Symbiogenesis

as donto believed on

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

Symbiogenesis has 207 facts recorded in Dontopedia across 38 references, with 13 live disagreements.

207 facts·178 predicates·38 sources·13 in dispute

Mostly:rdf:type(6), has per seed margins vs fedprox ft(5), analogous to(3)

Inbound mentions (51)

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.

`compares`Compares(4)

50 Client Final Tableex:50-client-final-table
Experiment 1ex:experiment-1
Message 2337ex:message-2337
Seed Experimentex:seed-experiment

`containsTerm`Contains Term(3)

Search 1ex:search-1
Search 2ex:search-2
Search Query 1ex:search-query-1

`inspiredBy`Inspired by(2)

Lisa Megawatts Symbio Gpt 10mex:lisa-megawatts-symbio-gpt-10m
Lisa Megawatts Symbio Slmex:lisa-megawatts-symbio-slm

`usedBy`Used by(2)

Metal Backend Rsex:metal-backend-rs
Metal Sgemm Kernelsex:metal-sgemm-kernels

`advocatesFor`Advocates for(1)

Lisamegawattsex:lisamegawatts

`affiliatedWithTeamDeveloping`Affiliated With Team Developing(1)

Lisamegawattsex:lisamegawatts

`appliesToModule`Applies to Module(1)

Metal Gpu Featureex:metal-gpu-feature

`appliesToModules`Applies to Modules(1)

Metal Gpu Featureex:metal-gpu-feature

`attemptedInformationRetrieval`Attempted Information Retrieval(1)

Omega Botex:omega-bot

`behindFeatureFlag`Behind Feature Flag(1)

Training Decisionex:training-decision

`canJoinPopulationAtAnyTimeIn`Can Join Population at Any Time in(1)

New Clientsex:new-clients

`claims`Claims(1)

Message 2026 03 20 06 16ex:message-2026-03-20-06-16

`comparesToConcept`Compares to Concept(1)

Content 1ex:content-1

`conductsReplicationStudy`Conducts Replication Study(1)

Xenonfunex:xenonfun

`confirmsInsight`Confirms Insight(1)

Xenonfunex:xenonfun

`cooccursWith`Cooccurs With(1)

Fedsymex:fedsym

`decouples`Decouples(1)

Harmonic Core Cargo Tomlex:harmonic-core-cargo-toml

`derivedFromLibrary`Derived From Library(1)

Fedsym Experimentex:fedsym-experiment

`developing`Developing(1)

Lisa Wattsex:lisa-watts

`emphasizesInsight`Emphasizes Insight(1)

Xenonfunex:xenonfun

`firstExplainedInDarwinianTerms`First Explained in Darwinian Terms(1)

Boris Kozo Polyanskyex:boris-kozo-polyansky

`focusesOnTopic`Focuses on Topic(1)

Xenonfunex:xenonfun

`hasResearchFocusOn`Has Research Focus on(1)

Lisa Wattsex:lisa-watts

`implicatesEvolutionaryProcess`Implicates Evolutionary Process(1)

Experimentsex:experiments

`include`Include(1)

Lisa Naming Choicesex:lisa-naming-choices

`includes`Includes(1)

Test Config Core Metal Symbiogenesis Resonateex:test-config-core-metal-symbiogenesis-resonate

`involvesDecouplingOf`Involves Decoupling of(1)

Action 3ex:action-3

`isFarSlowerThan`Is Far Slower Than(1)

Finetuneex:finetune

`isGenuinely`Is Genuinely(1)

Proposed Approachex:proposed-approach

`isInspiredBy`Is Inspired by(1)

Lisa Watts AI Researchex:lisa-watts-ai-research

`isOperationalDomainOf`Is Operational Domain of(1)

Weight Spaceex:weight-space

`isRepositoryFor`Is Repository for(1)

Symbiogenesis Githubex:symbiogenesis-github

`isSpeedInBetween`Is Speed in Between(1)

Finetuneex:finetune

`isSuitableForScaling`Is Suitable for Scaling(1)

Mlxex:mlx

`isUsedIn`Is Used in(1)

Boltzmann Selectionex:boltzmann-selection

`makesPitch`Makes Pitch(1)

Lisamegawattsex:lisamegawatts

`neverSeeEachOthersDataIn`Never See Each Others Data in(1)

Clientsex:clients

`neverSeesRawDataIn`Never Sees Raw Data in(1)

Serverex:server

`presupposesExistenceOf`Presupposes Existence of(1)

Textex:text

`presupposesSymbiogenesisViability`Presupposes Symbiogenesis Viability(1)

Lisamegawattsex:lisamegawatts

`readyNow`Ready Now(1)

Xor Parameter Efficiencyex:xor-parameter-efficiency

`slowerThan`Slower Than(1)

Finetuneex:finetune

`structure`Structure(1)

Phasesex:phases

`usesAsMetaphor`Uses As Metaphor(1)

Lisa Wattsex:lisa-watts

Other facts (199)

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.

