NoCap dev-page fact-check checklist (3/3)
Checklist for every scientific numeric claim in cf-pages-nocap-agent-dev/index.html (2026-07-17 jury-ready pass). Repeated values share one row. CSS/layout dimensions are out of scope.
Canonical table: NUMBERS_CANON.md (same folder).
Base path: /home/q/Downloads/software_i_built/modal/
Recompute date: 2026-07-17. CSV first-cross = first row with val_disjoint ≤ 3.3821; T_h = train_time_ms / 3.6e6; ΔT% = 100 × (3.866 − T_h) / 3.866.
Jury-pass status (2026-07-17 QA)
| Gate | Status |
|---|---|
| Overall | PASS — numbers match NUMBERS_CANON.md; no public Modal-profile leaks; claim language clean |
Number diff index.html ↔ canon | PASS — F1_seal 3.866 h; seeds 3.553 / 3.567 / 3.511; ΔT% +8.1 / +7.7 / +9.2; mean ≈+8.3% / ≈3.544 h; stock no-cross / N/A |
CSV recompute (public data/*) | PASS — F1 / s2029 / s4242 / F1_Claim-W2 / s2029 stock no-cross |
| Modal profile leak scan | PASS on deployable tree — profiles only in WORKSPACE_MAP.local.md (must-NOT-deploy); mailto [email protected] OK |
| Byline / Verify / chart / terms / reviewers | PASS — Carol Calin + mailto; #verify + chart_series.json/CSV fetch; challenge terms; #next reviewers; CHALLENGE_TERMS.md linked |
| Forbidden claims | PASS — no matched stock ΔT% speed claim; Public % vs F1_seal only |
Deploy exclude: WORKSPACE_MAP.local.md (must-NOT-deploy), EMAIL_SNIPPET.md (operator-only; prefer exclude).
Jury-ready additions (2026-07-17)
| Claim on page | Verified? | Source |
|---|---|---|
Public data/f1_seal.csv matches F1_seal first-cross | ✅ | data/f1_seal.csv ms 13,918,667.5 → 3.866297 h; same as nocap-repo/results/f1_seal/log.csv pull |
Public data/s2029_cand.csv | ✅ | ms 12,840,827.6 → 3.566897 h · +7.7% |
Public data/s2029_stock.csv — no cross | ✅ | terminal val 3.384017; no row ≤ 3.3821 |
Public data/s4242_cand.csv | ✅ | ms 12,640,470.5 → 3.511242 h · +9.2% |
Public data/f1_claim_w2.csv | ✅ | ms 13,889,386.9 → 3.858163 h (~0.2% vs F1_seal) |
| Seed 1337 labeled summary-only (no public CSV) | ✅ | SEED1337_SUMMARY.md; no data/s1337_*.csv |
First-passage chart sourced from same data/*.csv | ✅ | Verify section fetches same-origin CSVs; optional data/chart_series.json is derived from those files |
| Author byline Carol Calin + mailto contact on page | ✅ | jury-ready hero; no Modal profile names |
Challenge terms: val ≤ 3.3821 ↔ first-passage train_time_ms | ✅ | NoCap-Test README + FAIR_CLOCK + measure section |
| Challenge terms: non-4090 → seal baseline on your GPU ↔ F1_seal A100-SXM4-40GB | ✅ | NoCap-Test README + f1_seal SUMMARY |
| Challenge terms: 1 GPU ↔ attested pin | ✅ | hero pins + scoreboard |
| Challenge terms: fair comparison ↔ same pin + documented recipe | ✅ | RECIPE_CARD / PREREG + recipe section |
| Reviewer section: applied Jul 9; seals Jul 15–16 | ✅ | reviewer block + caveats |
| Compressed workspace table — codes only, honest isolation framing | ✅ | workspace section; real profiles only in WORKSPACE_MAP.local.md |
Evidence state & challenge
| Claim on page | Verified? | Source |
|---|---|---|
| Evidence state 2026-07-16 · 3/3 sealed | ✅ | nocap-repo/results/claim_stack_v1/MULTI_SEED_CLAIM.md |
| GPT-2 ≈124M · d12 · FineWeb | ✅ | challenge framing / production + README |
