Helmholtz

as donto believed on

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

Helmholtz has 66 facts recorded in Dontopedia across 11 references, with 3 live disagreements.

66 facts·58 predicates·11 sources·3 in dispute

Mostly:rdf:type(3), has tokens per second(2), has lower capacity than(2)

Inbound mentions (31)

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.

`hasBlock`Has Block(3)

G2 H4 Qpsk8 Helmholtzex:g2-h4-qpsk8-helmholtz
G8 H2 Bpsk16 Helmholtzex:g8-h2-bpsk16-helmholtz
G8 H4 Ofdm32 Helmholtzex:g8-h4-ofdm32-helmholtz

`characterizes`Characterizes(2)

Bandpass Signatureex:bandpass-signature
Frequency Division Multiplexingex:frequency-division-multiplexing

`appliedTo`Applied to(1)

Eliminating Python Loopsex:eliminating-python-loops

`closesBpbGapBetween`Closes Bpb Gap Between(1)

Clifford Readoutex:clifford-readout

`comparesFFNType`Compares Ffn Type(1)

Ffn Comparison Table 1ex:ffn-comparison-table-1

`comparesPerLayer`Compares Per Layer(1)

Spectral Characterization Postex:spectral-characterization-post

`concerns500StepTraining`Concerns500 Step Training(1)

Spectral Characterization Postex:spectral-characterization-post

`consistsOf`Consists of(1)

Full Cl 3 0 Swarmex:full-cl-3-0-swarm

`contrastsWith`Contrasts With(1)

Cayley Rotor Designex:cayley-rotor-design

`exhibitsSamePatternAs`Exhibits Same Pattern As(1)

Rotor L 6 Ofdmex:rotor-l-6-ofdm

`hasBpbValueFor`Has Bpb Value for(1)

Step 950ex:step-950

`hasCapacityValueFor`Has Capacity Value for(1)

Step 950ex:step-950

`hasConsistentlyHigherAlignDeltaThan`Has Consistently Higher Align Delta Than(1)

V3ex:v3

`hasDifferentEncodingStrategyFrom`Has Different Encoding Strategy From(1)

V3ex:v3

`hasHigherThroughputThan`Has Higher Throughput Than(1)

V3ex:v3

`hasLowerSpectralEntropy`Has Lower Spectral Entropy(1)

V3ex:v3

`hasSameInformationContentAs`Has Same Information Content As(1)

V3ex:v3

`hasSnrValueFor`Has Snr Value for(1)

Step 950ex:step-950

`hasThroughputValueFor`Has Throughput Value for(1)

Step 950ex:step-950

`improvesOver`Improves Over(1)

Clifford Readoutex:clifford-readout

`includes`Includes(1)

Manifold Methodsex:manifold-methods

`includesHigherSpectralEntropy`Includes Higher Spectral Entropy(1)

Distinct Dynamicsex:distinct-dynamics

`includesLowerSnr`Includes Lower Snr(1)

Distinct Dynamicsex:distinct-dynamics

`involvesModel`Involves Model(1)

Experiment 1ex:experiment-1

`mentionsEntity`Mentions Entity(1)

Message 01 44 Statsex:message-01-44-stats

`shareKnowledgeOf`Share Knowledge of(1)

Chat Participantsex:chat-participants

`showsEqualityBetween`Shows Equality Between(1)

Language Modeling Metricex:language-modeling-metric

`synchronizeAtDifferentRatesInHelmholtz`Synchronize at Different Rates in Helmholtz(1)

Groupsex:groups

Other facts (62)

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.

