docs: org about + registry + brand icon

README.md

@@ -6,15 +6,13 @@ tags:
 
 # DotCheck
 
-We train detection models, measure them on holdouts we publish, and ship the product that runs them.
+This organization publishes the open side of DotCheck’s production detectors: Apache-2.0 open weights (`.npz` heads on frozen upstream backbones), model cards, and the held-out numbers used in product docs.
 
-This Hub org is the open side of that work: Apache-2.0 open weights (`.npz` heads on frozen upstream backbones), model cards, and the same gate numbers we use in production. Scores are estimates, not verdicts. The tables are what we measured. Use them as you will.
-
-See what's real.
+Each card states encoding, architecture, fit data, and a Layer A / holdout table at threshold 0.5. Scores are estimates of \(P(\mathrm{AI})\) under that model, not authorship verdicts.
 
 | Hub id | Wire | Backbone / stack | Artifact |
 |--------|------|------------------|----------|
-| [`vermeer-image-v14_2`](https://huggingface.co/DotCheck/vermeer-image-v14_2) | `inhouse@14` | Pair stack: max-side 256 + native 224 | `siglip2_base_patch16_224_linear_head_v14_2.npz` |
+| [`vermeer-image-v14_2`](https://huggingface.co/DotCheck/vermeer-image-v14_2) | `inhouse@14` | Still pair stack (transport + center crop) | `siglip2_base_patch16_224_linear_head_v14_2.npz` |
 | [`valla-text-v11`](https://huggingface.co/DotCheck/valla-text-v11) | `inhouse-text@11` (+ lang heads) | TMR + Fakespot + logistic heads | `text_stack_head_v11.npz` + lang |
 | [`muybridge-video-v4_4`](https://huggingface.co/DotCheck/muybridge-video-v4_4) | `inhouse-video@4` | Pair stack + stamp-cap max | `siglip2_base_patch16_224_linear_head_v4_4.npz` |
 | [`helmholtz-audio-v3`](https://huggingface.co/DotCheck/helmholtz-audio-v3) | `inhouse-audio@3` | Dasheng-Base + logistic head | `dasheng_base_linear_head_audio_h3e.npz` |