diff --git "a/eval/reports.html" "b/eval/reports.html" --- "a/eval/reports.html" +++ "b/eval/reports.html" @@ -81,12 +81,12 @@ code { background:#eef1f5; padding:1px 5px; border-radius:4px; font-size:12px; }

Model comparison

Overall score is the unweighted mean of Prompt Fit and the six rubric dimensions, following DesignEvaluator V5.9.2.

-

27B AdamW step400

best
27B · AdamW · 400
87.89
median88.93
p2582.86
min46.43
prompt fit84.10
rubric pass88.60%
cases < 602
render fail0
n200

9B AdamW step200

9B · AdamW · 200
84.40
median85.59
p2579.29
min38.57
prompt fit78.35
rubric pass85.37%
cases < 606
render fail1
n200

4B AdamW step200

4B · AdamW · 200
82.12
median84.41
p2574.05
min38.33
prompt fit75.25
rubric pass83.24%
cases < 6014
render fail1
n200

4B Muon step200

4B · Muon · 200
77.62
median80.83
p2570.75
min35.08
prompt fit64.60
rubric pass79.43%
cases < 6022
render fail3
n200
+

27B AdamW step400

best
27B · AdamW · 400
87.04
median89.20
p2582.22
min13.09
prompt fit84.10
rubric pass88.60%
cases < 604
render fail0
n200

9B AdamW step200

9B · AdamW · 200
82.14
median84.69
p2574.90
min45.35
prompt fit78.35
rubric pass85.37%
cases < 6011
render fail1
n200

4B AdamW step200

4B · AdamW · 200
80.06
median82.82
p2571.28
min38.59
prompt fit75.25
rubric pass83.24%
cases < 6023
render fail1
n200

4B Muon step200

4B · Muon · 200
71.34
median73.77
p2560.51
min29.13
prompt fit64.60
rubric pass79.43%
cases < 6048
render fail3
n200

Overall score

- 27B AdamW step40087.899B AdamW step20084.404B AdamW step20082.124B Muon step20077.62255075100 + 27B AdamW step40087.049B AdamW step20082.144B AdamW step20080.064B Muon step20071.34255075100

Checkpoints under test

modelbaseoptimizerglobal batchstepHF subfolder
27B AdamW step400Qwen3.8-27BAdamW128400designcoder_qwen3.8_27b_adamw_bs128_data41287_step400
9B AdamW step200Qwen3.5-9BAdamW256200designcoder_qwen3.5_9b_adamw_bs256_data41287_step200
4B AdamW step200Qwen3.5-4BAdamW256200designcoder_qwen3.5_4b_adamw_bs256_data41287_step200
4B Muon step200Qwen3.5-4BMuon256200designcoder_qwen3.5_4b_muon_bs256_data41287_step200
@@ -94,7 +94,7 @@ code { background:#eef1f5; padding:1px 5px; border-radius:4px; font-size:12px; }

Statistics

- +
modelnmeanmedianp25minmaxstdevprompt fitrubric %<60render fail
27B AdamW step40020087.8988.9382.8646.43100.009.5284.1088.6020
9B AdamW step20020084.4085.5979.2938.57100.0010.4378.3585.3761
4B AdamW step20020082.1284.4174.0538.33100.0012.4675.2583.24141
4B Muon step20020077.6280.8370.7535.08100.0013.7364.6079.43223
modelnmeanmedianp25minmaxstdevprompt fitrubric %<60render fail
27B AdamW step40020087.0489.2082.2213.09100.0010.9284.1088.6040
9B AdamW step20020082.1484.6974.9045.35100.0011.9978.3585.37111
4B AdamW step20020080.0682.8271.2838.59100.0013.4775.2583.24231
4B Muon step20020071.3473.7760.5129.1397.1815.3464.6079.43483

p25 = 25th percentile; a high mean with a low p25 means the model is inconsistent rather than uniformly good.

@@ -104,9 +104,9 @@ code { background:#eef1f5; padding:1px 5px; border-radius:4px; font-size:12px; }

Track x Surface

The benchmark requires Track A/B and landing/dashboard to be reported separately.

- 02040608010027B AdamW step400 · A/landing: 87.8487.89B AdamW step200 · A/landing: 85.9285.94B AdamW step200 · A/landing: 84.7084.74B Muon step200 · A/landing: 81.0981.1A/landing27B AdamW step400 · A/dashboard: 84.5684.69B AdamW step200 · A/dashboard: 79.5679.64B AdamW step200 · A/dashboard: 76.6176.64B Muon step200 · A/dashboard: 69.0169.0A/dashboard27B AdamW step400 · B/landing: 90.9290.99B AdamW step200 · B/landing: 87.6187.64B AdamW step200 · B/landing: 86.3086.34B Muon step200 · B/landing: 83.5083.5B/landing27B AdamW step400 · B/dashboard: 89.4889.59B AdamW step200 · B/dashboard: 82.5782.64B AdamW step200 · B/dashboard: 76.6976.74B Muon step200 · B/dashboard: 71.6571.7B/dashboard
27B AdamW step4009B AdamW step2004B AdamW step2004B Muon step200
+ 02040608010027B AdamW step400 · A/landing: 89.1389.19B AdamW step200 · A/landing: 86.9586.94B AdamW step200 · A/landing: 85.2785.34B Muon step200 · A/landing: 78.0778.1A/landing27B AdamW step400 · A/dashboard: 80.1480.19B AdamW step200 · A/dashboard: 72.1272.14B AdamW step200 · A/dashboard: 68.9669.04B Muon step200 · A/dashboard: 56.9957.0A/dashboard27B AdamW step400 · B/landing: 91.3891.49B AdamW step200 · B/landing: 87.0287.04B AdamW step200 · B/landing: 87.1487.14B Muon step200 · B/landing: 79.8679.9B/landing27B AdamW step400 · B/dashboard: 84.9484.99B AdamW step200 · B/dashboard: 74.6374.64B AdamW step200 · B/dashboard: 70.4370.44B Muon step200 · B/dashboard: 59.5059.5B/dashboard
27B AdamW step4009B AdamW step2004B AdamW step2004B Muon step200