62 facts

Predicate	Value	Ref
Rdf:type	Ffn Type	[8]
Rdf:type	Model Component	[9]
Rdf:type	Model	[11]
Has Tokens Per Second	51000	[4]
Has Tokens Per Second	23200	[10]
Has Lower Capacity Than	V3	[5]
Has Lower Capacity Than	V3	[10]
Has Speed	28-39K tok/s	[1]
Significantly Faster	Baseline	[1]
Is Significantly Faster Than	Baseline	[1]
Referenced in Comparison	Rotor Pattern	[2]
Part of Current Implementation	True	[3]
Uses Scalar Decay	True	[3]
Has Exponential Memory Fade	True	[3]
Has Total Params Approx	8.5M	[4]
Is Faster Than	Gelu Mlp Baseline	[4]
Has Bpb At500	2.547	[4]
Abbreviated As	H	[5]
Exhibits Smoothing Effect	null	[5]
Has Weaker Snr Than	V3	[5]
Matches Exactly on	Language Modeling	[5]
Has Mean Spectral Entropy	1.967	[6]
Has Higher Cascade Index	V3	[6]
Has Higher Spectral Entropy	V3	[6]
Has Higher Spectral Entropy in	5 of 6 layers	[6]
Has Higher Spectral Entropy Than	V3	[6]
Trained for	500	[6]
Has Lower Spectral Entropy in	Layer 0	[6]
Has Mean Align Delta	0.318	[6]
Has Mean Cascade Index	0.007	[6]
Has Fiber Plus Sync Pipeline	Fiber Sync Pipeline	[6]
Has More Variance in Per Group Coherence	true	[6]
Has No Dominant Layer for Coherence	true	[6]
Has R Global at Layer5	0.344	[6]
Has Cascade Index2 To3 Times Higher Than	V3	[6]
Is Architecture	Helmholtz	[6]
Experiences More Disruption From Readout After Fiber	true	[6]
Exhibits Sharper Drop in R Global at Layer5 Than V3	V3	[6]
Distributes Energy Broadly	true	[6]
Preferred for Broad Energy Distribution	implied	[6]
Desynchronizes in Final Layer	Layer 5	[6]
Commits to Fiber Based Encoding	Broad Distribution	[6]
Spreads Coherence Uniformly	true	[6]
Has Coherence at	Layer 2	[6]
Has Coherence	0.714	[6]
Has Fiber Component	Fiber	[6]
Has Flatter Coherence Profile	true	[6]
Has Metric	speed	[7]
Has Speed Metric	tokens-per-second	[7]
Uses Operation	Multi Head Cumsum	[7]
Bpb At500	2.547	[8]
Tokens Per Second	51K	[8]
Params Total Model	~8.5M	[8]
Part of	Dynamics Stack	[9]
Has Bpb at Step950	2.257	[10]
Has Snr in Db	-6.4	[10]
Has Capacity in Bits	2.4	[10]
Sweaker Than in Snr	V3	[10]
Has Validation Ppl	4.2	[10]
Has Final Bpb	2.06	[10]
Has Snr Advantage Disadvantage	Lower Snr by 1.1 Db	[10]
Has Spectral Entropy	Higher Spectral Entropy	[10]

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.

