README: surface the 8-way public model comparison (incl. FST) in the main table

- Add the 8-way summary table (F1/Precision/Recall/FPR/sanity-FAIL) under Baseline
Results so FST and the other five detectors are visible without opening v1.1/.
- RESULTS_8WAY.md: state the scored track count (2,104: 1,388 AI / 716 real) instead
of the partition size, add FPR and FAIL columns from the stored per-track
probabilities, and note the production-pipeline (v9.7/cnn_v95) measurement.

README.md

@@ -48,6 +48,27 @@ Per-source pass/fail thresholds:
 | CLAM (MoM) | 194M | 0.824 | 16/28 | 78% | 70.5% |
 | SpecTTTra | 19M | 0.766 | 23/28 | 55% | 21.4% |
 
+## Public Model Comparison — 8-way (v1.1 test partition, 2026-07-04)
+
+Eight publicly-available detectors scored on identical files with the same runner
+(adapters in `artifactbench/models/`), τ = 0.5. n = 2,104 (1,388 AI / 716 real —
+the locally-restored real set; see the provenance caveat in `v1.1/RESULTS_8WAY.md`).
+
+| Rank | Model | Params | F1 | Precision | Recall (TPR) | FPR | Sanity FAIL |
+|---|---|---:|:---:|:---:|:---:|:---:|:---:|
+| 1 | **ArtifactNet v9.4 (public ONNX)** | **4.2M** | **0.952** | 0.932 | 97.3% | 13.8% | 8/28 |
+| 2 | AI-Music-Detection AST-60s | 90.8M | 0.840 | 0.848 | 83.1% | 28.9% | 16/28 |
+| 3 | CLAM (MoM) | 194.3M | 0.787 | 0.711 | 88.3% | 69.7% | 14/28 |
+| 4 | SpecTTTra α-120s | 18.7M | 0.777 | 0.880 | 69.5% | 18.4% | 22/28 |
+| 5 | Deezer ISMIR fakeprint LR | 3.6K | 0.754 | 0.906 | 64.6% | 13.0% | 18/28 |
+| 6 | FST (Mippia, arXiv:2601.13647) | 174.4M | 0.735 | 0.984 | 58.7% | 1.8% | 17/28 |
+| 7 | DeepFense EAT+Nes2Net | — | 0.650 | 0.589 | 72.4% | 97.8% | 11/28 |
+| 8 | SpecTTTra β-5s | 18.7M | 0.563 | 0.884 | 41.3% | 10.5% | 24/28 |
+
+The ArtifactNet production pipeline (v9.7 / cnn_v95, PyTorch — not the public ONNX
+export above) measured on the same partition: F1 0.984 / TPR 98.9% / FPR 4.2%
+(1/28 FAIL). Per-source TPR/FPR tables, ROC summaries, and notes: `v1.1/RESULTS_8WAY.md`.
+
 ## Usage
 
 ```python
@@ -133,6 +154,6 @@ See `v1.1/RESULTS_v1.1.md`.
 
 ## 8-way public model comparison (2026-07-04)
 
-Extends the v1.1 evaluation with five more publicly-available detectors on the
-purged test partition. ArtifactNet ranks first among all eight (parameter count
-does not predict F1 — see notes). Full table: `v1.1/RESULTS_8WAY.md`.
+Extends the v1.1 evaluation with five more publicly-available detectors. The summary
+table is in **Public Model Comparison** above; per-source breakdowns and notes are in
+`v1.1/RESULTS_8WAY.md`.

v1.1/RESULTS_8WAY.md

@@ -18,18 +18,25 @@ Adapters: `artifactbench/models/{ast_60s,deepfense,deezer_ismir,fst,spectttra_va
 Run with `--model ast_60s --model deepfense --model deezer_ismir --model fst --model spectttra_beta5s`
 alongside the existing `--model artifactnet --model spectttra --model clam`.
 
-## Results (n = 2,224, τ = 0.5)
+## Results (n = 2,104: 1,388 AI / 716 real, τ = 0.5)
 
-| Rank | Model | Params | F1 | Precision | Recall (TPR) |
-|---|---|---:|:---:|:---:|:---:|
-| 1 | **ArtifactNet v9.4 (ours)** | **4.2M** | **0.952** | 0.932 | 97.3% |
-| 2 | AI-Music-Detection AST-60s | 90.8M | 0.840 | 0.848 | 83.1% |
-| 3 | CLAM (MoM) | 194.3M | 0.787 | 0.711 | 88.3% |
-| 4 | SpecTTTra α-120s | 18.7M | 0.777 | 0.880 | 69.5% |
-| 5 | Deezer ISMIR fakeprint LR | 3.6K | 0.754 | 0.906 | 64.6% |
-| 6 | FST (Mippia) | 174.4M | 0.735 | 0.984 | 58.7% |
-| 7 | DeepFense EAT+Nes2Net | — | 0.650 | 0.589 | 72.4% |
-| 8 | SpecTTTra β-5s | 18.7M | 0.563 | 0.884 | 41.3% |
+| Rank | Model | Params | F1 | Precision | Recall (TPR) | FPR | Sanity FAIL |
+|---|---|---:|:---:|:---:|:---:|:---:|:---:|
+| 1 | **ArtifactNet v9.4 (ours, public ONNX)** | **4.2M** | **0.952** | 0.932 | 97.3% | 13.8% | 8/28 |
+| 2 | AI-Music-Detection AST-60s | 90.8M | 0.840 | 0.848 | 83.1% | 28.9% | 16/28 |
+| 3 | CLAM (MoM) | 194.3M | 0.787 | 0.711 | 88.3% | 69.7% | 14/28 |
+| 4 | SpecTTTra α-120s | 18.7M | 0.777 | 0.880 | 69.5% | 18.4% | 22/28 |
+| 5 | Deezer ISMIR fakeprint LR | 3.6K | 0.754 | 0.906 | 64.6% | 13.0% | 18/28 |
+| 6 | FST (Mippia) | 174.4M | 0.735 | 0.984 | 58.7% | 1.8% | 17/28 |
+| 7 | DeepFense EAT+Nes2Net | — | 0.650 | 0.589 | 72.4% | 97.8% | 11/28 |
+| 8 | SpecTTTra β-5s | 18.7M | 0.563 | 0.884 | 41.3% | 10.5% | 24/28 |
+
+*Update 2026-09-05.* The track count above is the number actually scored in this run
+(1,388 AI / 716 real); the earlier "n = 2,224" referred to the v1.1 partition size, not
+to the files available on the scoring host (see the provenance caveat below). FPR and
+sanity-FAIL columns were added from the stored per-track probabilities of the same run.
+The ArtifactNet production pipeline (v9.7 / cnn_v95, PyTorch, not the public ONNX
+export) on the same files: F1 0.984 / TPR 98.9% / FPR 4.2%, 1/28 FAIL (fma_hardneg).
 
 **Observation.** Parameter count does not predict performance: FST (174.4M) scores
 below Deezer's 3.6K-parameter logistic-regression baseline. ArtifactNet (4.2M) ranks