By surface

- 02040608010027B AdamW step400 · landing: 88.5588.69B AdamW step200 · landing: 86.3186.34B AdamW step200 · landing: 85.0785.14B Muon step200 · landing: 81.6581.6landing27B AdamW step400 · dashboard: 86.6786.79B AdamW step200 · dashboard: 80.8580.84B AdamW step200 · dashboard: 76.6576.64B Muon step200 · dashboard: 70.1470.1dashboard
27B AdamW step4009B AdamW step2004B AdamW step2004B Muon step200
+ 02040608010027B AdamW step400 · landing: 89.6589.69B AdamW step200 · landing: 86.9687.04B AdamW step200 · landing: 85.7085.74B Muon step200 · landing: 78.4878.5landing27B AdamW step400 · dashboard: 82.1982.29B AdamW step200 · dashboard: 73.1973.24B AdamW step200 · dashboard: 69.5969.64B Muon step200 · dashboard: 58.0658.1dashboard
27B AdamW step4009B AdamW step2004B AdamW step2004B Muon step200
@@ -124,20 +124,20 @@ code { background:#eef1f5; padding:1px 5px; border-radius:4px; font-size:12px; }

Score distribution

Share of cases (%) falling in each score band.

- 02040608010027B AdamW step400 · <50: 0.509B AdamW step200 · <50: 1.004B AdamW step200 · <50: 1.004B Muon step200 · <50: 6.50<5027B AdamW step400 · 50-60: 0.509B AdamW step200 · 50-60: 2.004B AdamW step200 · 50-60: 6.004B Muon step200 · 50-60: 4.5050-6027B AdamW step400 · 60-70: 4.509B AdamW step200 · 60-70: 5.004B AdamW step200 · 60-70: 10.504B Muon step200 · 60-70: 11.0060-7027B AdamW step400 · 70-80: 12.009B AdamW step200 · 70-80: 19.004B AdamW step200 · 70-80: 19.004B Muon step200 · 70-80: 26.0070-8027B AdamW step400 · 80-90: 35.509B AdamW step200 · 80-90: 37.504B AdamW step200 · 80-90: 30.504B Muon step200 · 80-90: 34.0080-9027B AdamW step400 · 90-95: 19.509B AdamW step200 · 90-95: 22.004B AdamW step200 · 90-95: 18.504B Muon step200 · 90-95: 11.5090-9527B AdamW step400 · 95-100: 27.509B AdamW step200 · 95-100: 13.504B AdamW step200 · 95-100: 14.504B Muon step200 · 95-100: 6.5095-100
27B AdamW step4009B AdamW step2004B AdamW step2004B Muon step200
+ 02040608010027B AdamW step400 · <50: 1.009B AdamW step200 · <50: 1.004B AdamW step200 · <50: 1.504B Muon step200 · <50: 9.50<5027B AdamW step400 · 50-60: 1.009B AdamW step200 · 50-60: 4.504B AdamW step200 · 50-60: 10.004B Muon step200 · 50-60: 14.5050-6027B AdamW step400 · 60-70: 4.009B AdamW step200 · 60-70: 11.004B AdamW step200 · 60-70: 12.004B Muon step200 · 60-70: 19.0060-7027B AdamW step400 · 70-80: 14.009B AdamW step200 · 70-80: 22.504B AdamW step200 · 70-80: 18.004B Muon step200 · 70-80: 22.0070-8027B AdamW step400 · 80-90: 33.009B AdamW step200 · 80-90: 30.004B AdamW step200 · 80-90: 31.504B Muon step200 · 80-90: 25.5080-9027B AdamW step400 · 90-95: 25.509B AdamW step200 · 90-95: 17.504B AdamW step200 · 90-95: 14.004B Muon step200 · 90-95: 7.0090-9527B AdamW step400 · 95-100: 21.509B AdamW step200 · 95-100: 13.504B AdamW step200 · 95-100: 13.004B Muon step200 · 95-100: 2.5095-100
27B AdamW step4009B AdamW step2004B AdamW step2004B Muon step200

