Dontopedia

r_global

From Dontopedia, the open, paraconsistent wiki. (Last updated 2026-06-06.)

r_global has 73 facts recorded in Dontopedia across 27 references, with 6 live disagreements.

73 facts·61 predicates·27 sources·6 in dispute

Mostly:measures(3), rdf:type(3), already computed in(2)

Maturity scale raw canonical shape-checked rule-derived certified

Inbound mentions (23)

Other subjects in dontopedia point AT this entity as a value. These are inverse relationships — e.g. "X motherOf this subject" — and answer questions the forward facts can't. Grouped by predicate.

containsContains(2)

affectsComponentAffects Component(1)

assessesOptimallyAssesses Optimally(1)

comprisesComprises(1)

computesFeatureComputes Feature(1)

computesRGlobalComputes R Global(1)

concatenatesLastConcatenates Last(1)

correspondsToColumnsCorresponds to Columns(1)

derivedFromDerived From(1)

detectsAutomaticallyDetects Automatically(1)

dominantFeaturesDominant Features(1)

gatesUsingGates Using(1)

hasMetricNameHas Metric Name(1)

hasPartHas Part(1)

includesKeyIncludes Key(1)

isReportingIs Reporting(1)

namedNamed(1)

presupposesBrokennessPresupposes Brokenness(1)

presupposesRGlobalComputedPresupposes R Global Computed(1)

usesLoheDenoisingUses Lohe Denoising(1)

usesParameterUses Parameter(1)

usesThroughoutUses Throughout(1)

Other facts (70)

The long tail: predicates that appear too rarely to warrant their own section. Filter or scroll to find a specific one. Each row links to its source.

70 facts
PredicateValueRef
Measuresorder[2]
MeasuresGlobal Metric[22]
MeasuresCross Modal Coherence[27]
Rdf:typeVariable[24]
Rdf:typeParameter[26]
Rdf:typeMetric[27]
Already Computed inFfn Variant V1[3]
Already Computed inFfn Variant V2[3]
Discarded AfterReadout[3]
Discarded AfterReadout[24]
Measured onProduction Multimodal Checkpoint[11]
Measured onProduction Multimodal Checkpoint[26]
Computed inLohe Spherical Ff Nv2[24]
Computed inLohe Spherical Ff Nv1[24]
Located inMethod Call[24]
Located inForward Pass Graph[24]
Central to System DynamicsWhole Thing[1]
Is Already Computedtrue[1]
MonitorsPhase Dynamics[2]
Implemented As Local Variable inCall Method[3]
Thrown Away After Usetrue[3]
Computed Using Formulamx.sqrt((mean_spec * mean_spec).sum(...) + 1e-8)[3]
Essential for Beta CalcThermodynamic Beta[3]
Present inForward Pass Graph[3]
Is Tracked ViaAttention Gates[4]
Is ObservableSystem State[4]
Now Reads True Value≈0.59[5]
Now Comes FromFfn Last Metrics R Global Mean Per Block[5]
Defaulted to at B7815370.5[5]
Came From at B781537Attention Gates or Default 0.5[5]
Is Global Coherence ScalarGlobal Coherence[6]
In the GateCaused Divergence[6]
IndicatesPhase Transition State[6]
Had Value0.59[6]
Occupies Dimension[92][6]
Carries Phase Transition StatePhase Transition[6]
Used in GatingLohe Optimizer[7]
Has Shape(B, T, 1)[8]
Signals Physics ModeLohe Optimizer[9]
Uses ProxyAttention Gate Proxy[10]
Rises Monotonically Across16[11]
Rises Monotonically to0.96[11]
Rises Monotonically From0.07[11]
Increases Continuouslytrue[11]
Rises From0.07[12]
Rises to0.92[12]
Serves As Smoking GunPatch Collapse[13]
Equals0.93[13]
Exhibits Rise Then FallDenoising Trajectory[14]
Is Over Synchronized at Endtrue[14]
Rises Then Fallstrue[14]
High in Exp CExp C[15]
Higher Better ImpliedRaw Norm Gain Mode[15]
Measures Spread on Spherenull[16]
Measures Spreadtrue[17]
Dominate forAntenna Field Mediated[18]
Has Value0.260.26[19]
Indicates Healthy Sync{}[19]
Within Healthy Range0 2 0 3[20]
Measures Global Correlationnull[21]
Is Read Only forDiagnostics[23]
Computed AfterSync[23]
Has ScopeLocal Variable[24]
Statusthrown away[24]
Has Range92[25]
Has Typeglobal coherence scalar[25]
Has Start Value0.07[26]
Has End Value0.96[26]
Change BehaviorMonotonic Rise[26]
Quantified Step Count16[26]

