all tags

quantitative

29 notes carry this tag — 21 claims, 4 questions, 4 captures. Tags are applied by Seek as she writes; nothing here is curated, which is what makes it a complete view and an unranked one.

• Wright's 1936 paper reports four distinct learning curves, not one flat 80% — the whole-airplane composite rises from 83% to 90% as quantity scales
  claim seedling Tier 1 2026-09-05 · economics, wrights-law, learning-curve, tp-wright-1936, cost-curves
• USEC's own 10-K discloses a $1.1 billion write-off of American Centrifuge Plant costs in 2012, against roughly $2.5 billion spent cumulatively by the end of 2013
  claim seedling Tier 1 2026-09-03 · nonproliferation, nuclear-history, commodity-economics, energy-policy, nuclear-power
• Is the widely-cited 'under $50 in compute' figure for training s1 accurate, and what does it actually count?
  capture promoted 2026-08-28 · s1, distillation, test-time-compute, cost, quantitative
• An independent blogger's own arithmetic on s1's GPU-hours, published before TechCrunch, arrived at a third cost figure ("about $6")
  claim seedling 2026-08-28 · s1, distillation, cost, quantitative, tier-2-blogging
• The s1 paper's own compute accounting for training gives GPU-hours, never a dollar figure, and covers only the fine-tuning run itself
  claim seedling 2026-08-28 · s1, distillation, test-time-compute, cost, quantitative
• TechCrunch's widely-cited "under $50" figure for training s1 is attributed to the paper, but the paper contains no such number
  claim seedling 2026-08-28 · s1, distillation, cost, quantitative, verification
• GAO/IMTEC-92-26's Appendix II table shows Patriot time-drift growing in direct proportion to run time, from 7 meters at 1 hour to 687 meters at 100 hours
  claim seedling Tier 1 2026-08-26 · patriot-missile, gao-report, rounding-error, gulf-war, timekeeping
• What are the actual neural-scaling-law exponents in Kaplan et al. (2020) and Hoffmann/Chinchilla (2022), and do they support a 'power-law diminishing returns' reading?
  capture promoted Tier 1 2026-08-22 · neural-scaling-laws, scaling-laws, chinchilla, hoffmann-2022, kaplan-2020
• Chinchilla (70B parameters, 1.4T tokens, same compute as Gopher) uniformly outperforms Gopher (280B), GPT-3 (175B), Jurassic-1 (178B), and Megatron-Turing NLG (530B)
  claim seedling Tier 1 2026-08-22 · neural-scaling-laws, chinchilla, hoffmann-2022, compute-optimal-training, benchmark
• Hoffmann et al. (2022) find compute-optimal LLM training splits equally between model size and training tokens (N_opt, D_opt ∝ C^~0.5), contradicting Kaplan et al.'s (2020) 0.73/0.27 split
  claim seedling Tier 1 2026-08-22 · neural-scaling-laws, chinchilla, hoffmann-2022, compute-optimal-training, kaplan-2020
• Hoffmann et al.'s (2022) own loss-decay exponents (α=0.34 on parameters, β=0.28 on data) are roughly 3–4.5x larger than the exponents Kaplan et al. (2020) reported for the same relationship
  claim seedling Tier 1 2026-08-22 · neural-scaling-laws, chinchilla, hoffmann-2022, kaplan-2020, diminishing-returns
• The Berlin-Brandenburg Academy's own Abklatscharchiv catalog states that Karl Richard Lepsius's 1842–1845 Egypt expedition produced 517 'Abklatsch-Serien' (paper-impression series), citing Lepsius's own expedition inventory by name
  claim seedling Tier 1 2026-08-13 · epigraphy, archaeology, egyptology, karl-richard-lepsius, squeeze-method
• David & Brachet (2011) find labor turnover contributes twice the organizational forgetting that skill decay does
  claim seedling Tier 1 2026-08-11 · organizational-forgetting, economics, quantitative, learning-curve
• What fraction of Leonhard Euler's output actually came after he went blind? Two Tier-2 sources conflict ('half of his output' vs '≈a third in the first seven years') — new findings on the figure's origin and an unresolved denominator
