E001-SC1 Observable Semantic Slack
Loading the held-out semantic-consistency result, the test run on data the controller never saw. No conclusion appears here until the page knows what evidence stands behind it.
Reading data/e001-semantic-consistency-v1.json…
E001-SC1 · held-out software experiment
Same useful work · result loading
The page will say whether adaptive switching kept the held-out learning and beat the fixed comparator that was frozen before the run.
The infrastructure is simulated, not measured, unless the artifact says otherwise.Every policy gets the same training examples and has to reach the same amount of useful work. The result shows whether switching consistency mode helped, what it saved, and where the controller refused to guess.
The fixed comparator is chosen from calibration data only. Six untouched stress families pair the adaptive and fixed policies against the best whole-policy choice in hindsight. Sampling intervals and infrastructure uncertainty stay separate rather than being merged into one number.
Exact optimizer commits, mode changes, membership, lineage, work, WAN accounting for traffic on the long link between sites, assumptions, uncertainty, and missing evidence all appear below.
Paired held-out effects
Ranking under uncertainty
Each held-out stress family is split into labeled regions of infrastructure uncertainty, showing where adaptive wins, where a fixed policy wins, where the ranking flips, and where the controller declines to rank them at all.
| Family | Uncertainty region | Ranking | Comparator | Reason |
|---|
Selected held-out family
The adaptive and comparator mode intervals on one shared axis, marked with membership, WAN, abstention, merge, and rejoin events.
Untouched family ledger
| Family | Learning delta | Completion ratio | WAN ratio | Work replayed | Energy | Ranking / abstention |
|---|
| Wall / logical tick | Mode / action | Stress / membership | Lineage | Update age / disagreement | Useful / attempted / replayed | WAN payload / time | Held-out NLL | Evidence / abstention |
|---|
not loaded
The raw optimizer-commit trace is not loaded yet. Opening this panel in Full trace fetches the separate artifact it is bound to.
Aligned experiment time
Matched recovery trace
The same failure hits all four policies. The shared clock shows what stops, what is restored, what must be redone, and when useful work catches up.
Every policy track shares one time axis taken from the artifact. Recovery is mechanical: each policy reaches the same durable frontier, the last point safely written to storage, after explicit preemption, restore, replay, and membership changes.
Every mark is drawn from a saved recovery episode. Event IDs, nanosecond bounds, what happened to each unit of work, checkpoint lineage, and matched-frontier hashes stay available in the structured trace.
Measured small-model learning
LC1 · paired interrupted calibration
The page will state the measured answer once the saved artifact loads.
Lower held-out loss means the model predicts better. The cards show how much training each policy attempted to get there, and the lines show what it learned over the same failure clock.
Policy medians summarize matched strata. Inside each stratum, the paired effect subtracts fixed-local progress per FLOP from adaptive progress per FLOP, so zero means the two are equally efficient.
Artifact, dataset, runtime, paired evaluation, and per-run records remain attached below.
Median trajectory
Paired causal contrast
Paired interval not loaded.
Preregistered decision
| Stratum | Fixed progress/FLOP | Adaptive progress/FLOP | Adaptive − fixed |
|---|
| Run | Stratum | Final NLL | Attempted tokens | Energy | Active time |
|---|
not loaded
LC3 · measured equal-work endpoint
Same 524,288 useful tokens · six held-out pairs
The page will state the result once the saved LC3 artifact loads.
Useful work is the fair ruler here. Both policies end with the same 524,288 tokens actually kept by the model. Adaptive avoided work it would have had to redo and finished the schedule sooner with no meaningful loss of learning, but its GPU drew too much energy to pass.
LC3 closes the trap in LC1, where stopping early shrank the denominator and flattered the result. Six paired evaluation strata all end at the same amount of canonical work. Read learning as paired NLL noninferiority, physical work as attempted-FLOP savings, schedule time as opportunity ticks, and energy as a paired device-energy ratio.
Exact pairs, run records, protocol predecessors, hashes, and the modeled-mechanics boundary remain attached below.
Observed policy medians
Paired 90% intervals
Frozen LC3 decision
| Stratum | Fixed NLL | Adaptive NLL | Adaptive − fixed NLL | Attempted-FLOP saved | Ticks saved | Energy ratio |
|---|
| Run | Split | Policy | Interrupted | Final NLL | Attempted / canonical tokens | Opportunity ticks | Device energy | Checkpoints |
|---|
| Artifact path | Conclusion | SHA-256 |
|---|
not loaded
E002-PW1 · measured local power
32 real GPU runs · measurement rejected
The page will state the measurement boundary once the saved artifact loads.
No policy claim is available yet.The experiment ran; the measuring stick failed. All 32 planned comparisons started from the exact same trained model. Asking the sensor every 20 milliseconds did not make it answer that fast: its reading effectively changed about every 495 milliseconds. That is too slow to decide whether sparse continuation saves energy.