66 undated

—

hasSpeedblah/watt-activation/part-383

28-39K tok/s

—

significantlyFasterblah/watt-activation/part-383

ex:baseline

—

isSignificantlyFasterThanblah/watt-activation/part-383

ex:baseline

—

referencedInComparisonblah/watt-activation/part-393

ex:rotor-pattern

—

partOfCurrentImplementationblah/watt-activation/part-391

ex:true

—

usesScalarDecayblah/watt-activation/part-391

ex:true

—

hasExponentialMemoryFadeblah/watt-activation/part-391

ex:true

—

hasTotalParamsApproxblah/watt-activation/part-386

8.5M

—

hasTokensPerSecondblah/watt-activation/part-386

51000

—

isFasterThanblah/watt-activation/part-386

ex:gelu-mlp-baseline

—

hasBpbAt500blah/watt-activation/part-386

2.547

—

abbreviatedAsblah/watt-activation/part-451

ex:h

—

exhibitsSmoothingEffectblah/watt-activation/part-451

null

—

hasLowerCapacityThanblah/watt-activation/part-451

ex:v3

—

hasWeakerSnrThanblah/watt-activation/part-451

ex:v3

—

matchesExactlyOnblah/watt-activation/part-451

ex:language-modeling

—

hasMeanSpectralEntropyblah/watt-activation/part-452

1.967

—

hasHigherCascadeIndexblah/watt-activation/part-452

ex:v3

—

hasHigherSpectralEntropyblah/watt-activation/part-452

ex:v3

—

hasHigherSpectralEntropyInblah/watt-activation/part-452

5 of 6 layers

—

hasHigherSpectralEntropyThanblah/watt-activation/part-452

ex:v3

—

trainedForblah/watt-activation/part-452

500

—

hasLowerSpectralEntropyInblah/watt-activation/part-452

ex:layer-0

—

hasMeanAlignDeltablah/watt-activation/part-452

0.318

—

hasMeanCascadeIndexblah/watt-activation/part-452

0.007

—

hasFiberPlusSyncPipelineblah/watt-activation/part-452

ex:fiber-sync-pipeline

—

hasMoreVarianceInPerGroupCoherenceblah/watt-activation/part-452

true

—

hasNoDominantLayerForCoherenceblah/watt-activation/part-452

true

—

hasRGlobalAtLayer5blah/watt-activation/part-452

0.344

—

hasCascadeIndex2To3TimesHigherThanblah/watt-activation/part-452

ex:v3

—

isArchitectureblah/watt-activation/part-452

Helmholtz

—

experiencesMoreDisruptionFromReadoutAfterFiberblah/watt-activation/part-452

true

—

exhibitsSharperDropInRGlobalAtLayer5ThanV3blah/watt-activation/part-452

ex:v3

—

distributesEnergyBroadlyblah/watt-activation/part-452

true

—

preferredForBroadEnergyDistributionblah/watt-activation/part-452

implied

—

desynchronizesInFinalLayerblah/watt-activation/part-452

ex:layer-5

—

commitsToFiberBasedEncodingblah/watt-activation/part-452

ex:broad-distribution

—

spreadsCoherenceUniformlyblah/watt-activation/part-452

true

—

hasCoherenceAtblah/watt-activation/part-452

ex:layer-2

—

hasCoherenceblah/watt-activation/part-452

0.714

—

hasFiberComponentblah/watt-activation/part-452

ex:fiber

—

hasFlatterCoherenceProfileblah/watt-activation/part-452

true

—

labelblah/watt-activation/381

Helmholtz

—

hasMetricblah/watt-activation/381

speed

—

hasSpeedMetricblah/watt-activation/381

tokens-per-second

—

usesOperationblah/watt-activation/381

ex:multi-head-cumsum

—

typeblah/watt-activation/384

ex:FFNType

—

labelblah/watt-activation/384

helmholtz

—

bpbAt500blah/watt-activation/384

2.547

—

tokensPerSecondblah/watt-activation/384

51K

—

paramsTotalModelblah/watt-activation/384

~8.5M

—

typeblah/watt-activation/383

ex:ModelComponent

—

labelblah/watt-activation/383

Helmholtz

—

partOfblah/watt-activation/383

ex:dynamics-stack

—

labelblah/watt-activation/449

Helmholtz

—

hasBpbAtStep950blah/watt-activation/449

2.257

—

hasSnrInDbblah/watt-activation/449

-6.4

—

hasCapacityInBitsblah/watt-activation/449

2.4

—

hasTokensPerSecondblah/watt-activation/449

23200

—

sweakerThanInSnrblah/watt-activation/449

ex:v3

—

hasLowerCapacityThanblah/watt-activation/449

ex:v3

—

hasValidationPplblah/watt-activation/449

4.2

—

hasFinalBpbblah/watt-activation/449

2.06

—

hasSnrAdvantageDisadvantageblah/watt-activation/449

ex:lower-snr-by-1.1-db

—

hasSpectralEntropyblah/watt-activation/449

ex:higher-spectral-entropy

—

typeblah/watt-activation/453

ex:Model

References (11)

11 references

[1]Part 3833 facts
ctx:discord/blah/watt-activation/part-383
[2]Part 3931 fact
ctx:discord/blah/watt-activation/part-393
[3]Part 3913 facts
ctx:discord/blah/watt-activation/part-391
[4]Part 3864 facts
ctx:discord/blah/watt-activation/part-386
[5]Part 4515 facts
ctx:discord/blah/watt-activation/part-451
[6]Part 45226 facts
ctx:discord/blah/watt-activation/part-452
[7]3814 facts
ctx:discord/blah/watt-activation/381
- full textwatt-activation-381
  text/plain3 KBdoc:agent/watt-activation-381/184349fd-7bc1-4a89-ae08-89b469117e85
  Show excerpt
  [2026-03-19 00:15] xenonfun: on occasion but generally session ahs been running like 16 hrs [2026-03-19 00:16] lisamegawatts: mine is always trying to weasel out of work i feel like [2026-03-19 00:45] lisamegawatts: ## Executive Summary Th…
[8]3845 facts
ctx:discord/blah/watt-activation/384
- full textwatt-activation-384
  text/plain3 KBdoc:agent/watt-activation-384/9e57a6f6-7f88-4fd5-81f0-454435de5e3e
  Show excerpt
  [2026-03-19 01:53] xenonfun: ⏺ Pure FFN comparison — clean, no contention, sequential: ``` ┌─────────────────────┬──────────┬───────┬──────────────────────┐ │ FFN Type │ BPB @500 │ tok/s │ Params (total model) │ ├──────────…
[9]3833 facts
ctx:discord/blah/watt-activation/383
[10]44911 facts
ctx:discord/blah/watt-activation/449
- full textwatt-activation-449
  text/plain2 KBdoc:agent/watt-activation-449/f6da4410-b0fa-46d2-9cb5-99d205dd2e55
  Show excerpt
  [2026-03-21 03:48] xenonfun: ⏺ You're right. At step 950: ``` ┌──────────┬───────────┬──────────┐ │ Metric │ Helmholtz │ v3 │ ├──────────┼───────────┼──────────┤ │ BPB │ 2.257 │ 2.257 │ ├──────────┼───────────┼…
[11]4531 fact
ctx:discord/blah/watt-activation/453