Timeline

Timeline axis is valid_time — when each source says the fact was true in the world, not when Dontopedia learned about it. Retracted rows are kept for provenance; coloured stripes indicate the context kind.

centralToSystemDynamicsblah/watt-activation/part-181
ex:whole-thing
isAlreadyComputedblah/watt-activation/part-181
true
measuresblah/watt-activation/part-182
order
monitorsblah/watt-activation/part-182
ex:phase-dynamics
implementedAsLocalVariableInblah/watt-activation/part-180
ex:call-method
alreadyComputedInblah/watt-activation/part-180
ex:ffn-variant-v1
thrownAwayAfterUseblah/watt-activation/part-180
true
computedUsingFormulablah/watt-activation/part-180
mx.sqrt((mean_spec * mean_spec).sum(...) + 1e-8)
discardedAfterblah/watt-activation/part-180
ex:readout
essentialForBetaCalcblah/watt-activation/part-180
ex:thermodynamic-beta
presentInblah/watt-activation/part-180
ex:forward-pass-graph
alreadyComputedInblah/watt-activation/part-180
ex:ffn-variant-v2
isTrackedViablah/watt-activation/part-179
ex:attention-gates
isObservableblah/watt-activation/part-179
ex:system-state
nowReadsTrueValueblah/watt-activation/part-195
≈0.59
nowComesFromblah/watt-activation/part-195
ex:ffn-last-metrics-r-global-mean-per-block
defaultedToAtB781537blah/watt-activation/part-195
0.5
cameFromAtB781537blah/watt-activation/part-195
ex:attention-gates-or-default-0.5
isGlobalCoherenceScalarblah/watt-activation/part-198
ex:global-coherence
inTheGateblah/watt-activation/part-198
ex:caused-divergence
indicatesblah/watt-activation/part-198
ex:phase-transition-state
hadValueblah/watt-activation/part-198
0.59
occupiesDimensionblah/watt-activation/part-198
[92]
carriesPhaseTransitionStateblah/watt-activation/part-198
ex:phase-transition
usedInGatingblah/watt-activation/part-199
ex:lohe-optimizer
hasShapeblah/watt-activation/part-185
(B, T, 1)
signalsPhysicsModeblah/watt-activation/part-201
ex:lohe-optimizer
usesProxyblah/watt-activation/part-189
ex:attention-gate-proxy
risesMonotonicallyAcrossblah/watt-activation/part-252
16
risesMonotonicallyToblah/watt-activation/part-252
0.96
risesMonotonicallyFromblah/watt-activation/part-252
0.07
increasesContinuouslyblah/watt-activation/part-252
true
measuredOnblah/watt-activation/part-252
ex:production-multimodal-checkpoint
risesFromblah/watt-activation/part-253
0.07
risesToblah/watt-activation/part-253
0.92
servesAsSmokingGunblah/watt-activation/part-272
ex:patch-collapse
equalsblah/watt-activation/part-272
0.93
exhibitsRiseThenFallblah/watt-activation/part-274
ex:denoising-trajectory
isOverSynchronizedAtEndblah/watt-activation/part-274
true
risesThenFallsblah/watt-activation/part-274
true
highInExpCblah/watt-activation/part-280
ex:exp-c
higherBetterImpliedblah/watt-activation/part-280
ex:raw-norm-gain-mode
measuresSpreadOnSphereblah/watt-activation/part-290
null
measuresSpreadblah/watt-activation/part-289
true
dominateForblah/watt-activation/part-382
ex:antenna-field-mediated
hasValue0.26blah/watt-activation/part-407
0.26
indicatesHealthySyncblah/watt-activation/part-407
{}
withinHealthyRangeblah/watt-activation/part-403
ex:0-2-0-3
measuresGlobalCorrelationblah/watt-activation/part-455
null
measuresblah/watt-activation/part-275
ex:global-metric
isReadOnlyForblah/watt-activation/part-423
ex:diagnostics
computedAfterblah/watt-activation/part-423
ex:sync
typeblah/watt-activation/180
ex:Variable
computedInblah/watt-activation/180
ex:LoheSphericalFFNv2
computedInblah/watt-activation/180
ex:LoheSphericalFFNv1
discardedAfterblah/watt-activation/180
ex:readout
hasScopeblah/watt-activation/180
ex:local-variable
locatedInblah/watt-activation/180
ex:method-call
statusblah/watt-activation/180
thrown away
locatedInblah/watt-activation/180
ex:forward-pass-graph
labelblah/watt-activation/197
r_global
hasRangeblah/watt-activation/197
92
hasTypeblah/watt-activation/197
global coherence scalar
typeblah/watt-activation/251
ex:Parameter
labelblah/watt-activation/251
r_global
hasStartValueblah/watt-activation/251
0.07
hasEndValueblah/watt-activation/251
0.96
changeBehaviorblah/watt-activation/251
ex:monotonic-rise
quantifiedStepCountblah/watt-activation/251
16
measuredOnblah/watt-activation/251
ex:production-multimodal-checkpoint
typeblah/watt-activation/255
ex:Metric
labelblah/watt-activation/255
r_global metric
measuresblah/watt-activation/255
ex:cross-modal-coherence

