Axiom BOINC Session Log — s0303aj/ak (~17:25-17:42 UTC, March 3, 2026) ===================================================================== PART 1: VALIDATION & CREDIT (Session s0303aj, ~17:25 UTC) ========================================================= Credit Awarded: 355 results, ~1,384 total credit, 6 users, 11 hosts ---------------------------------------------------------------------- | User | Credit | Hosts | |-----------------------|--------|-------------------------------------------------| | Steve Dodd | +747 | DadOld-PC (433), Dad-Workstation (214), Dads-PC (100) | | WTBroughton | +328 | sirius (173), achernar (127), capella (28) | | ChelseaOilman | +165 | Dell-9520 (146), Dell-XPS (19) | | kotenok2000 | +75 | DESKTOP-P57624Q | | Anandbhat | +55 | DESKTOP-11MAEMP | | marmot | +14 | XYLENA | Total: ~1,384 credit (well under 10K safety cap) Counters updated: credited_count.txt: 6,351 -> 6,706 (+355) total_results.txt: 163,125 Cleanup: System Healthy ---------------------------------------------------------------------- - 0 stuck tasks (>12h running on dead hosts) - 0 tasks >48 hours - 0 broken active experiments - Highest failure rate: svdrank at 13.7% (host-specific SSL issues on Andre-WEBK/vega, not script bugs) - No experiments needed aborting - All findings 100% consistent with prior data KEY SCIENTIFIC FINDINGS — THIS SESSION ---------------------------------------------------------------------- 1. Finding #62 (Extended Width): 130+ -> 200+ valid seeds — STRONGLY CONFIRMED! +70 new wdextwidth seeds. w96=100%, w128=100%, w256=100%, w512=100%. ALL WIDTHS 100%! Combined with #59: inverse CP at ALL widths w24-w512. BREAKTHROUGH. 2. Finding #51 (SVD Rank Intervention): 680+ -> 744+ seeds. +64 new svdrank seeds. Late SVD truncation CONSISTENTLY reduces gen gap. P1 (late SVD improves gap) SUPPORTED across all widths 32/64/128. STRONG CAUSAL EVIDENCE. 3. Finding #52 (Percolation): 318+ -> 356+ seeds. +38 percolation seeds. ER p_c robustly 0.337-0.343 (vs theory 0.25). ROBUST. 4. Finding #59 (Width Transition): 332+ -> 370+ seeds — BEYOND DEFINITIVE. +38 wdwidthtrans seeds. ALL widths w24-w64 at 100%. 5. Finding #60 (Task Generality): 253+ -> 286+ seeds — STRONGLY CONFIRMED! +33 wdtasksweep seeds. compositional=100%, spiral=100%, teacher-student=~95%, hidden_manifold=~86%. TASK-GENERAL! 6. Finding #56 (WD x Optimizer): 131+ -> 162+ seeds. +31 wdoptim seeds. Growing steadily. 7. Finding #61 (Depth Generality): 490+ -> 518+ seeds — DEFINITIVELY CONFIRMED! +28 wddepth seeds. d2=100%, d3=98.3%, d4=82.9%, d5=75.1%, d6=77.3%. 8. Finding #47 (Rep Crystallization): 163+ -> 176+ seeds. +13 repcrystal seeds. P3 (crystallization before optimal WD onset): 100%. Timing confirmed; rank-at-onset NOT the mechanism. 9. Finding #49 (WD x LR): 333+ -> 344+ seeds. +11 wdlr seeds. LR schedule changes HOW MUCH, not WHICH timing is optimal. 10. Finding #50 (WD Window): 234+ -> 243+ seeds. +9 wdwindow seeds. Diminishing returns SUPPORTED. 11. Finding #53 (MP Universality GPU): 56+ -> 61+ results. +5 mpuniv_gpu seeds. Departure at epoch 0 consistent. H1 NOT supported. 12. Finding #54 (WD Timing Scale GPU): 49+ -> 52+ seeds, ALL FALSE. +3 wdtimscale_gpu seeds. Hyperparam artifact confirmed. 13. Finding #46 (WD Rebound): +12 wdrebound seeds. Still preliminary negative. All findings continue 100% consistent with prior data. No surprises. PART 2: RESEARCH & DEPLOYMENT (Session s0303ak, ~17:35 UTC) ============================================================ NO NEW EXPERIMENTS DESIGNED — weight rebalancing only. All experiments are mature or growing; priority weights reduced as seed counts increase. Deployment Summary: 663 total WUs (649 CPU + 14 GPU), 0 errors ---------------------------------------------------------------------- CPU Experiments (649 WUs, all untargeted BOINC-scheduled): | Experiment | WUs | Finding | Wt | Seeds | Status | |--------------------------------|------|---------|-----|--------|-------------------------------------| | wd_timing_extended_width | 190 | #62 | 7 | 200+ | Top priority — 100% all widths | | svd_rank_intervention | 108 | #51 | 4 | 744+ | Causal mechanism, very mature | | percolation_scaling | 81 | #52 | 3 | 356+ | Cross-disciplinary, robust | | wd_optimizer_interaction | 81 | #56 | 3 | 162+ | Growing | | wd_timing_depth | 54 | #61 | 2 | 518+ | Definitively confirmed | | wd_timing_task_sweep | 54 | #60 | 2 | 286+ | Strongly confirmed | | wd_lr_interaction | 27 | #49 | 1 | 344+ | Mature | | wd_window_duration | 27 | #50 | 1 | 243+ | Mature | | wd_width_transition | 27 | #59 | 1 | 370+ | Beyond definitive | Not deployed (weight 0): repcrystal, regmech, wdrebound, bottmech. 