The 2×2 cadence-by-continuation design is intact, but the energy quantity it was meant to estimate cannot be used. Both frozen update-count invalidators fired and the logger calibration hit its limit. Arm medians and contrasts stay visible as diagnostics, never as evidence about mechanism.
The 32-run ledger, per-phase metrics, bound source identities, and lazy raw telemetry are attached below.
Validity before effect size
Observed arm medians
No-failure controls
Raw diagnostic contrasts
| Run | Split / block | Arm | Attempted / canonical | Final NLL | Samples / effective updates | Raw board energy | Checkpoints | Raw trace SHA-256 |
|---|
Phase energy is kept for diagnosis. It cannot support a causal comparison between policies, because the logger invalidators fired.
| Phase | Duration | Idle-subtracted board energy | Energy / canonical token | Effective update equivalents |
|---|
not loaded
Raw telemetry is not loaded yet. Opening this panel in Full trace fetches the separate point artifact.
| Point | Time | Phase at timestamp | Board power | GPU util. | Memory util. | SM / memory clock | Temperature | P-state |
|---|
E002-PW2 · valid cumulative energy
32 real GPU runs · measurement valid
The page will state the mechanism once the saved artifact loads.
Mechanism estimate loading.Checkpointing less often fixed the measured energy problem without giving back the recovery win. Both policies kept exactly 524,288 useful tokens. Sparse continuation attempted less work, finished 40 scheduling ticks sooner, and stayed inside the frozen GPU-energy limit. One version of the calculation subtracts an estimated idle baseline; that version is less certain and cannot rule out that the effect is zero.
PW2 repeats PW1’s frozen 2×2 design, this time with a cumulative-energy counter the hardware actually supports. No measurement invalidator fired. Positive total and checkpoint-group interactions pass the frozen primary attribution gates, and sparse continuation holds the learning, work, schedule, and energy gates. The sensitivity-only, idle-subtracted interaction crossed zero, so the result does depend on how the idle baseline is handled.
The 32-run ledger, selected-run counter phases, bound identities, and lazy raw counter points are attached below.
Measurement validity
Paired causal contrasts
Counter support by phase
Frozen decision
| Run | Split / block | Arm | Attempted / canonical | Final NLL | Polls / counter updates | Run energy | Checkpoints | Counter trace SHA-256 |
|---|
| Phase | Evidence class | Duration | Energy | Energy / canonical token | Counter updates | Effective update equivalents |
|---|
not loaded
Raw counter telemetry is not loaded yet. Opening this panel in Full trace fetches the separate point artifact.
| Point | Time | Phase at timestamp | Cumulative board energy | Counter update | Ancillary board power | GPU util. | Temperature |
|---|
E002-PW3 · physical rack mechanism
Physical result loading
The page will explain what moved, what the rack meter saw, and whether learning stayed fixed.
Nothing about this rack is inferred from the earlier laptop experiment.The same training and recovery work runs several ways. The shaped run is allowed to move only the operations that have real timing slack. The rack meter then says whether spreading those operations out really did soften the electrical spike.
Paired blocks hold useful work, state generations, failures, and learning commitments fixed. The only thing that changes is when checkpoint and rejoin traffic is allowed to be released.
Exact event intervals, clock alignment, sensor coverage, semantic obligations, and chunk hashes appear below.
Same clock · measured boundary
Paired power traces from the rack PDU, the unit that distributes power to the rack, drawn above the checkpoint and recovery event rails for the job.
Paired physical effects
Five frozen policies
| Policy | Rack ramp | 0.1–10 Hz energy | Useful-token rate | Rack J / token | p95 recovery | Held-out NLL |
|---|
Preregistered decision
| Arm / job | Operation | Generation | Ready → release | Actual interval | Bytes | Outcome |
|---|
not loaded
E001 v1 screening artifact
| Policy | Local steps / decision | Progress per FLOP | Inter-site bytes | Time to target | Base + compute energy | Falsifier status |
|---|
Waiting for a generated artifact. Result cells remain not run.
E001 v1 screening evidence chain
Source observationsUNMEASURED · UNAVAILABLE FROM ARTIFACT
Unfitted sensitivity priorPRIOR · NOT FITTED
Progress per FLOPUNMEASURED
Held-out time to targetUNMEASURED
Published source measurements are read from the observatory artifact. Uncertainty that comes from rounding in the paper is tracked separately from real run-to-run variation.
| Observation | Value | Evidence | Action |
|---|
The attached literature covers one narrow delay setting. It says nothing about progress per FLOP, longer local-update intervals, frontier-scale transfer, how a multi-site run behaves when it is interrupted, or a controller that acts during an outage.
Repeated small-model delay calibrationMeasure several delay intervals and optimizers instead of extrapolating from a single step.
Held-out optimizer, model, and site combinationsTest combinations that were deliberately left out when the prior was built.
Controlled 30B to 100B-plus multi-site runVary delay and cadence under one stated policy, counting the evaluation the same way every time.
| After | Observed state | Decision | Applies to | Evidence |
|---|
not run