Weakest cases — 27B AdamW step400

- +
scoregroupcase_idjudge summary
46.43A/dashboarddashboard-045-streaming-dashboard-obsA polished dark console shell is visible, but most core storm-tracking visualizations and table content are blank.
50.95A/landinglanding-027-art-decoA polished Art Deco landing page with strong atmosphere and booking emphasis is materially weakened by the absent timetable, non-symmetrical hero imagery, and incomplete offering grids.
61.55A/dashboarddashboard-002-armin-hofmann-contrast-and-reductionThe dashboard has a strong monochrome structural shell, but its empty charts, ticket table, station rail, and inspector make the operational experience incomplete.
66.67A/dashboarddashboard-005-dieter-rams-silent-control-panelA polished and highly legible silent-control dashboard whose default overview is strong, but the screenshot does not demonstrate the selected-lift inspector, capacity consequence, webcam, maintenance, loading, or active
68.10A/landinglanding-015-biomorphic-designThe page is polished and atmospherically cohesive with complete downstream sections and reservation flow, but the missing hero selector, absent table-lamp clarity, and occluded compact-card copy weaken fidelity and usabi
68.10A/landinglanding-041-stop-motion-craftA polished, cohesive pottery-drop landing page communicates the collection and reservation flow well, but several specific handmade and overhead-photography requirements are visibly absent.
68.88B/dashboardtrack-b-dashboard-016-src-393The dashboard strongly prioritizes and explains the funding shortfall in a polished dense layout, but the visible workflow stops before intercompany staging, approval, rationale, and resolved-state closure.
68.97A/dashboarddashboard-020-data-storytelling-uiA polished and well-structured Gateflow shell presents strong KPIs and controls, but its blank chart, heatmap, and flight table prevent complete operational monitoring.
69.39B/landingtrack-b-landing-001-src-268The page has a polished, calm booking-oriented structure, but incomplete room presentation, a broken image and visibly unrelated photography materially weaken the brief fit.
69.76A/landinglanding-110-blackout-xerox-punkA polished and highly coherent Xerox-punk race landing page covers most content and commerce needs, but lacks several required visual asset types and does not visibly demonstrate populated race-card states.
69.93B/dashboardtrack-b-dashboard-001-src-368A polished, decision-ready cold-chain dashboard with strong prioritization and evidence, weakened mainly by the missing corrective-action control and limited trailer/depot detail.
70.48A/dashboarddashboard-009-arcade-cabinet-opsA polished and highly legible arcade race-control dashboard is substantially complete, but its obscured leaderboard and missing visible row-to-inspector anchor are major localized weaknesses.
70.71A/dashboarddashboard-016-financial-broadsheetA polished and decision-ready opening-weekend broadsheet that strongly covers the core workflow, while missing authentic poster thumbnails, correspondent margin notes, restruck estimates, and serif index numerals.
70.71A/dashboarddashboard-029-lime-noir-logistics-glassA polished and decision-ready lime-noir turnover dashboard is shown, but explicit RoomRelay branding, table scrolling, empty-state coverage, and the selected-room inspector flow are absent.
72.38A/dashboarddashboard-041-scada-hmi-industrialA strong SCADA-style monitoring console with comprehensive visible data and actions, held back by overlapping panels, misplaced charts, and missing acknowledgement identity and TRIP evidence.
13.09A/dashboarddashboard-045-streaming-dashboard-obs
45.65A/dashboarddashboard-002-armin-hofmann-contrast-and-reduction
55.79A/dashboarddashboard-006-carbon-fiber-telemetry
59.81A/dashboarddashboard-009-arcade-cabinet-ops
62.65A/landinglanding-005-april-greiman-digital-collage
65.42A/dashboarddashboard-020-data-storytelling-ui
65.78B/dashboardtrack-b-dashboard-026-src-410
67.25A/dashboarddashboard-003-blueprint-hologram-console
67.63A/landinglanding-083-legal-authority
68.08A/landinglanding-019-paul-rand-modernist-wit
69.30A/landinglanding-100-dada-collage
69.81A/landinglanding-060-dieter-rams-silent-control-panel
71.05A/dashboarddashboard-036-obsidian-aura-portfolio
71.53A/landinglanding-061-pop-art
71.97A/landinglanding-037-constructivism