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.

Helmholtz

Inbound mentions (31)

hasBlockHas Block(3)

characterizesCharacterizes(2)

appliedToApplied to(1)

closesBpbGapBetweenCloses Bpb Gap Between(1)

comparesFFNTypeCompares Ffn Type(1)

comparesPerLayerCompares Per Layer(1)

concerns500StepTrainingConcerns500 Step Training(1)

consistsOfConsists of(1)

contrastsWithContrasts With(1)

exhibitsSamePatternAsExhibits Same Pattern As(1)

hasBpbValueForHas Bpb Value for(1)

hasCapacityValueForHas Capacity Value for(1)

hasConsistentlyHigherAlignDeltaThanHas Consistently Higher Align Delta Than(1)

hasDifferentEncodingStrategyFromHas Different Encoding Strategy From(1)

hasHigherThroughputThanHas Higher Throughput Than(1)

hasLowerSpectralEntropyHas Lower Spectral Entropy(1)

hasSameInformationContentAsHas Same Information Content As(1)

hasSnrValueForHas Snr Value for(1)

hasThroughputValueForHas Throughput Value for(1)

improvesOverImproves Over(1)

includesIncludes(1)

includesHigherSpectralEntropyIncludes Higher Spectral Entropy(1)

includesLowerSnrIncludes Lower Snr(1)

involvesModelInvolves Model(1)

mentionsEntityMentions Entity(1)

shareKnowledgeOfShare Knowledge of(1)

showsEqualityBetweenShows Equality Between(1)

synchronizeAtDifferentRatesInHelmholtzSynchronize at Different Rates in Helmholtz(1)

Other facts (62)

Timeline

References (11)

See also

Keep researching

`hasBlock`Has Block(3)

`characterizes`Characterizes(2)

`appliedTo`Applied to(1)

`closesBpbGapBetween`Closes Bpb Gap Between(1)

`comparesFFNType`Compares Ffn Type(1)

`comparesPerLayer`Compares Per Layer(1)

`concerns500StepTraining`Concerns500 Step Training(1)

`consistsOf`Consists of(1)

`contrastsWith`Contrasts With(1)

`exhibitsSamePatternAs`Exhibits Same Pattern As(1)

`hasBpbValueFor`Has Bpb Value for(1)

`hasCapacityValueFor`Has Capacity Value for(1)

`hasConsistentlyHigherAlignDeltaThan`Has Consistently Higher Align Delta Than(1)

`hasDifferentEncodingStrategyFrom`Has Different Encoding Strategy From(1)

`hasHigherThroughputThan`Has Higher Throughput Than(1)

`hasLowerSpectralEntropy`Has Lower Spectral Entropy(1)

`hasSameInformationContentAs`Has Same Information Content As(1)

`hasSnrValueFor`Has Snr Value for(1)

`hasThroughputValueFor`Has Throughput Value for(1)

`improvesOver`Improves Over(1)

`includes`Includes(1)

`includesHigherSpectralEntropy`Includes Higher Spectral Entropy(1)

`includesLowerSnr`Includes Lower Snr(1)

`involvesModel`Involves Model(1)

`mentionsEntity`Mentions Entity(1)

`shareKnowledgeOf`Share Knowledge of(1)

`showsEqualityBetween`Shows Equality Between(1)

`synchronizeAtDifferentRatesInHelmholtz`Synchronize at Different Rates in Helmholtz(1)