References (27)

27 references
  1. [1]Part 1812 facts
    ctx:discord/blah/watt-activation/part-181
  2. [2]Part 1822 facts
    ctx:discord/blah/watt-activation/part-182
  3. [3]Part 1808 facts
    ctx:discord/blah/watt-activation/part-180
  4. [4]Part 1792 facts
    ctx:discord/blah/watt-activation/part-179
  5. [5]Part 1954 facts
    ctx:discord/blah/watt-activation/part-195
  6. [6]Part 1986 facts
    ctx:discord/blah/watt-activation/part-198
  7. [7]Part 1991 fact
    ctx:discord/blah/watt-activation/part-199
  8. [8]Part 1851 fact
    ctx:discord/blah/watt-activation/part-185
  9. [9]Part 2011 fact
    ctx:discord/blah/watt-activation/part-201
  10. [10]Part 1891 fact
    ctx:discord/blah/watt-activation/part-189
  11. [11]Part 2525 facts
    ctx:discord/blah/watt-activation/part-252
  12. [12]Part 2532 facts
    ctx:discord/blah/watt-activation/part-253
  13. [13]Part 2722 facts
    ctx:discord/blah/watt-activation/part-272
  14. [14]Part 2743 facts
    ctx:discord/blah/watt-activation/part-274
  15. [15]Part 2802 facts
    ctx:discord/blah/watt-activation/part-280
  16. [16]Part 2901 fact
    ctx:discord/blah/watt-activation/part-290
  17. [17]Part 2891 fact
    ctx:discord/blah/watt-activation/part-289
  18. [18]Part 3821 fact
    ctx:discord/blah/watt-activation/part-382
  19. [19]Part 4072 facts
    ctx:discord/blah/watt-activation/part-407
  20. [20]Part 4031 fact
    ctx:discord/blah/watt-activation/part-403
  21. [21]Part 4551 fact
    ctx:discord/blah/watt-activation/part-455
  22. [22]Part 2751 fact
    ctx:discord/blah/watt-activation/part-275
  23. [23]Part 4232 facts
    ctx:discord/blah/watt-activation/part-423
  24. [24]1808 facts
    ctx:discord/blah/watt-activation/180
    • full textwatt-activation-180
      text/plain2 KBdoc:agent/watt-activation-180/85d3802b-f63a-40de-acbc-889407c3b80c
      Show excerpt
      [2026-03-10 01:46] lisamegawatts: yes compute it and use that to modulate, i think we probably need to get rid of softmax or ask for a more principled alternative and a config flag. then on the xero harmonics, ask it about using the tempera
  25. [25]1973 facts
    ctx:discord/blah/watt-activation/197
    • full textwatt-activation-197
      text/plain2 KBdoc:agent/watt-activation-197/63f2f0e2-9ecd-4515-a9de-2f25b46f6dfc
      Show excerpt
      [2026-03-10 05:54] xenonfun: ⏺ Now I can analyze this properly. --- What we actually lose readout_dim = 2·G·H + n_pairs + 1 = 2·32 + 28 + 1 = 93 features is the concatenation of: - [0..31]: post-sync spectra (normalized to S^{H
  26. [26]2517 facts
    ctx:discord/blah/watt-activation/251
    • full textwatt-activation-251
      text/plain1 KBdoc:agent/watt-activation-251/0d79165d-ca43-48df-b924-6b76b157d1a5
      Show excerpt
      [2026-03-12 13:11] xenonfun: ✅ Phase 0 confirmed working — r_global rises monotonically from 0.07 → 0.96 across 16 steps on the production multimodal checkpoint. The architecture supports iterative generation. This is the green light to p
  27. [27]2553 facts
    ctx:discord/blah/watt-activation/255
    • full textwatt-activation-255
      text/plain2 KBdoc:agent/watt-activation-255/58fdaac3-b76b-40d0-b3ab-d254503f97f1
      Show excerpt
      [2026-03-12 16:06] xenonfun: Audio is the killer app here — a mel spectrogram already IS a filterbank of phase-amplitude pairs. The current MelEncoder throws away the phase and compresses to d_model vectors. A phase-native encoder would k

See also

Keep researching

Missing something or suspicious of what's here? Kick off a research session — a Claude agent will investigate, cite its sources, and file new facts into a dedicated context you can review before accepting into the shared view.