Quality bar val_disjoint ≤ 3.3821 | ✅ | f1_seal/SUMMARY.md; FAIR_CLOCK; all claim-stack seals |
| One GPU | ✅ | PREREG / challenge |
Headline / hero
| Claim on page | Verified? | Source / recompute |
|---|---|---|
| Seeds 1337, 2029, 4242 | ✅ | MULTI_SEED_CLAIM + three SEED*_SUMMARY |
| T = 3.553 / 3.567 / 3.511 h | ✅ | sealed summaries; CSV for 2029 & 4242 (below) |
| ΔT% = +8.1% / +7.7% / +9.2% | ✅ | MULTI_SEED; recompute from rounded T vs 3.866 |
| F1_seal 3.866 h | ✅ | f1_seal/SUMMARY.md; CSV ms 13,918,667.5 |
| Mean T ≈3.544 h | ✅ | mean(3.553,3.567,3.511)=3.543667 |
| Mean ΔT% ≈+8.3% | ✅ | mean(+8.1,+7.7,+9.2)=+8.337… |
| Terminal CE ≈3.37 | ✅ | terminals 3.3710 / 3.3695 / 3.3702 |
| 3 / 3 seeds under F1_seal | ✅ | MULTI_SEED |
| Each ≥6% under F1 | ✅ | all three ≥7.7% |
| Public % vs F1_seal only; matched cand-vs-stock ΔT% = N/A | ✅ | claim language |
| Matched cand-vs-stock ΔT% N/A | ✅ | stock no-cross / not run |
| GPU A100-SXM4-40GB · torch 2.13.0+cu130 | ✅ | f1_seal SUMMARY |
Scoreboard rows
F1_seal
| Claim | ✅ | Source |
|---|---|---|
| @4736 · val 3.381538 | ✅ | SUMMARY + nocap-repo/results/f1_seal/log.csv |
| 3.866 h · 13,918,667.5 ms | ✅ | SUMMARY (CSV ÷3.6e6 = 3.866297; publish 3.866) |
| Terminal 3.379813 | ✅ | SUMMARY + CSV |
Candidate s1337
| Claim | ✅ | Source |
|---|---|---|
| @4736 · val 3.3741 · T 3.553 h · +8.1% · term 3.3710 | ✅ | SEED1337_SUMMARY.md; MULTI_SEED |
| Workspace Claim-W1 | ✅ | same; real profile in WORKSPACE_MAP.local.md |
| Local cand CSV | ⚠️ absent | rely on sealed SUMMARY (see ACCURACY_NOTES) |
Candidate s2029
| Claim | ✅ | Source / CSV |
|---|---|---|
| @4736 · val 3.3726 · T 3.567 h · +7.7% · term 3.3695 | ✅ | SUMMARY; CSV first-cross val 3.372588, ms 12,840,827.6 → 3.566897 h, ΔT% 7.74 → +7.7%; term 3.369498 |
| Workspace Claim-W2 | ✅ | SUMMARY; real profile in WORKSPACE_MAP.local.md |
| CSV path | nocap-repo/results/claim_stack_v1/seed2029/cand_log.csv |
Candidate s4242
| Claim | ✅ | Source / CSV |
|---|---|---|
| @4736 · val 3.3733 · T 3.511 h · +9.2% · term 3.3702 | ✅ | SUMMARY; CSV val 3.373309, ms 12,640,470.5 → 3.511242 h, ΔT% 9.18 → +9.2%; term 3.370249 |
| cand-only / no stock arm | ✅ | SEED4242_SUMMARY.md |
| CSV path | nocap-repo/results/claim_stack_v1/seed4242/cand_log.csv |
Bar widths (display-only)
| Width | Ratio | ✅ |
|---|---|---|
| 91.9% | 3.553/3.866 | ✅ |
| 92.3% | 3.567/3.866 | ✅ |
| 90.8% | 3.511/3.866 | ✅ |
Stock control & F1_Claim-W2 notes
| Claim on page | ✅ | Source / CSV |
|---|---|---|
| Stock s1337 terminal 3.3848 · no cross | ✅ | SEED1337_SUMMARY |
| Stock s2029 terminal 3.3840 · no cross | ✅ | SUMMARY; CSV terminal 3.384017, no row ≤3.3821 |
| Seed 4242 stock not run | ✅ | SEED4242_SUMMARY / MULTI_SEED |
| Failure ledger “2 / 2 no cross” (paired stocks) | ✅ | correct — not a stale “2/2 seeds sealed” claim |
| F1_Claim-W2 @4768 · val 3.380977 · 3.858 h | ✅ | local F1_BCAROL_SUMMARY; CSV ms 13,889,386.9 → 3.858163 |
| ~0.2% vs F1_seal | ✅ | SUMMARY |
| Seed 2029 7.5% vs F1_Claim-W2 | ✅ | SUMMARY; recompute ≈7.55% |
| 4242 first spawn A100-80GB reject; seal is 40GB | ✅ | SEED4242_SUMMARY |
Fair-clock / methodology
| Claim | ✅ | Source |
|---|---|---|
Primary = first-passage train_time_ms at val≤3.3821 | ✅ | FAIR_CLOCK; PREREG |
