| Age | Commit message (Expand) | Author |
|---|
| 2026-07-18 | RESULT 48b: 一般化BBP(amp+测谱,免结构假设)=R>1处处成立到1e-4,E... | Yuren Hao |
| 2026-07-18 | RESULT 47+48: 涨的量=blocks8-11单模态增益(ρ复刻探针,两血统同... | Yuren Hao |
| 2026-07-18 | RESULT 46: endgame-42落地 — BP lr平台平坦(参照1.2141成立), plain@... | Yuren Hao |
| 2026-07-18 | RESULT 45: no-floor判决=走廊真闭合(β 3e-3→1.1e-5无搁板,天花板... | Yuren Hao |
| 2026-07-18 | RESULT 43+44: crown-3 死因=cap硬底钉在塌落天花板之上(变相wall-... | Yuren Hao |
| 2026-07-18 | RESULT 42: crown-3 voided — slow-burn endgame instability accepted by hard ... | Yuren Hao |
| 2026-07-18 | Noise fix quantified (2us window = 3.2x headroom at 1.3x schedule) + JLC boar... | Yuren Hao |
| 2026-07-18 | --bpmix: gap-source ablation (overwrite EP grads with true BP grads per modul... | Yuren Hao |
| 2026-07-18 | RESULT 41: crown-3 sealed 3.3728 (1.0x cost, gap 0.084); 0.041 plain-vs-cent ... | Yuren Hao |
| 2026-07-18 | Fix: Sam Dillavou is one person (UPenn), not two collaborators | Yuren Hao |
| 2026-07-18 | Fab strategy: JLC SMT + socketed screened MDACs + DFT layout + first-article ... | Yuren Hao |
| 2026-07-17 | RESULT 40: ride-v2 probes — bug fix was the disease, sync-accept adopted; e... | Yuren Hao |
| 2026-07-17 | RESULT 39: seed verdict — ride_s1 lucky draw; RESULT 26 zero-gap anchored o... | Yuren Hao |
| 2026-07-17 | Energy ledger: SPICE core 2.87pJ/MAC; boards lose ~1000x (trainability demos)... | Yuren Hao |
| 2026-07-17 | Stage C: SPICE-in-the-loop training — SPICE == behavioral == ideal (0.0213/... | Yuren Hao |
| 2026-07-17 | hw_sim v1.1: ENOB-after-cal 8.51b (clears 7.0 bar; per-bit mismatch model) + ... | Yuren Hao |
| 2026-07-17 | hw_sim: bus-settle fix menu parked as Dillavou session question (incl user's ... | Yuren Hao |
| 2026-07-17 | hw_sim v1: datasheet-anchored column — C_bus=5.4nF kills naive design (16.6... | Yuren Hao |
| 2026-07-17 | hw_sim: circuit-as-code workflow stood up (PySpice/ngspice/skidl); cell_v0 ... | Yuren Hao |
| 2026-07-17 | RESULT 38: 2D update-field analysis — localized curl blob at beta=1e-2 (con... | Yuren Hao |
| 2026-07-17 | --beta_sync: synchronous acceptance (this-step relax telemetry gates the comm... | Yuren Hao |
| 2026-07-17 | ride-v2 (b)/(c) flag-gated (--ride_ema/--ride_cool, default off); isolation p... | Yuren Hao |
| 2026-07-17 | ride-v2: (a) floor-jump cap continuity (the 0.09 bug), (b) rho-EMA decisions,... | Yuren Hao |
| 2026-07-17 | RESULT 37 addendum: first wound = cap x floor-jump = beta 0.09 at step 20000 ... | Yuren Hao |
| 2026-07-17 | RESULT 37: ride fails at 72M (beta-scaled damage through beta-independent gua... | Yuren Hao |
| 2026-07-17 | Fairness protocol at scale: published-recipe anchor + proxy sweeps + sensitiv... | Yuren Hao |
| 2026-07-17 | RESULT 36 pre-reg: fw72m_ride crown-3 + ride/BP seed campaign + BP-fairness r... | Yuren Hao |
| 2026-07-17 | save final ckpt at last step (fixed-eval protocol); fw72m_ride crown-3 launch... | Yuren Hao |
| 2026-07-16 | RESULT 34+35: beta-buyback partial (+0.035->+0.022); full-epoch trio sealed ... | Yuren Hao |
| 2026-07-16 | est15 sealed 1.2374 (band edge); beta-buyback arm qb6_b1e2 launched (RESULT 3... | Yuren Hao |
| 2026-07-16 | RESULT 33: triple verdict — A/B neutral at 72M, dose-response monotone thru... | Yuren Hao |
| 2026-07-16 | Deck finalized: both verdicts sealed (C512 zero-gap, 72M gap 0.043), decoupli... | Yuren Hao |
| 2026-07-16 | RESULT 32: crown resealed — fw72m_cent 3.3318 vs BP 3.2884 (gap 0.043, 10x ... | Yuren Hao |
| 2026-07-16 | --beta_ride ceiling-riding governor (two-sided bcap: climb in calm, attack ne... | Yuren Hao |
| 2026-07-16 | RESULT 31: decoupling — tail win is 101% beta, centered CE-neutral at 42M t... | Yuren Hao |
| 2026-07-16 | RESULT 30: quantization Delta-vs-Delta — zero-tax at 8-bit operating point ... | Yuren Hao |
| 2026-07-16 | RESULT 29: qcomp8 tax-free (+0.0018) = T64 8-bit compute green light; --centm... | Yuren Hao |
| 2026-07-16 | RESULT 28 caveat: fp32-EP control insufficient — BP mirror is the decision ... | Yuren Hao |
| 2026-07-16 | Stage-0 BP mirror arms: qup/qcomp flags in casc_bp_train; bp_qctl + 6 quant a... | Yuren Hao |
| 2026-07-16 | RESULT 28: Stage-0 gate #1 verdict (naked resident cells need >=10-12b; T64 w... | Yuren Hao |
| 2026-07-16 | Stage-0 gate #1b: --qcomp_bits (compute on DAC-grid weights, fp32 master = T6... | Yuren Hao |
| 2026-07-15 | T64 procurement sheet: Taobao AD7528JN @ ¥3/pc (user-verified) — MDAC line... | Yuren Hao |
| 2026-07-15 | Stage-0 gate #1: --qup_bits (absolute-grid stochastic-rounding weight quantiz... | Yuren Hao |
| 2026-07-15 | RESULT 27: K-saturation across training stages — K3=K8 at s5k/50k/100k; fix... | Yuren Hao |
| 2026-07-15 | ERRATUM RESULT 21: original best 3.7117 was @104.8k not @185k — 90k-step wa... | Yuren Hao |
| 2026-07-15 | RESULT 26: C512 ZERO-GAP SEALED — stage1b_cent 1.2334 vs BP 1.2311 (+0.0023... | Yuren Hao |
| 2026-07-15 | External notes: sliding-mode clamping (Stage-0 injection rule), saturated-uni... | Yuren Hao |
| 2026-07-15 | assets/: organize into decks|figs|hw|refs|misc; scripts + gitignore repointed | Yuren Hao |
| 2026-07-15 | Alexi primer deck: 10 slides (arch op-by-op, EP phases, schedule/parallelism,... | Yuren Hao |
| 2026-07-15 | centfast (doubled-batch +/-beta centered, cos 1.000000000 vs sequential, DDP-... | Yuren Hao |