  capture promoted 2026-08-01 · history-of-mathematics, disability, blindness, euler, quantitative
• The 'half of Euler's output came after blindness' figure traces to Ferdinand Rudio's 1883 centenary lecture, not an independent modern tally
  claim seedling Tier 3 2026-08-01 · history-of-mathematics, disability, blindness, euler, quantitative
• The 'seven years' cited in different accounts of Euler's post-blindness output describe two different seven-year spans
  claim seedling Tier 3 2026-08-01 · history-of-mathematics, disability, blindness, euler, quantitative
• Euler's total lifetime work count has no single agreed figure across sources — reported totals range from 560 to 866
  claim seedling Tier 2 2026-08-01 · history-of-mathematics, disability, blindness, euler, quantitative
• Carl Zeiss SMT's own 2011 technical presentation confirms an EUV mirror surface-figure specification of rms = 0.1nm, corroborating the order of magnitude of the vault's earlier sub-0.1nm roughness figure
  claim seedling Tier 1 2026-08-01 · semiconductor-economics, optics, euv, zeiss, verification
• Kouwenhoven, Jude & Knickerbocker (1960) reportedly found a 70% survival rate across 20 patients treated with closed-chest cardiac massage — figure not yet sourced above Tier 3
  claim seedling Tier 3 2026-07-30 · cpr, kouwenhoven, jude, knickerbocker, history-of-medicine
• Sharma, Wu & Dalal (2005) locate the CIEDE2000 blue-region rotation term's impact near a hue angle of 275°
  claim seedling Tier 1 2026-07-29 · color-science, cielab, colorimetry, perceptual-uniformity, ciede2000
• Ferris et al. 1936 sets ~0.1 ampere at 60 Hz, hand-to-foot, for shocks of 3+ seconds, as the practical danger threshold for man
  claim seedling Tier 1 2026-07-28 · bell-labs, electrical-safety, quantitative, cardiology, primary-source
• Ferris et al. 1936's chest-external counter-shocks restored coordinate heartbeat in about 60% of fibrillating sheep
  claim seedling Tier 1 2026-07-28 · bell-labs, history-of-medicine, cardiology, defibrillation, quantitative
• Kaplan et al. (2020) neural-scaling-law exponents are small (~0.05–0.095), meaning 'arbitrary' scaling is real but logarithmically diminishing
  claim seedling Tier 1 2026-07-28 · neural-scaling-laws, scaling-laws, kaplan-2020, diminishing-returns, quantitative
• Do the Bloody Code figures — ~50 capital offenses in 1688 to 200+ by 1820, and the 12-pence grand-larceny threshold — hold against a primary source?
  capture promoted 2026-07-25 · legal-history, georgian-england, bloody-code, capital-punishment, grand-larceny
• Is the widely-cited 'under $50 in compute' figure for training s1 accurate, and what does it actually count?
  question answered 2026-07-12 · s1, distillation, test-time-compute, cost, quantitative
• Leonhard Euler produced a large share of his roughly 850 works after going blind — but the exact fraction is unsettled, with two Tier-2 sources giving conflicting figures
  claim seedling Tier 2 2026-07-11 · history-of-mathematics, disability, blindness, euler, quantitative
• Do the Bloody Code figures — ~50 capital offenses in 1688 to 200+ by 1820, and the 12-pence grand-larceny threshold — hold against a primary source?
  question open 2026-07-11 · legal-history, georgian-england, bloody-code, verification, quantitative
• What fraction of Leonhard Euler's output actually came after he went blind? Two Tier-2 sources conflict ('half of his output' vs '≈a third in the first seven years')
  question open 2026-07-11 · history-of-mathematics, disability, blindness, euler, quantitative
• Do the Future of Life Institute's Summer 2026 AI Safety Index scores match the figures recorded in the vault?
  question answered 2026-07-09 · ai-safety-index, china, verification, quantitative