CACTI Mismatch-Only Benchmark Suite (P-benchmark + 4 I-benchmark tiers)L3-004
Example Data
View demos →A reference solver demo is available — see input / ground truth / reconstruction visualizations on the demos page.
Genesis spec
- Parent L2:
- L2-004
- Spec type:
- —
- d_spec:
- —
On-chain
Not registered on-chain yet.
P-benchmark — rho=50 distribution over Ω
The headline ranking score. Solvers are evaluated on 50 instances drawn parametrically across the full Ω, scored as a distribution. This is what the leaderboard's rank-1 chases.
| Authored baseline | overall_Q | ≈ Q_int |
|---|---|---|
| GAP-TV | 0.45 | 45 |
| PnP-FFDNet | 0.61 | 61 |
| EfficientSCI | 0.67 | 67 |
To take rank 1 of the P-benchmark, beat the highest authored Q above. See L2 spec for the epsilon_fn used per draw.
I-benchmark tiers — 4 fixed-Ω operating points
Each tier is a single-instance evaluation at a specific Ω with its own ε floor. Solvers may pass individual tiers without saturating the full P-benchmark.
| Tier | ρ | ε (dB) | Ω parameters | d_ibench | Top authored baseline |
|---|---|---|---|---|---|
| T1_nominal | 1 | 26 | H=256, W=256, B=8 … | 0.14 | EfficientSCI = 34.5 (PSNR_dB)Q=0.92 |
| T2_low | 3 | 24 | H=256, W=256, B=8 … | 0.28 | EfficientSCI = 29.4 (PSNR_dB)Q=0.75 |
| T3_moderate | 5 | 22.5 | H=512, W=512, B=16 … | 0.48 | EfficientSCI = 25.8 (PSNR_dB)Q=0.62 |
| T4_blind | 10 | 21 | H=1024, W=1024, B=32 … | 0.71 | EfficientSCI = 20.1 (PSNR_dB)Q=0.34 |
Current standing
Example data
Two independent samples from the InverseNet benchmark, each shown as measurement → ground truth → reference reconstruction. The reconstruction is the reference solver's output — the bar a competing solver must beat to win the bounty.
sample_01 · kobe
reference PSNR 27.99 dB
sample_02 · traffic
reference PSNR 20.88 dB
Get this benchmark
Everything you need to run the reference solver locally, browse the full benchmark specification, or submit a competing solution.
Download sample data
sample_01 · 4143245 B
- snapshot.npz354.6 KB
- meta.json1.3 KB
- solution.npz1.79 MB
- ground_truth.npz1.72 MB
- ground_truth.png28.6 KB
- solution.png30.6 KB
- snapshot.png39.4 KB
sample_02 · 2792190 B
- snapshot.npz256.2 KB
- meta.json1.3 KB
- solution.npz1.75 MB
- ground_truth.npz548.3 KB
- ground_truth.png38.1 KB
- solution.png38.3 KB
- snapshot.png47.5 KB
sample_03 · 2613734 B
- snapshot.npz250.7 KB
- meta.json1.3 KB
- solution.npz1.73 MB
- ground_truth.npz429.7 KB
- ground_truth.png30.7 KB
- solution.png31.6 KB
- snapshot.png38.0 KB
sample_04 · 2337413 B
- snapshot.npz254.4 KB
- meta.json1.3 KB
- solution.npz1.66 MB
- ground_truth.npz254.7 KB
- ground_truth.png17.1 KB
- solution.png23.4 KB
- snapshot.png36.9 KB
sample_05 · 3105035 B
- snapshot.npz311.2 KB
- meta.json1.3 KB
- solution.npz1.76 MB
- ground_truth.npz816.6 KB
- ground_truth.png29.0 KB
- solution.png31.4 KB
- snapshot.png41.6 KB
sample_06 · 3228468 B
- snapshot.npz332.7 KB
- meta.json1.3 KB
- solution.npz1.75 MB
- ground_truth.npz917.8 KB
- ground_truth.png31.4 KB
- solution.png31.3 KB
- snapshot.png42.4 KB
Browse on GitHub
- Benchmark card (YAML spec)
- L3 principle JSON (canonical on-chain spec)
- Demo README (full pipeline walkthrough)
- Bounty index (compete with a better solver)
Run the reference solver locally
git clone https://github.com/integritynoble/pwm-public.git cd pwm-public python3 scripts/regenerate_demos_inversenet.py --only cacti (uses pwm_core.recon.cacti_solvers.gap_tv_cacti) cat /tmp/out/meta.json
Expected reference PSNR: 27.99 dB (seed ). Byte-stable across runs at the same git SHA.