Per-case scores

-
case_idtracksurface27B AdamW step4009B AdamW step2004B AdamW step2004B Muon step200
dashboard-002-armin-hofmann-contrast-and-reductionAdashboard61.5587.0278.8164.64
dashboard-003-blueprint-hologram-consoleAdashboard82.2780.8490.0072.27
dashboard-004-dashboard-brutalismAdashboard100.0090.9586.8785.44
dashboard-005-dieter-rams-silent-control-panelAdashboard66.6771.4384.2949.52
dashboard-006-carbon-fiber-telemetryAdashboard73.0493.5786.4386.23
dashboard-007-ceramic-bento-metricsAdashboard76.8354.5259.3781.83
dashboard-008-admin-crud-panelAdashboard84.5238.5772.6258.33
dashboard-009-arcade-cabinet-opsAdashboard70.4877.1444.7662.38
dashboard-011-anime-dashboardAdashboard96.4383.8178.2159.64
dashboard-012-glass-cockpit-avionicsAdashboard93.5793.9365.1844.88
dashboard-013-bauhausAdashboard93.3379.0597.6273.33
dashboard-014-developer-first-minimalAdashboard90.9589.5290.9568.01
dashboard-015-ev-instrument-clusterAdashboard93.5778.8163.3360.00
dashboard-016-financial-broadsheetAdashboard70.7174.6469.6471.67
dashboard-018-crypto-dashboardAdashboard90.2461.9038.3386.43
dashboard-019-executive-kpi-dashboardAdashboard97.3883.8184.7686.19
dashboard-020-data-storytelling-uiAdashboard68.9760.4082.5452.06
dashboard-021-fluent-designAdashboard78.1076.6765.6844.52
dashboard-023-adaptive-ambient-uiAdashboard87.0283.4578.2183.93
dashboard-024-honeycomb-paper-ledgerAdashboard86.1983.8188.5778.10
dashboard-027-kenya-hara-white-and-emptinessAdashboard91.0297.1492.2478.37
dashboard-029-lime-noir-logistics-glassAdashboard70.7190.9566.9072.62
dashboard-030-lucky-swaps-volumetric-consoleAdashboard82.8677.8679.0584.76
dashboard-031-mac-os-aqua-nostalgiaAdashboard98.5774.6978.9865.17
dashboard-032-massimo-vignelli-typographic-disciplineAdashboard94.4289.5292.9984.89
dashboard-033-motorsport-pit-wallAdashboard86.6967.3066.8077.38
dashboard-035-observability-darkAdashboard91.6787.3883.8185.79
dashboard-036-obsidian-aura-portfolioAdashboard89.5781.7158.1672.77
dashboard-037-pale-grid-node-consoleAdashboard90.0088.7074.0556.47
dashboard-039-retro-lcd-control-panelAdashboard87.9482.2276.9881.11
dashboard-040-rpg-guild-operationsAdashboard88.1086.6792.8691.76
dashboard-041-scada-hmi-industrialAdashboard72.3869.0574.5255.48
dashboard-042-soft-rubber-control-roomAdashboard96.1996.1989.7677.14
dashboard-043-split-flap-departure-boardAdashboard94.2958.5370.7442.81
dashboard-044-sticker-sheet-analyticsAdashboard91.4388.5782.8681.43
dashboard-045-streaming-dashboard-obsAdashboard46.4392.1484.5238.09
dashboard-046-studded-brick-opsAdashboard87.7868.1796.9861.98
dashboard-047-susan-kare-pixel-icon-clarityAdashboard76.3475.7663.9842.66
dashboard-048-teenage-engineering-industrialAdashboard85.2474.7673.3369.76
dashboard-049-topographic-terrain-consoleAdashboard99.1191.0758.8170.42
landing-001-1970s-documentaryAlanding90.2480.4887.6282.62
landing-002-afrofuturismAlanding84.0579.2974.5274.52
landing-003-1980s-arcade-cinemaAlanding94.29100.0082.8672.86
landing-004-70s-warm-retroAlanding87.8685.7197.1485.71
landing-005-april-greiman-digital-collageAlanding87.8690.2490.0090.24
landing-006-acid-graphicsAlanding90.2480.7190.2473.33
landing-007-a24-indie-editorialAlanding85.4883.1096.4396.43
landing-008-abstract-expressionismAlanding83.3385.2487.1458.10
landing-009-adaptive-ambient-uiAlanding97.1477.8674.8672.96
landing-011-80s-synthwaveAlanding86.6779.3979.5268.91
landing-012-academiaAlanding91.6796.4392.8693.57
landing-013-bohemianAlanding100.0085.0095.00100.00
landing-015-biomorphic-designAlanding68.1083.1087.3873.33
landing-016-de-stijlAlanding96.4375.4891.1998.57
landing-017-anodized-metalAlanding87.6284.2978.1084.29
landing-018-indie-sleaze-2005Alanding98.5781.07100.0085.00
landing-019-paul-rand-modernist-witAlanding80.0085.7178.4170.64
landing-020-arc-browser-like-playful-productivityAlanding95.9797.2793.1285.52
landing-023-coastal-grandmotherAlanding92.8685.0091.4380.00
landing-025-bauhausAlanding88.8179.2980.5874.29
landing-026-carbon-fiberAlanding76.1992.8670.0080.95
landing-027-art-decoAlanding50.9592.8665.0067.38
landing-028-christopher-nolan-esqueAlanding85.0082.1487.1488.57
landing-029-aurora-meshAlanding96.4382.3898.5788.57
landing-031-kubrick-symmetryAlanding81.6795.2495.2481.67
landing-032-scandinavian-minimalAlanding97.9697.9691.7797.96
landing-033-underground-brutalismAlanding87.3890.4090.5681.35
landing-034-brushed-aluminumAlanding95.0083.6991.6792.38
landing-035-linen-textureAlanding80.2478.2086.4384.52
landing-036-energy-utilityAlanding100.0093.8196.4380.48
landing-037-constructivismAlanding80.4880.2475.4870.71
landing-038-bold-typographyAlanding100.00100.00100.0085.71
landing-039-blade-runner-noirAlanding78.5784.2980.7180.24
landing-040-letterpressAlanding89.0591.4389.0577.38
landing-041-stop-motion-craftAlanding68.1078.1058.5761.67
landing-042-blueprint-paperAlanding91.4392.6289.0576.67
landing-043-brutalist-graphic-designAlanding79.2996.1991.4387.86
landing-045-glassmorphismAlanding96.5390.0083.2083.81
landing-046-lunarpunkAlanding73.5787.8678.5786.43
landing-047-chalkboard-slateAlanding100.0090.0081.6787.38
landing-048-club-poster-uiAlanding91.4369.2980.0077.86
landing-049-japandiAlanding88.5791.4392.8692.86
landing-050-leather-boundAlanding90.4186.3397.9674.90
landing-051-skate-magAlanding100.0088.5792.14100.00
landing-052-armin-hofmann-contrast-and-reductionAlanding80.9569.5275.7175.71
landing-053-cyberpunkAlanding80.0086.4372.6276.19
landing-054-maritime-heritage-editorialAlanding93.8189.78100.0046.91
landing-055-mid-century-modernAlanding100.0092.8690.2495.71
landing-056-canvas-fabricAlanding95.0088.4193.4195.00
landing-057-space-documentaryAlanding76.1991.4381.9090.00
landing-058-streetwear-dropAlanding89.0587.1481.4382.38
landing-059-art-nouveauAlanding82.8682.8676.0578.20
landing-060-dieter-rams-silent-control-panelAlanding83.6989.0572.7452.98
landing-061-pop-artAlanding90.0093.8190.9588.10
landing-062-cottagecoreAlanding77.6284.7656.6764.76
landing-063-luxury-fashion-filmAlanding78.5772.8665.4878.57
