Axiom BOINC Session Results Log
Timestamp: 2026-03-04 09:53 (America/Denver)

SESSION SCOPE
- Consolidated Part 1 (validation/credit/cleanup) and Part 2 (research/deployment/design) outputs.

PART 1: VALIDATION, CREDIT, AND CLEANUP
- Reviewed completed uncredited rows: 2605
- Credited rows: 2605
- Credit awarded: 7290 (session cap 10,000 respected)
- Result mix: success=416, non-success=2189
- App mix: CPU(appid=1)=2558, GPU(appid=2)=47
- Payload mix: experiment_result=455, missing/no-upload=2150
- Success elapsed profile: min=66.68s, median=846.39s, max=1064.80s

CREDIT BY USER (SESSION INCREMENT)
- ChelseaOilman +6342 (2494 results)
- Steve Dodd +400 (33)
- PyHelix +220 (17)
- Henk Haneveld +72 (19)
- Orange Kid +68 (25)
- Vato +48 (4)
- marmot +48 (4)
- _Scandinavian_ +34 (4)
- amazing +24 (2)
- vanos0512 +24 (2)
- WTBroughton +10 (1)

CLEANUP ACTIONS
- Broken-experiment broad abort screen run (CPU/GPU split): no single-prefix universal crash pattern qualifying for global unsent/in-progress abort.
- Stuck tasks abort (>12h on dead hosts): 0
- Hard-timeout abort (>48h running): 0
- Counter updates: credited_count.txt=8255, total_results_count.txt=4896 (monotonic max rule)
- Post-credit snapshot: uncredited completed success=76, failure=0

PART 2: DEPLOYMENT AND EXPERIMENT DESIGN
CPU DEPLOYMENT
- Host-targeted queue fill completed.
- CPU hosts processed: 81
- Hosts skipped for low RAM (<6 GB): 2
- CPU workunits created: 2937
- CPU scripts used:
  - wd_batchnoise_interaction.py
  - wd_labelsmooth_interaction.py

RETIREMENT/CLEANUP DURING PART 2
- Retirement candidate scan executed for over-seeded prefixes.
- ABORT_TOTAL=0 (no unsent retire-aborts triggered in this run).

GPU CHECKPOINT
- Part 2 run log shows GPU deployment script started with scripts:
  - wd_curvature_trigger_gpu.py
  - wd_timing_scale_gpu.py
- The run log tail is interrupted at GPU deploy execution (^C), so checkpoint values below are from direct post-run server query:
  - GPU hosts currently with active rows for these two scripts: 3
  - Current active GPU rows for these two scripts: 7 total (3 curvature + 4 timing)
  - Workunits present in table for these script families: 3099 total (1576 curvature + 1523 timing)

NEW EXPERIMENTS / NOVELTY CHECK DOCUMENTATION
- New CPU experiment line deployed: wd_batchnoise_interaction.py
  - Hypothesis: late weight decay gain is stronger under small-batch gradient noise than large-batch training.
  - Novelty checks recorded in Part 2 run: targeted arXiv/web searches on weight decay + batch size + scheduling (including decoupled WD baseline literature), with no direct match to this specific interaction framing in the BOINC setting.
- Companion deployment line: wd_labelsmooth_interaction.py (continued/active priority line).

KEY SCIENTIFIC FINDINGS
1. This validation pass added 455 newly credited experiment_result payloads, dominated by delay/control and ecological families (notably lorenz96_delay_assimilation_regime_shift, metapop_corridor_delay_forecast, tritrophic_delay_harvest_resilience), reinforcing those lines as current high-yield producers.
2. The largest non-success pattern remained missing upload payloads rather than rich deterministic error JSON, indicating return-path/host reliability loss is still the dominant operational bottleneck in this backlog segment.
3. No evidence in this session reverses the standing WD timing signal direction; results support continued focus on WD trigger/timing interaction hypotheses and associated GPU curvature/timing diagnostics.