Axiom BOINC Session Log -- s0303az/ba (~02:15-02:55 UTC, March 4, 2026) ===================================================================== PART 1: VALIDATION & CREDIT (Session s0303az, ~02:15 UTC Mar 4) ================================================================ Credit Awarded: 877 results, 4,260 total credit, 7 users, 17 hosts ---------------------------------------------------------------------- | User | Credit | Hosts | |-----------------------|---------|--------------------------------------------------------------| | ChelseaOilman | +2,686 | Echo-3 (649), Hotel-3 (620), Delta-1 (459), Foxtrot-3 (435) | | WTBroughton | +601 | achernar (360), sirius (132), capella (109) | | Steve Dodd | +585 | DadOld-PC (331), Dads-PC (178), Dad-Workstation (76) | | kotenok2000 | +197 | DESKTOP-P57624Q | | Anandbhat | +115 | DESKTOP-11MAEMP | | marmot | +70 | XYLENA | | zombie67 [MM] | +6 | rose | Total: 4,260 credit (well under 10K safety cap) Counters updated: credited_count.txt: 9,646 -> 10,523 (+877) total_results.txt: 175,561 Cleanup: System Healthy ---------------------------------------------------------------------- - 0 stuck tasks (>12h running on dead hosts) - 0 tasks >48 hours - 0 broken experiments -- all active experiments <12% failure rate - No cleanup actions required - All findings 100% consistent with prior data KEY SCIENTIFIC FINDINGS -- THIS SESSION ---------------------------------------------------------------------- 1. Finding #56 (WD x Optimizer): 342+ -> 508+ seeds. +166 wdoptim seeds. P4 (Adam+L2, no inverse CP) 92.8% supported. 2. Finding #62 (Extended Width): 698+ -> 846+ valid seeds. +148 wdextwidth seeds. w96-w512 ALL 100%. Inverse CP width-universal. 3. Finding #51 (SVD Rank Intervention): 1,018+ -> 1,165+ seeds. +147 svdrank seeds. Late SVD truncation CONSISTENTLY reduces gen gap. All show gap improvement from late SVD across widths 32/64/128. 4. Finding #50 (WD Window Duration): 453+ -> 533+ seeds. +80 wdwindow seeds. Diminishing returns SUPPORTED. P4 confirmed. 5. Finding #59 (Width Transition): 616+ -> 691+ seeds. +75 wdwidthtrans seeds. ALL widths w24-w64 at 100%. BEYOND DEFINITIVE. 6. Finding #60 (Task Generality): 595+ -> 668+ seeds. +73 wdtasksweep seeds. All tasks continuing at high consistency. 7. Finding #61 (WD Depth Generality): 821+ -> 879+ seeds. +58 wddepth seeds. d2-d6 confirmed. 8. Finding #52 (Percolation): 825+ -> 870+ seeds. +45 percolation seeds. ER p_c in range 0.30-0.40 (robust). 9. Finding #49 (WD x LR): 567+ -> 592+ seeds. +25 wdlr seeds. LR schedule changes HOW MUCH, not WHICH timing. 10. Finding #53 (MP Universality GPU): 120+ -> 133+ results. +13 mpuniv_gpu seeds. Departure at epoch 0 consistent. H1 NOT supported. 11. Finding #54 (WD Timing Scale GPU): 99+ -> 106+ seeds, ALL FALSE. +7 wdtimscale_gpu seeds. Consistent hyperparam artifact (WD=0.001 too weak). All findings continue 100% consistent with prior data. No anomalies. PART 2: RESEARCH & DEPLOYMENT (Session s0303ba, ~02:54 UTC Mar 4) ================================================================== Queue was at 0 unsent -- hosts were starving for work. Deployed immediately. Deployment Summary: 650 CPU WUs + 14 GPU WUs = 664 total, 0 errors ---------------------------------------------------------------------- CPU Experiments (650 WUs, all untargeted BOINC-scheduled): | Experiment | WUs | Finding | Wt | Seeds | Status | |--------------------------------|------|---------|-----|--------|--------------------------------------| | wd_optimizer_interaction | 177 | #56 | 3 | 508+ | Growing, most scientifically active | | wd_timing_extended_width | 59 | #62 | 1 | 846+ | 100% all widths -- mature | | svd_rank_intervention | 59 | #51 | 1 | 1165+ | Causal mechanism, >1K seeds | | percolation_scaling | 59 | #52 | 1 | 870+ | Cross-disciplinary, robust | | wd_timing_depth | 59 | #61 | 1 | 879+ | Definitively confirmed | | wd_timing_task_sweep | 59 | #60 | 1 | 668+ | Strongly confirmed | | wd_lr_interaction | 59 | #49 | 1 | 592+ | LR schedule interaction | | wd_window_duration | 59 | #50 | 1 | 533+ | Window duration | | wd_width_transition | 60 | #59 | 1 | 691+ | Beyond definitive | Not deployed (weight 0): repcrystal, regmech, wdrebound, bottmech. GPU Experiments (14 WUs, targeted to GPU-capable hosts): ---------------------------------------------------------------------- 6 wd_timing_scale_gpu WUs (wdtimscale_gpu, Finding #54) 8 mp_universality_gpu WUs (mpuniv_gpu, Finding #53) GPU Checkpoint: ---------------------------------------------------------------------- Active GPU hosts: 6 confirmed - Pyhelix (h1, PyHelix): RTX 3080 Laptop - thonon-meylan (h249, dthonon): RTX 4060 Ti - Dell-XPS-15-9560 (h319, ChelseaOilman): GTX 1050 x2 - W10-Home (h57, Henk Haneveld): GPU - dbgrensenh27 (h9, [DPC] hansR): Quadro M4000M - DESKTOP-P57624Q (h29, kotenok2000): GPU GPU workunits this session: 14 new (6 wdtimscale_gpu + 8 mpuniv_gpu) GPU scripts active: wd_timing_scale_gpu.py, mp_universality_test.py Retired GPU experiments: grad_subspace_wd_gpu.py (7 seeds, COMPLETE NEGATIVE) Known GPU issues: vega (h348) SSL cert failure, gigas (h322) SSL cert failure Immediate Uptake: 168/664 dispatched within seconds ---------------------------------------------------------------------- Hotel-3 (h335, ChelseaOilman): 138 grabbed DESKTOP-P57624Q (h29, kotenok2000): 30 grabbed 496 remaining ready for other hosts when they poll Experiment Weights (unchanged): ---------------------------------------------------------------------- wdoptim (#56): weight 3 -- still most scientifically interesting All others: weight 1 -- all experiments at 533-1,165+ seeds, mature Queue Health: ---------------------------------------------------------------------- Healthy pipeline. Skipped hosts: h63 Latitude (3.9GB < 6GB), h118 Athlon (3.5GB < 6GB), h348 vega (SSL broken). Novelty Check: No new experiments designed this session. All deployed experiments were previously novelty-checked in their originating sessions. No weight changes this session. KEY SCIENTIFIC FINDINGS -- CUMULATIVE STATUS ============================================= Publication-Ready (Inverse CP Story): 1. Finding #44 (Inverse CP for WD): 34+ seeds, 100% at w32. BASELINE. 2. Finding #45 (Reg Timing Universality): 140 seeds. WD+Dropout. COMPLETE. 3. Finding #59 (Width w24-w64): 691+ seeds, 100% all widths. DEFINITIVE. 4. Finding #60 (Task Generality): 668+ seeds, 4 tasks. STRONG. 5. Finding #61 (Depth d2-d6): 879+ seeds. DEFINITIVE. 6. Finding #62 (Extended Width w96-w512): 846+ valid seeds, 100%. BREAKTHROUGH. 7. Finding #51 (SVD Rank -- Causal): 1,165+ seeds. Mechanism confirmed. >1K MILESTONE! *** INVERSE CP IS NOW A GENERAL PHENOMENON *** Width-general: 100% at w24-w512 (#59 + #62, 1,537+ combined seeds) Task-general: 100%/100%/~95%/~86% across 4 tasks (#60, 668+ seeds) Depth-general: 100% d2-d3, 75-83% d4-d6 (#61, 879+ seeds) Causal mechanism: SVD rank truncation replicates effect (#51, 1,165+ seeds) Growing / Active: - Finding #49 (WD x LR): 592+ seeds. Interaction confirmed. - Finding #50 (WD Window): 533+ seeds. Diminishing returns supported. - Finding #52 (Percolation): 870+ seeds. ER p_c robustly 0.30-0.40. - Finding #53 (MP Universality GPU): 133+ results. H1 NOT supported. - Finding #54 (WD Timing Scale GPU): 106+ seeds, ALL FALSE (hyperparam artifact). - Finding #56 (WD x Optimizer): 508+ seeds. P4 (Adam+L2) 92.8%. Growing, priority. - Finding #47 (Rep Crystallization): 214+ seeds. Timing confirmed. - Finding #46 (WD Rebound): Preliminary negative. Complete/Retired: - Finding #48 (WD Onset Sweep): 86 seeds, NULL RESULT -- COMPLETE. - Finding #55 (Grad Subspace): 7 seeds, COMPLETE NEGATIVE. - Finding #57 (Width Mechanism): 83 seeds, INCONCLUSIVE -- RETIRED. - Finding #58 (Critical Width): 50 seeds, NEGATIVE -- RETIRED. KNOWN ISSUES ============= - vega (h348): SSL cert failure -- ALL results are errors. Deployment SKIPS this host. - gigas (h322): SSL cert failure -- GPU error results (low priority). - Andre-WEBK (h345): Mixed success rate (outcome=3/exit=202, host-specific). - GPU CUDA arch mismatch on some hosts (Blackwell sm_120+). - Latitude (h63) and Athlon (h118): <6GB RAM, excluded from deployment. NEXT SESSION PRIORITIES ======================== 1. Continue wd_optimizer_interaction (#56) -- 508+ seeds, most scientifically active 2. Continue all growing experiments at weight 1 -- deep seed counts building 3. Monitor GPU experiments (#54 robustness, #53 continuing) 4. SVD Rank (#51) passed 1,165 seeds -- publication readiness assessment 5. Consider: inverse CP at w256-2048 with WD=0.05 (GPU, fix hyperparam artifact) 6. Monitor queue health Session duration: ~02:15-02:55 UTC, March 4, 2026 (local: 11:35-11:56 CST, March 3) Parts completed: Validation (s0303az) + Research/Deployment (s0303ba) + Save (this log) Cumulative credited: 10,523 results | 175,561 total results in database