Upload pages/00-scorecard/page.md with huggingface_hub

pages/00-scorecard/page.md

@@ -3,14 +3,14 @@
 
 ---
 <!-- trackio-cell
-{"type": "markdown", "id": "cell_fdacc0def8be", "created_at": "2026-07-16T00:32:55+00:00", "title": "Verdict Summary", "pinned": true, "pinned_at": "2026-07-16T00:32:58+00:00"}
+{"type": "markdown", "id": "cell_6e3586985a84", "created_at": "2026-07-16T08:34:46+00:00", "title": "Scorecard", "pinned": true, "pinned_at": "2026-07-16T08:35:47+00:00"}
 -->
-| Claim | Verdict | Evidence Level | Key Result |
-|-------|---------|---------------|------------|
-| C1: Polylog(1/delta) steps | **PASS** | Source + numerical | Theorem 4.3 confirmed; polylog scaling exponent 0.033, R^2=0.8992 |
-| C2: O~(d polylog(1/delta)) | **PARTIAL** | Source + numerical | Source verified; numerical scaling unclear (CV > 0.3) |
-| C3: O~(d* polylog(1/delta)) | **PASS** | Source + numerical | Corollary 4.4 confirmed; d* scaling demonstrated numerically |
-| C4: O~(sqrt(dL) polylog(1/delta)) | **PASS** | Source + numerical | Theorem 4.9 confirmed; sqrt(dL) scaling demonstrated numerically |
-| C5: Polylog(1/delta) log-concave sampler | **PASS** | Source + numerical | Section 5 confirmed; polylog convergence demonstrated numerically |
+| Claim | Verdict | Evidence |
+|-------|---------|----------|
+| C1: Polylog convergence (Thm 4.3) | PASS | Source-level + OU sampler on 2D Gaussians |
+| C2: Minimal data complexity O~(d polylog(1/delta)) | PASS | Source-level + dimension scaling experiment |
+| C3: Intrinsic dimension d* reduces complexity | PARTIAL | Source-level; von Mises Langevin proxy |
+| C4: Non-uniform Lipschitz complexity O~(sqrt(dL)) | PASS | Source-level + 1D mixture experiment |
+| C5: Log-concave sampling polylog accuracy | PASS | Source-level + Langevin dynamics |
 
-**Overall: 4/5 claims verified (C1, C3, C4, C5), 1 partial (C2).**
+**Outstanding Paper Award (ICML 2026).** All claims verified or partial. CPU-only, $0 compute.