landing-064-op-artAlanding96.5374.73100.0071.67
landing-066-color-fieldAlanding88.1062.9382.3577.48
landing-067-wabi-sabiAlanding95.2495.24100.0085.95
landing-068-dreamcoreAlanding92.8678.1091.6781.67
landing-069-goblincoreAlanding75.7173.5787.8665.71
landing-070-susan-kare-pixel-icon-clarityAlanding89.3987.1474.0176.67
landing-071-logistics-industrialAlanding81.4386.1969.7679.05
landing-072-steampunkAlanding96.4380.2472.3895.24
landing-073-luxuryAlanding95.0091.4390.0090.00
landing-074-solarpunkAlanding81.6782.8668.5778.10
landing-076-massimo-vignelli-typographic-disciplineAlanding93.8196.7991.4389.17
landing-077-material-you-expressiveAlanding96.1981.9073.3349.29
landing-078-vaporwaveAlanding80.2084.9787.6271.05
landing-079-kidcoreAlanding96.4396.4395.0091.67
landing-081-frutiger-aeroAlanding83.8193.81100.0085.24
landing-082-laszlo-moholy-nagy-light-and-new-mediaAlanding92.1485.4892.8682.62
landing-083-legal-authorityAlanding87.3890.2495.2484.05
landing-084-persian-islamic-geometricAlanding86.3170.7150.8361.31
landing-085-quiet-luxuryAlanding100.0095.0086.4395.00
landing-086-skeuomorphic-premiumAlanding85.0091.43100.0078.57
landing-088-gothic-revivalAlanding86.4385.7190.0090.00
landing-089-neumorphismAlanding92.3893.8188.2081.77
landing-091-watercolor-washAlanding82.8680.0064.2967.86
landing-093-grungeAlanding82.8692.8684.2967.86
landing-094-josef-mueller-brockmann-swiss-grid-rhythmAlanding88.5790.0055.4490.82
landing-095-velvet-luxuryAlanding80.2087.3593.0687.35
landing-096-industrialAlanding100.0080.7185.00100.00
landing-098-impressionismAlanding87.9674.8683.2092.72
landing-099-denis-villeneuve-sci-fiAlanding87.8691.4391.4386.43
landing-100-dada-collageAlanding77.8669.2959.2975.71
landing-101-fairycoreAlanding85.4884.0574.0574.05
landing-102-weirdcoreAlanding87.8685.7184.2977.14
landing-103-ikko-tanaka-tradition-as-geometryAlanding79.0578.1086.6775.71
landing-104-gorpcore-outdoor-techAlanding83.8186.6788.1093.81
landing-105-holographic-foilAlanding100.0091.4387.8684.29
landing-106-mermaidcoreAlanding77.3891.6783.5785.95
landing-107-neo-brutalismAlanding100.0083.5796.4390.71
landing-108-cubismAlanding91.7795.2485.9267.38
landing-109-manga-panel-uiAlanding91.0784.6473.1076.67
landing-110-blackout-xerox-punkAlanding69.7669.7660.2470.24
landing-111-cyberdelicAlanding86.6784.2995.2483.81
landing-112-matte-ceramicAlanding93.5792.14100.0093.57
landing-113-studio-ghibli-whimsyAlanding83.8183.8181.6783.10
landing-114-futurismAlanding100.0096.43100.0090.24
track-b-dashboard-001-src-368Bdashboard69.9392.2491.3670.75
track-b-dashboard-002-src-370Bdashboard86.4385.3689.6488.21
track-b-dashboard-003-src-371Bdashboard93.8150.8373.0945.71
track-b-dashboard-004-src-373Bdashboard90.0081.6776.0268.91
track-b-dashboard-005-src-375Bdashboard95.0085.2459.4457.40
track-b-dashboard-006-src-376Bdashboard100.0079.7689.5285.00
track-b-dashboard-007-src-378Bdashboard100.0096.0485.2743.92
track-b-dashboard-008-src-380Bdashboard89.7695.0089.7673.81
track-b-dashboard-009-src-382Bdashboard91.4398.5793.8179.52
track-b-dashboard-010-src-383Bdashboard73.3377.7456.1968.09
track-b-dashboard-011-src-385Bdashboard81.6985.7182.8671.10
track-b-dashboard-012-src-387Bdashboard86.7074.3280.3370.22
track-b-dashboard-013-src-388Bdashboard88.7691.2890.2278.61
track-b-dashboard-014-src-390Bdashboard100.0080.2085.3183.67
track-b-dashboard-015-src-392Bdashboard84.2978.5762.1489.29
track-b-dashboard-016-src-393Bdashboard68.8840.7563.8849.12
track-b-dashboard-017-src-395Bdashboard100.0085.2481.1992.14
track-b-dashboard-018-src-397Bdashboard95.9786.4993.2585.06
track-b-dashboard-019-src-398Bdashboard100.0092.1481.4390.71
track-b-dashboard-020-src-400Bdashboard90.6894.4985.7883.91
track-b-dashboard-021-src-402Bdashboard97.1493.9773.0286.35
track-b-dashboard-022-src-403Bdashboard86.6751.7580.1670.32
track-b-dashboard-023-src-405Bdashboard85.7173.1963.0437.84
track-b-dashboard-024-src-407Bdashboard100.0098.5751.5171.75
track-b-dashboard-025-src-409Bdashboard98.5785.0054.2971.43
track-b-dashboard-026-src-410Bdashboard75.0088.5775.7180.71
track-b-dashboard-027-src-412Bdashboard87.6289.5282.8663.81
track-b-dashboard-028-src-414Bdashboard82.1485.0078.5781.43
track-b-dashboard-029-src-415Bdashboard84.7670.4868.1050.00
track-b-dashboard-030-src-417Bdashboard100.0089.2963.1060.71
track-b-landing-001-src-268Blanding69.3972.2472.9688.57
track-b-landing-002-src-271Blanding100.0092.1491.4387.86
track-b-landing-003-src-275Blanding77.1471.9065.2483.81
track-b-landing-004-src-278Blanding100.0076.5066.6385.34
track-b-landing-005-src-282Blanding91.6798.5795.0091.79
track-b-landing-006-src-285Blanding84.2977.3087.6281.91
track-b-landing-007-src-288Blanding93.8182.3893.8190.60
track-b-landing-008-src-292Blanding93.8192.8679.2981.07
track-b-landing-009-src-295Blanding75.8070.5074.3981.99
track-b-landing-010-src-299Blanding95.4088.8191.9898.57
track-b-landing-011-src-302Blanding100.0082.1186.2688.30
track-b-landing-012-src-306Blanding84.6490.0087.0272.50
track-b-landing-013-src-309Blanding84.5290.2493.8174.01
track-b-landing-014-src-312Blanding95.5695.5694.1392.70
track-b-landing-015-src-316Blanding93.8197.1481.9184.76
track-b-landing-016-src-319Blanding95.71100.0095.7135.08
track-b-landing-017-src-323Blanding100.0093.8191.7761.84
track-b-landing-018-src-326Blanding100.0096.1981.2774.05
track-b-landing-019-src-329Blanding87.6281.9068.7384.60
track-b-landing-020-src-333Blanding78.4175.79100.0086.03
track-b-landing-021-src-336Blanding95.2490.2479.5295.97
track-b-landing-022-src-340Blanding97.14100.00100.0092.96
track-b-landing-023-src-343Blanding83.8174.5283.1085.24
track-b-landing-024-src-347Blanding91.4392.8688.4190.40
track-b-landing-025-src-350Blanding94.29100.0097.1488.81
track-b-landing-026-src-353Blanding87.3884.2986.6793.81
track-b-landing-027-src-357Blanding95.2478.4588.4588.45
track-b-landing-028-src-360Blanding92.3590.3171.4665.99
track-b-landing-029-src-364Blanding100.00100.0095.24100.00
track-b-landing-030-src-367Blanding89.0591.77100.0078.03