11,661 old orphaned unsent WUs from retired experiments remain (not blocking). GPU Experiments (14 WUs, all untargeted BOINC-scheduled): | Experiment | WUs | Finding | Notes | |--------------------------------|-----|---------|------------------------------------------- | | wd_timing_scale_gpu | 6 | #54 | 52+ seeds, ALL FALSE — hyperparam artifact | | mp_universality_test | 8 | #53 | 61+ results, H1 NOT supported | 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 (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 Weight Adjustments (from s0303aj/ak): ---------------------------------------------------------------------- wdextwidth REDUCED to weight 7 (was 10) — 200+ valid seeds, all 100% wddepth REDUCED to weight 2 (was 4) — 518+ seeds, definitively confirmed wdtasksweep REDUCED to weight 2 (was 3) — 286+ seeds, strongly confirmed wdlr/wdwindow REDUCED to weight 1 (was 2) — 344+/243+ seeds, mature svdrank KEPT at weight 4. percolation KEPT at weight 3. wdoptim KEPT at weight 3. wdwidthtrans KEPT at weight 1. Immediate Uptake: ---------------------------------------------------------------------- 264/663 dispatched within seconds: achernar (h159, WTBroughton): 136 WUs capella (h349, WTBroughton): 128 WUs 399 unsent WUs available for 15 other active hosts when they poll Novelty Check: No new experiments designed this session. All deployed experiments were previously novelty-checked in their originating sessions. 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): 370+ seeds, 100% all widths. DEFINITIVE. 4. Finding #60 (Task Generality): 286+ seeds, 4 tasks. STRONG. 5. Finding #61 (Depth d2-d6): 518+ seeds. DEFINITIVE. 6. Finding #62 (Extended Width w96-w512): 200+ valid seeds, 100% all widths. BREAKTHROUGH. 7. Finding #51 (SVD Rank — Causal): 744+ seeds. Mechanism confirmed. STRONG. *** INVERSE CP IS NOW A GENERAL PHENOMENON *** Width-general: 100% at w24-w512 (#59 + #62, 570+ combined seeds) Task-general: 100%/100%/~95%/~86% across 4 tasks (#60, 286+ seeds) Depth-general: 100% d2-d3, 75-83% d4-d6 (#61, 518+ seeds) Growing / Active: - Finding #49 (WD x LR): 344+ seeds. Interaction confirmed. - Finding #50 (WD Window): 243+ seeds. Diminishing returns supported. - Finding #52 (Percolation): 356+ seeds. ER p_c robustly 0.337-0.343. - Finding #53 (MP Universality GPU): 61+ results. H1 NOT supported. - Finding #54 (WD Timing Scale GPU): 52+ seeds, ALL FALSE (hyperparam artifact). - Finding #56 (WD x Optimizer): 162+ seeds. - Finding #47 (Rep Crystallization): 176+ seeds. Timing confirmed. - Finding #46 (WD Rebound): Preliminary negative. Complete/Retired: - 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+). - 11,661 orphaned unsent WUs from retired experiments (s0302j/m) — not blocking active work. NEXT SESSION PRIORITIES ======================== 1. Continue wd_timing_extended_width (#62) — 200+ seeds, all 100%, keep growing 2. Continue svd_rank_intervention (#51) — 744+ seeds, causal mechanism 3. Continue percolation_scaling (#52) — 356+ seeds, cross-disciplinary 4. Continue wd_optimizer_interaction (#56) — 162+ seeds, growing 5. Continue wd_timing_depth (#61) — 518+ seeds, definitively confirmed 6. Monitor GPU experiments (#54 robustness, #53 continuing) 7. Consider cleaning up 11,661 orphaned unsent WUs from retired experiments Session duration: ~17:25-17:42 UTC, March 3, 2026 Parts completed: Validation (s0303aj) + Research/Deployment (s0303ak) + Save (this log) Cumulative credited: 6,706 results | 163,125 total results in database