NoCap-Test evidence · think2earn docs
Source code ↗

Campaign overview

Audience: strangers who want the plot, not the ops manual.
Evidence state: 2026-07-16 · 3/3 claim-stack candidates under F1_seal.
Live report: https://nocap.think2earn.com
Detail timeline: CAMPAIGN_TIMELINE.html · Failures: FAILURE_LEDGER.html

Numbers from sealed summaries — under-claim; no trophy language.


One-line spine

Failed finals → fair clock + F1_seal → Muon compose misses → bank systems + cheap split → Tier A keeps defaults → claim-stack 3/3 under sealed F1 (candidate vs F1, not matched stock speedup).


What the challenge is

Train GPT-2 ≈124M on FineWeb until official disjoint val ≤ 3.3821, on one GPU, with less first-passage train time than a stock baseline. This campaign’s public comparator is our sealed stock AdamW run F1_seal (T=3.866 h on Clock-W).


1 — Early finals

RunOutcomeTakeaway
F2 (short warmdown)Terminal ≈ 3.405 · never crossedQuality miss.
F2b (long warmdown)Crossed @4736 · terminal ≈ 3.377Quality hit. Historical wall ≈ −2.9% vs notebook F1 retracted as speed (infra/noise).

Notebook-era historical F1 (~229.1 min) exists but torch was unlogged → provisional only.


2 — Fair-clock reform

Unpinned torch builds drifted step time (~6% observed). Speed claims now require:

F1_seal (Clock-W): stock AdamW · cross @4736 · val 3.381538 · T=3.866 h.


3 — Full claims that did not under-baseline

RunQualitySpeed vs F1_seal
claim_muon_parMiss (final 3.386130, never crossed)N/A
claim_muon_wsdHit (~3.376 terminal)T3.899 h · ~0.8% slower

Board-v3 800-step screens: no promote. Proxies are not claims.


4 — Component bank (research workspaces)

On Research-W\* accounts, levers were screened and banked:

Through 2026-07-14 evening: bank existed; 0 under-baseline full recipes yet.


5 — Claim-stack seals

Recipe: Keller Jordan Muon + B32 path + timing-cheap split-QKV candidate, vs joint-QKV stock control on the same stack, full 4768.

SeedWorkspaceCandidate Tvs F1_sealStock control
1337Claim-W13.553 h+8.1%matched · never crossed
2029Claim-W23.567 h+7.7%matched · never crossed
4242Claim-W13.511 h+9.2%not run (candidate-only)

All candidates first-cross at step 4736. Mean T3.544 h · mean ΔT% ≈ +8.3%.

Same-workspace AdamW check F1_Claim-W2 on Claim-W2: T=3.858 h (~+0.2% vs F1_seal).

Hard honesty: do not claim matched speedup vs joint stock — stock never crossed where measured.


What strangers should remember

  1. Multi-seed under-baseline vs sealed F1 is sealed and documented.
  2. That is campaign evidence under sealed F1.
  3. Failures stay visible (FAILURE_LEDGER.html).
  4. Attribution: Muon → Keller Jordan; timing-cheap split-QKV → this campaign’s banked mechanism.

Next step in the public narrative: package the writeup / submission — not more GPUs unless an operator asks.