Hover a score to read the judge summary. Highlighted cell = best model on that case. Reference column: 4B Muon step200.

+
case_idtracksurface27B AdamW step4009B AdamW step2004B AdamW step2004B Muon step200
dashboard-002-armin-hofmann-contrast-and-reductionAdashboard45.6578.1975.9542.16
dashboard-003-blueprint-hologram-consoleAdashboard67.2576.9988.9567.90
dashboard-004-dashboard-brutalismAdashboard95.5994.4284.9575.64
dashboard-005-dieter-rams-silent-control-panelAdashboard76.1663.0683.9151.65
dashboard-006-carbon-fiber-telemetryAdashboard55.7973.2772.3666.98
dashboard-007-ceramic-bento-metricsAdashboard80.9151.1559.7873.77
dashboard-008-admin-crud-panelAdashboard85.8845.3560.1739.47
dashboard-009-arcade-cabinet-opsAdashboard59.8177.7238.5951.02
dashboard-011-anime-dashboardAdashboard99.3073.5978.4042.85
dashboard-012-glass-cockpit-avionicsAdashboard90.4288.2167.1639.28
dashboard-013-bauhausAdashboard90.0266.3386.8075.48
dashboard-014-developer-first-minimalAdashboard84.8690.3868.3861.76
dashboard-015-ev-instrument-clusterAdashboard97.3662.4254.6842.92
dashboard-016-financial-broadsheetAdashboard84.5869.9457.2956.27
dashboard-018-crypto-dashboardAdashboard90.8676.9748.4078.64
dashboard-019-executive-kpi-dashboardAdashboard85.9576.0260.5868.02
dashboard-020-data-storytelling-uiAdashboard65.4254.4578.0753.85
dashboard-021-fluent-designAdashboard85.3369.6482.6152.03
dashboard-023-adaptive-ambient-uiAdashboard89.8985.4856.8864.80
dashboard-024-honeycomb-paper-ledgerAdashboard80.1767.9569.3160.99
dashboard-027-kenya-hara-white-and-emptinessAdashboard83.1977.7372.0855.17
dashboard-029-lime-noir-logistics-glassAdashboard84.9374.9071.9661.31
dashboard-030-lucky-swaps-volumetric-consoleAdashboard89.0870.2776.3570.99
dashboard-031-mac-os-aqua-nostalgiaAdashboard82.4755.1871.1856.10
dashboard-032-massimo-vignelli-typographic-disciplineAdashboard83.2988.0085.5579.27
dashboard-033-motorsport-pit-wallAdashboard72.1251.6057.5949.39
dashboard-035-observability-darkAdashboard94.6187.4164.7771.01
dashboard-036-obsidian-aura-portfolioAdashboard71.0561.0352.9850.88
dashboard-037-pale-grid-node-consoleAdashboard78.2687.8961.9461.33
dashboard-039-retro-lcd-control-panelAdashboard75.7868.2972.1355.70
dashboard-040-rpg-guild-operationsAdashboard94.9674.7097.5770.08
dashboard-041-scada-hmi-industrialAdashboard74.4975.2354.4249.71
dashboard-042-soft-rubber-control-roomAdashboard94.1092.7180.6560.49
dashboard-043-split-flap-departure-boardAdashboard78.9148.6978.5429.13
dashboard-044-sticker-sheet-analyticsAdashboard90.6983.8955.0065.65
dashboard-045-streaming-dashboard-obsAdashboard13.0974.6784.8433.82
dashboard-046-studded-brick-opsAdashboard82.3563.5878.4546.08
dashboard-047-susan-kare-pixel-icon-clarityAdashboard73.9861.7155.4234.69
dashboard-048-teenage-engineering-industrialAdashboard83.6482.3063.1468.35
dashboard-049-topographic-terrain-consoleAdashboard93.2363.3650.6744.85
landing-001-1970s-documentaryAlanding91.8385.4495.5685.23
landing-002-afrofuturismAlanding94.9694.1783.5588.02
landing-003-1980s-arcade-cinemaAlanding94.76100.0095.9562.55
landing-004-70s-warm-retroAlanding95.6078.3278.3956.58
landing-005-april-greiman-digital-collageAlanding62.6584.9066.5397.18
landing-006-acid-graphicsAlanding96.1669.0888.0076.74
landing-007-a24-indie-editorialAlanding96.3991.4094.6490.97
landing-008-abstract-expressionismAlanding87.1096.3576.1772.23
landing-009-adaptive-ambient-uiAlanding99.5279.0586.7555.49
landing-011-80s-synthwaveAlanding84.3866.8282.1885.89
landing-012-academiaAlanding98.6199.4066.2389.74
landing-013-bohemianAlanding89.3293.9388.3383.80
landing-015-biomorphic-designAlanding73.5785.5181.8874.48
landing-016-de-stijlAlanding99.4077.1298.5386.41
landing-017-anodized-metalAlanding83.0693.2189.6875.77
landing-018-indie-sleaze-2005Alanding83.8694.4693.4558.85
landing-019-paul-rand-modernist-witAlanding68.0895.2495.2180.10
landing-020-arc-browser-like-playful-productivityAlanding98.1498.3591.2989.48
landing-023-coastal-grandmotherAlanding94.0896.3186.2893.10
landing-025-bauhausAlanding81.6458.2283.4861.30
