-v1/README.md) records hi
 
 ## Vision config
 
-Load `load_dataset("tasksource/tasksource-jev-typed-decisions", "vision")` for the multimodal pilot (12,997 train, 185 test, 380 validation).
+Load `load_dataset("tasksource/tasksource-jev-typed-decisions", "vision")` for the multimodal pilot (60,097 train, 175 test, 375 validation).
 It covers ai2d, aokvqa, clevr/color, clevr/count, clevr/material, clevr/shape, clevr/size, clevr/yesno, figureqa, intergps, mapqa/yesno, nlvr2, rico-widget/element, rico-widget/grid7, scienceqa-img, tallyqa/count, tqa, visual7w, vsr/yesno, with genuine source labels.
 `state`, `question`, `kind`, `options`, and `target` retain their existing decision semantics.
 `images` is an ordered sequence of Hugging Face Image features. Source bytes are preserved
@@ -385,7 +385,7 @@ See [vision/sources.yaml](vision/sources.yaml), [vision/release-audit.json](visi
 [vision/build-manifest.json](vision/build-manifest.json), and [vision/quality-audit.json](vision/quality-audit.json)
 when the materialized release was audited.
 
-Sampling: uniform reservoir over complete eligible source splits; deterministic shared family caps, round-robin task views, whole decision groups. Sampling caps and checkpoint provenance are recorded in the manifests.
+Sampling: full-source uniform reservoirs, shared per-family export caps, with per-task checkpoint provenance. Sampling caps and checkpoint provenance are recorded in the manifests.
 Visual task views share one export budget per dataset family, including CLEVR, TallyQA, and RICO;
 individual task IDs remain in `source`, and family totals appear in `release-audit.json`.
 Only native labeled splits are used. Source action identities group related GUI decisions;