| Sliding/EMA do not define T | ✅ | FAIR_CLOCK |
| ~6% single-seed noise floor; 6–10% wants multi-seed | ✅ | FAIR_CLOCK; f1_seal SUMMARY |
| Historical F1 ~229.1 min provisional only | ✅ | f1_seal SUMMARY |
| Published % use F1_seal 3.866 only | ✅ | hard rules + MULTI_SEED |
Timeline / failure ledger (campaign history)
| Claim | ✅ | Source |
|---|---|---|
| Canon ~30% step-time tax parked | ✅ | LESSONS / README |
| Parallel ~1.6% (notebook-era family) | ✅ | LESSONS / FAIR_CLOCK |
| F2 WD256 ended ~3.405 · no cross | ✅ | evidence_ledger / finals |
| F2b WD1024 ended ~3.377 · crossed; −2.9% historical speed retracted | ✅ | f2b PULL_SUMMARY + FAIR_CLOCK |
| claim_muon_par ~3.386 · no cross | ✅ | claim_muon_par/SUMMARY |
| claim_muon_wsd ~3.376 · T 3.899 h ≈0.8% slower than F1 | ✅ | claim_muon_wsd log / board_v3 |
| B32×accum16 ~6% faster core step vs B16; 524,288 tok/update | ✅ | systems_v1 / STACK |
| Cheap split-QKV ≈+0.015 CE @800; +0.0147 @2400 seed 5151 | ✅ | STACK / p2400 SUMMARY |
| Tier A 5/5 no-go (mom 0.90/0.98, Adam WD 0.05/0.2, WD 160) | ✅ | TIER_A_SEAL |
| Curation 800 screens · no promote · claim uses raw FineWeb | ✅ | SCREEN800_SUMMARY |
| Adam LR 0.003 · 800 lead / 2400 gates failed | ✅ | adam_lr persistence SUMMARY |
| Dual-polar ~+0.015 CE · timing ≈1.026 failed | ✅ | theory_split_qkv SUMMARY |
| Split alone ≪ ~0.06 CE equal-speed quality bar | ✅ | STACK / ledger |
Recipe card
| Claim | ✅ | Source |
|---|---|---|
| GPT-2 d12 · raw FineWeb · seq 1024 · val batch 16 · frozen unigram | ✅ | PREREG / kit |
| B32 × accum16 · 524,288 tok · 4,768 updates | ✅ | PREREG |
| Muon 0.02 · mom 0.95 · Muon WD 0.1 · Adam 0.0018 · Adam WD 0.1 · NS 5 | ✅ | PREREG |
| wu 476 · wd 1,192 · zloss 0 · EMA off · val every 128 · save every 256 | ✅ | PREREG / MULTI_SEED |
| bf16 · compile on · single process · parallel block | ✅ | PREREG / kit |
Candidate: muon_split_qkv=1, match_joint=0 | ✅ | PREREG |
| Stock = same stack, joint QKV | ✅ | MULTI_SEED |
Workspaces / paths / next
| Claim | ✅ | Notes |
|---|---|---|
| Clock-W = F1_seal clock | ✅ | real profile in WORKSPACE_MAP.local.md only |
| Claim-W1 = seeds 1337 + 4242 | ✅ | |
| Claim-W2 = seed 2029 + F1_Claim-W2 re-seal | ✅ | public label was F1_Claim-W2 |
| Evidence paths list sealed SUMMARY files | ✅ | match NUMBERS_CANON |
| Optional stock arm on 4242 only if demanded | ✅ | not “optional 3rd candidate seed” — 3rd cand already sealed |
| Public % vs F1_seal only; matched cand-vs-stock ΔT% = N/A | ✅ | footer + hedges |
| Public HTML has no Modal profile real names | ✅ | codes only; see WORKSPACE_MAP.local.md |
Stale-language scan (this index.html)
| Pattern | Found? |
|---|---|
| “2/2” as sealed candidate count | No (only “2 / 2 no cross” for paired stock arms — correct) |
| “optional 3rd seed” for the claim | No (optional stock arm only) |
| mean 3.560 h / 7.9% (old 2-seed mean) | No |
| Published % vs historical F1 / 229.1 | No (229.1 marked provisional only) |
| Matched stock first-cross ΔT% as a speed claim | No (explicitly N/A) |
Numeric mismatches requiring frontend fix: none found against sealed 3/3 sources and CSV recomputes for 2029/4242/F1.
2026-07-17 pre-deploy QA: JURY PASS (see Jury-pass status above). No deploy performed in that QA pass.
