ops: DotCheck Apache-2.0 open heads + card

README.md

@@ -27,15 +27,15 @@ model-index:
         metrics:
           - name: mean_P_AI_real
             type: mean_score_real
-            value: 0.005
+            value: 0.018
           - name: mean_P_AI_ai
             type: mean_score_ai
-            value: 0.982
+            value: 0.989
           - name: balanced_accuracy
             type: balanced_accuracy
-            value: 0.989
+            value: 0.995
         source:
-          name: audio_gates_audio_v3 / Data.json
+          name: audio_gates_audio_h3e / Data.json
           url: https://dotcheck.ai/docs
 ---
 
@@ -50,7 +50,7 @@ Open package: live `.npz` head, card, license, notices. In product, soundtrack w
 | Hub id | `DotCheck/helmholtz-audio-v3` |
 | Wire id | `inhouse-audio@3` |
 | Label | Helmholtz |
-| Artifact | `dasheng_base_linear_head_audio_v3.npz` |
+| Artifact | `dasheng_base_linear_head_audio_h3e.npz` |
 | Backbone | [`mispeech/dasheng-base`](https://huggingface.co/mispeech/dasheng-base) (Apache-2.0) |
 | Protocol | 4 s / center crop / up to 2 windows / max aggregate (serve may single-pass prewindowed clips) |
 | Output | `p ∈ [0,1]` — estimated P(AI) |
@@ -93,12 +93,13 @@ Gates: [dotcheck.ai/docs](https://dotcheck.ai/docs).
 | Split | Content |
 |-------|---------|
 | Fit reals | LibriTTS + FSD50K (concat, no loop-tile) |
-| Fit AI | parler-mini / melo-en / dasheng-audiogen + fal train (Kokoro / Eleven Turbo / mixes) + 22 Imagine few-shot |
+| Fit AI | parler-mini / melo-en / dasheng-audiogen + fal train (Kokoro / Eleven Turbo / mixes) + Imagine in the linear mix |
 | Holdout speech | CodecFake / DFADD (sha-disjoint) |
 | Holdout music/SFX | AudioGen odd-seed partition |
 | Holdout fal | prompt-id holdout (same gens, disjoint sha) |
+| Holdout Imagine | owned exam (Layer A); not a separate public claim row |
 
-Evidence: `audio_gates_audio_v3.json` · `AUDIO_GATES_OK`. Imagine-owned audio is a **watch**, not a public claim.
+Evidence: `audio_gates_audio_h3e.json` · `AUDIO_GATES_OK`. Public floats are the overall holdout (`claims.audio`).
 
 ## Evaluation
 
@@ -106,9 +107,9 @@ Evidence: `audio_gates_audio_v3.json` · `AUDIO_GATES_OK`. Imagine-owned audio i
 
 | Metric | Target | Measured |
 |--------|--------|---------:|
-| mean P(AI) \| real | ≤ 0.12 | **0.005** |
-| mean P(AI) \| AI | ≥ 0.85 | **0.982** |
-| bal_acc @ thr | ≥ 0.90 | **0.989** |
+| mean P(AI) \| real | ≤ 0.12 | **0.018** |
+| mean P(AI) \| AI | ≥ 0.85 | **0.989** |
+| bal_acc @ thr | ≥ 0.90 | **0.995** |
 
 ## Intended use
 

dasheng_base_linear_head_audio_h3e.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:caf141a5d82c589d93bdaec5932870e40ca343a39ccbe6d6d6e273e4a81348e3
+size 7108