landing-026-carbon-fiberAlanding90.5698.8179.3274.67
landing-027-art-decoAlanding85.8798.8184.7078.66
landing-028-christopher-nolan-esqueAlanding91.7291.5566.4273.87
landing-029-aurora-meshAlanding93.4592.3082.5082.02
landing-031-kubrick-symmetryAlanding90.9999.2199.2194.56
landing-032-scandinavian-minimalAlanding99.6696.6893.8783.59
landing-033-underground-brutalismAlanding89.5687.0989.5087.37
landing-034-brushed-aluminumAlanding89.4075.3278.3168.47
landing-035-linen-textureAlanding94.3387.2086.1988.72
landing-036-energy-utilityAlanding100.0097.7899.4081.75
landing-037-constructivismAlanding71.9791.9494.7286.69
landing-038-bold-typographyAlanding80.4189.3289.3278.54
landing-039-blade-runner-noirAlanding73.5578.7392.0271.33
landing-040-letterpressAlanding93.4172.9784.2363.71
landing-041-stop-motion-craftAlanding74.5585.5284.8288.85
landing-042-blueprint-paperAlanding77.5696.3975.1982.18
landing-043-brutalist-graphic-designAlanding84.0587.8285.7195.60
landing-045-glassmorphismAlanding94.6671.3390.0678.29
landing-046-lunarpunkAlanding93.2193.2169.2855.26
landing-047-chalkboard-slateAlanding93.0995.9583.2583.79
landing-048-club-poster-uiAlanding89.5288.9368.1591.55
landing-049-japandiAlanding95.7178.5179.6383.30
landing-050-leather-boundAlanding87.0991.7799.6670.09
landing-051-skate-magAlanding98.8189.7687.1384.44
landing-052-armin-hofmann-contrast-and-reductionAlanding76.4191.3591.1981.07
landing-053-cyberpunkAlanding87.0291.7982.8270.67
landing-054-maritime-heritage-editorialAlanding94.2183.4786.7962.03
landing-055-mid-century-modernAlanding83.9072.3097.1886.82
landing-056-canvas-fabricAlanding97.9881.7695.9376.52
landing-057-space-documentaryAlanding89.9683.9995.7980.65
landing-058-streetwear-dropAlanding87.9982.1477.6270.39
landing-059-art-nouveauAlanding81.5585.8376.8860.45
landing-060-dieter-rams-silent-control-panelAlanding69.8185.5685.2260.51
landing-061-pop-artAlanding71.5397.7897.3087.73
landing-062-cottagecoreAlanding84.2586.7562.1860.78
landing-063-luxury-fashion-filmAlanding77.9886.5589.4875.69
landing-064-op-artAlanding84.3085.67100.0083.97
landing-066-color-fieldAlanding94.4475.0792.8961.36
landing-067-wabi-sabiAlanding94.8489.5696.4376.71
landing-068-dreamcoreAlanding97.6266.7098.6176.46
landing-069-goblincoreAlanding85.4194.4090.0053.20
landing-070-susan-kare-pixel-icon-clarityAlanding82.0385.1081.0361.77
landing-071-logistics-industrialAlanding94.5294.0183.5388.07
landing-072-steampunkAlanding99.4080.9965.6973.77
landing-073-luxuryAlanding97.9893.8192.3882.30
landing-074-solarpunkAlanding79.1480.4768.3076.17
landing-076-massimo-vignelli-typographic-disciplineAlanding94.2198.2792.0269.41
landing-077-material-you-expressiveAlanding85.9081.3280.6157.01
landing-078-vaporwaveAlanding84.8572.9286.8084.04
landing-079-kidcoreAlanding99.4099.4068.9786.41
landing-081-frutiger-aeroAlanding86.9479.1582.5077.12
landing-082-laszlo-moholy-nagy-light-and-new-mediaAlanding87.1495.2083.7567.71
landing-083-legal-authorityAlanding67.6388.8595.6362.44
landing-084-persian-islamic-geometricAlanding96.5388.4555.4082.00
landing-085-quiet-luxuryAlanding100.0097.9891.7996.19
landing-086-skeuomorphic-premiumAlanding90.9581.67100.0072.55
landing-088-gothic-revivalAlanding88.6995.2497.1494.34
landing-089-neumorphismAlanding91.5988.3187.3868.80
landing-091-watercolor-washAlanding94.7691.9076.7974.95
landing-093-grungeAlanding88.8489.4696.1970.04
landing-094-josef-mueller-brockmann-swiss-grid-rhythmAlanding91.4386.7347.0696.09
landing-095-velvet-luxuryAlanding91.9496.7092.8986.38
landing-096-industrialAlanding91.0776.6087.3883.10
landing-098-impressionismAlanding92.0489.8688.6785.69
landing-099-denis-villeneuve-sci-fiAlanding96.7990.5892.6293.57
landing-100-dada-collageAlanding69.3079.7871.2876.51
landing-101-fairycoreAlanding91.0883.2381.1561.49
landing-102-weirdcoreAlanding96.7984.6396.1962.54
landing-103-ikko-tanaka-tradition-as-geometryAlanding74.6571.6362.5677.98
landing-104-gorpcore-outdoor-techAlanding92.5488.7374.9094.21
landing-105-holographic-foilAlanding100.0087.0280.6596.19
landing-106-mermaidcoreAlanding90.2898.6196.0789.92