ROUND_20260717_pilot (L6 merge · 2026-07-17)
Copy/structure merges into report.html / muon-explained.html only (index.html untouched). Deploy=no. Codes only.
| Change | Verified? | Source / note |
|---|---|---|
#compute triple 72.03 / 110.18 / 147.91 h + 128.28 NoCap-named | ✅ | GPU_HOURS_LOCAL.md · GPU_HOURS_VOLUMES.local.md · GPU_HOURS_BILLED.local.md |
| Ratios 1.34× / 1.16× / 2.05×; notebook ~19.63 h | ✅ | same billed note |
First paragraph states train_time_ms ≠ billed | ✅ | L3 paste |
#insight-kill-bank-seal: F1_seal 3.866 h; mean ΔT% ≈ +8.3%; wsd ~0.8% slower; matched ΔT% N/A | ✅ | NUMBERS_CANON.md · ledger / claim_muon_wsd |
#numbers lead: ≥6% under F1; published % vs F1_seal only; matched ΔT% N/A | ✅ | canon hedges |
#failures hard non-claim (NL) | ✅ | same non-claim; stock no-cross / 4242 not run |
#verification bridge to #compute + Muon primer | ✅ | inventory clocks ≠ claim ΔT% |
muon-explained.html claim-boundary: NS×5 = Keller Ortho; sealed speed = full stack on report — not Muon-alone % | ✅ | L5; no sealed ΔT% moved onto Muon page |
L1 anim-review-fp-pilot.html | ✅ review-only | Accept demo; port to preview/report deferred |
| L2 shell 720 + sticky docnav | ✅ | CSS/chrome only; no claim numbers |
| Modal profile names on public HTML | ✅ none | codes only; mailto author OK |
Numeric mismatches from this round: none vs canon / GPU_HOURS.
ROUND_20260717_r2 (L6 merge · 2026-07-17)
Copy/structure into preview-schema.html / report.html / muon-explained.html (+ CSS skim). index.html untouched. Deploy=no. Codes only. English public copy only.
| Change | Verified? | Source / note |
|---|---|---|
| Schema Figure 1 teach-beat gate → clock → cross | ✅ | THRESH/BASE/S1337 from JSON/canon; legend ΔT% +7.7 / +9.2 / +8.1 unchanged |
#insight-clock-denom: F1_seal 3.866 h; matched ΔT% N/A; no historical-F1 % | ✅ | NUMBERS_CANON.md |
Verify-in-5 / #recompute: 3.3821 · 3.866 · s2029 → +7.7% | ✅ | same worked example; stock N/A; compute tiers separate |
Report #compute tiers 72.03 / 110.18 / 147.91 | ✅ untouched | L3/L6: no rewrite this round |
Muon #split-vs-ortho: split ≠ Ortho/NS; sealed % = full stack on report | ✅ | L5; no Muon-alone sealed % |
| Modal profile names on public HTML | ✅ none | codes only; mailto author OK |
| Nederlands on merged public surfaces | ✅ none | EN voice only |
Numeric mismatches from this round: none vs canon.
Public kit + TRAINER_DELTA (2026-07-17)
| Change | Verified? | Source / note |
|---|---|---|
kit/train_gpt2_split_qkv_cheap_v1.py SHA becfce38…ac9cbc | ✅ | PREREG + Claim-W1/W2 volume kit_override pull |
kit/muon_opt_split_qkv_cheap_v1.py SHA 9086531f…6b1b20 | ✅ | same |
TRAINER_DELTA.md / .html | ✅ | flags muon_split_qkv=1 · match_joint=0; no new sealed % |
| Intro / report / RECIPE / DOCS_INDEX link kit | ✅ | public paths only |
Muon-class jargon explainers (2026-07-17)
| Change | Verified? | Source / note |
|---|---|---|
New qkv-explained.html (Query/Key/Value primer) | ✅ | no sealed %; points to split / F1_seal / report |
Upgraded f1-seal-explained.html worked ΔT% table | ✅ | canon 3.866 / +8.1/+7.7/+9.2 / ≈+8.3% |
Upgraded split-qkv-explained.html + QKV pointer | ✅ | dual-polar 1.026 · flags unchanged |
Upgraded not-claimed-explained.html side-by-side + fig | ✅ | matched ΔT% N/A hedge only |
| Wired intro / report / GLOSSARY / DOCS_INDEX | ✅ | codes only; no new sealed numbers |
Prior checklist staleness
The previous FACT_CHECK_CHECKLIST.md in this folder still described 2/2 seals and mean ≈3.560 h / ≈7.9%. That is obsolete; this file supersedes it.