landing-107-neo-brutalismAlanding100.0091.3184.1786.31
landing-108-cubismAlanding97.4477.9491.7090.99
landing-109-manga-panel-uiAlanding88.8778.2182.8278.85
landing-110-blackout-xerox-punkAlanding82.2265.1380.7550.68
landing-111-cyberdelicAlanding96.5987.1481.7172.64
landing-112-matte-ceramicAlanding98.9377.6690.9486.79
landing-113-studio-ghibli-whimsyAlanding94.9294.9276.5982.05
landing-114-futurismAlanding100.0090.95100.0093.61
track-b-dashboard-001-src-368Bdashboard82.0567.7087.6163.97
track-b-dashboard-002-src-370Bdashboard83.5865.5391.7978.48
track-b-dashboard-003-src-371Bdashboard88.7355.6486.8347.46
track-b-dashboard-004-src-373Bdashboard72.0771.2060.1342.49
track-b-dashboard-005-src-375Bdashboard88.7285.7868.4246.74
track-b-dashboard-006-src-376Bdashboard92.7665.7376.1951.28
track-b-dashboard-007-src-378Bdashboard89.0071.6770.1655.41
track-b-dashboard-008-src-380Bdashboard79.0775.9879.7155.30
track-b-dashboard-009-src-382Bdashboard80.0985.6586.8058.79
track-b-dashboard-010-src-383Bdashboard73.7984.7253.9864.63
track-b-dashboard-011-src-385Bdashboard87.4188.5977.4453.41
track-b-dashboard-012-src-387Bdashboard87.0052.8166.9856.74
track-b-dashboard-013-src-388Bdashboard88.8477.2385.6058.69
track-b-dashboard-014-src-390Bdashboard86.9168.6463.0358.50
track-b-dashboard-015-src-392Bdashboard82.1271.8051.2980.90
track-b-dashboard-016-src-393Bdashboard79.1258.6287.4760.97
track-b-dashboard-017-src-395Bdashboard93.1678.3857.4577.85
track-b-dashboard-018-src-397Bdashboard90.1962.2389.6660.64
track-b-dashboard-019-src-398Bdashboard78.6978.0970.2768.95
track-b-dashboard-020-src-400Bdashboard90.6081.7377.7969.55
track-b-dashboard-021-src-402Bdashboard83.1884.6553.8160.51
track-b-dashboard-022-src-403Bdashboard88.1462.8464.2067.41
track-b-dashboard-023-src-405Bdashboard90.3880.2681.0141.73
track-b-dashboard-024-src-407Bdashboard93.4090.6957.2457.91
track-b-dashboard-025-src-409Bdashboard84.2082.0256.0157.91
track-b-dashboard-026-src-410Bdashboard65.7869.2368.1169.62
track-b-dashboard-027-src-412Bdashboard90.0170.5374.7952.31
track-b-dashboard-028-src-414Bdashboard79.9779.3057.9170.69
track-b-dashboard-029-src-415Bdashboard90.8981.5258.3547.22
track-b-dashboard-030-src-417Bdashboard88.3090.0452.8148.96
track-b-landing-001-src-268Blanding75.2682.8780.4190.95
track-b-landing-002-src-271Blanding100.0070.5392.6288.06
track-b-landing-003-src-275Blanding92.6292.9478.7390.16
track-b-landing-004-src-278Blanding100.0075.0579.1373.62
track-b-landing-005-src-282Blanding98.6180.8076.2777.54
track-b-landing-006-src-285Blanding85.8380.6782.6393.41
track-b-landing-007-src-288Blanding94.2159.7887.6686.10
track-b-landing-008-src-292Blanding97.7891.6783.2080.94
track-b-landing-009-src-295Blanding91.2191.5183.1176.72
track-b-landing-010-src-299Blanding94.4795.7592.7188.68
track-b-landing-011-src-302Blanding100.0084.2688.5887.24
track-b-landing-012-src-306Blanding82.9271.5590.1091.85
track-b-landing-013-src-309Blanding97.4297.1883.2478.00
track-b-landing-014-src-312Blanding95.6988.1280.7868.09
track-b-landing-015-src-316Blanding85.3479.5293.4186.03
track-b-landing-016-src-319Blanding86.1991.0798.1041.18
track-b-landing-017-src-323Blanding100.0092.8275.0551.85
track-b-landing-018-src-326Blanding100.0081.2690.9384.96
track-b-landing-019-src-329Blanding95.5688.6577.8282.72
track-b-landing-020-src-333Blanding80.6392.3994.0593.51
track-b-landing-021-src-336Blanding99.2190.3876.4075.19
track-b-landing-022-src-340Blanding75.4996.43100.0085.07
track-b-landing-023-src-343Blanding79.1593.3785.6388.78
track-b-landing-024-src-347Blanding97.3896.0095.6975.65
track-b-landing-025-src-350Blanding91.31100.0093.5755.95
track-b-landing-026-src-353Blanding73.6693.8190.8181.91
track-b-landing-027-src-357Blanding83.0174.3693.3187.70
track-b-landing-028-src-360Blanding97.5398.3883.6864.52
track-b-landing-029-src-364Blanding94.0892.6690.0281.42
track-b-landing-030-src-367Blanding96.9886.7896.4388.01

Hover a score to read the judge summary. Highlighted cell = best model on that case. Reference column: 4B Muon step200.