199 facts

Predicate	Value	Ref
Rdf:type	Algorithm	[30]
Rdf:type	Algorithm	[31]
Rdf:type	Search Term	[33]
Rdf:type	Library	[34]
Rdf:type	Dependency	[37]
Rdf:type	Software Module	[38]
Has Per Seed Margins Vs Fedprox Ft	+1.2pp	[25]
Has Per Seed Margins Vs Fedprox Ft	+2.1pp	[25]
Has Per Seed Margins Vs Fedprox Ft	+1.1pp	[25]
Has Per Seed Margins Vs Fedprox Ft	+2.7pp	[25]
Has Per Seed Margins Vs Fedprox Ft	+1.8pp	[25]
Analogous to	Continual Learning in AI	[2]
Analogous to	Mitochondria	[9]
Analogous to	Mitochondria	[27]
Avoids	multi-round synchronization	[25]
Avoids	waiting for stragglers	[25]
Avoids	proximal term requiring the global model broadcast back	[25]
Contrasts With	Gradual Mutations and Individual Competition	[3]
Contrasts With	Standard Methods	[9]
Has Alternative Name	Serial Endosymbiotic Theory	[3]
Has Alternative Name	Endosymbiotic Theory	[3]
Outperforms	Neat Style Nas Baseline	[13]
Outperforms	Tensorneat Hybrid	[23]
Avoids Need to	restart training	[25]
Avoids Need to	synchronize rounds	[25]
Benefits From	Client Diversity	[25]
Benefits From	Diversity	[31]
Exceeds Capabilities of	Fedprox	[25]
Exceeds Capabilities of	Scaffold	[25]
Faster Than	Neat in Unit	[29]
Faster Than	Neat Algorithm	[30]
Has Comparison Advantage Over	Fedprox	[31]
Has Comparison Advantage Over	Scaffold	[31]
In Neural Networks	Neural Networks	[1]
Commits to	Bio Inspired AI Ontology	[1]
Has Git Hub	github.com/MonumentalSystems/Symbiogenesis	[1]
Explores	Independent Neural Modules Merging	[1]
Addresses	Catastrophic Forgetting	[1]
Is Novel Framework for	Continual Learning	[1]
Is Novel	true	[1]
Preserves Previously Learned Capabilities	Co Adapting Neural Modules	[1]
Presupposes	Biological Symbiogenesis Exists	[1]
Inspired by	Biological Symbiogenesis	[1]
Is Evaluated As	novel metaphor	[2]
Is Novel Metaphor for	Solving Catastrophic Interference	[2]
Embodies Teleological Preservation	Function	[2]
Is	Continual Learning	[2]
Provides Biological Inspiration for	Evolving AI Project	[2]
Involves	Merger of Organisms	[2]
Presupposes Existence of	Endosymbiosis Events	[3]
Superior to	Natural Selection Alone	[3]
Draws From Historical Context of	Early 20th Century Biology	[3]
Demonstrates That	Major Evolutionary Advancements From Symbiotic Mergers	[3]
Inspires Continual Learning Solution	Lisa Watts AI Research	[3]
Is Possible for	Other Organelles	[3]
Is Evaluated As Leading	true	[3]
Positively Evaluated by	Text	[3]
Is Leading Evolutionary Theory of	Origin of Eukaryotic Cells From Prokaryotic Organisms	[3]
Ontologically Commits to	Symbiosis As Evolutionary Driver	[3]
Axiological Value of	Cooperation Over Competition	[3]
Is Metaphor for	Continual Learning in AI	[3]
Models Biological Process	symbiogenesis	[4]
Commits to Evolutionary Paradigm	null	[5]
Is Ongoing Process	null	[6]
Achieves Convergence in Less Time	true	[7]
Uses Fusion in Distillation	true	[7]
Uses Dynamic Lr	true	[7]
Biologically Inspired	Endosymbiosis	[7]
Lacks Optimizations	true	[7]
Achieves Convergence in Less Compute	true	[7]
Never Overfits Due to	Gelation Signal	[7]
Performs Better Wide Than Deep	true	[7]
Superior to Standard Training on Some Tasks	true	[7]
Close Replication on Cifar10	~CLOSE	[7]
Is Nas Method	true	[7]
Achieves Convergence in Less Params	true	[7]
Replicates Paper on Mnist	YES	[7]
Uses Fusion in Moe	true	[7]
Related to	Symbio	[8]
Has Branch	Metalverificationreplication	[8]
Connects Tasks at	Boundary	[9]
Implies Modular Architecture	Structural Territory	[9]
Teleologically Separates Tasks	Task	[9]
Is Key Insight	confirmed	[9]
Protects Knowledge by Separation	Old Knowledge	[9]
Gives to Each Task	Structural Territory	[9]
Provides Ga Results	Per Layer Lr Multipliers	[10]
Compared to	Symbiogpt	[11]
Superior in Speed	Neat in Unit	[12]
Has Mean Wall Time Per Seed	2.7s	[12]
Is Much Faster Per Seed Than	Neat in Unit	[12]
Completes Tasks in Time	1-2 seconds of log time	[12]
Has Efficiency Advantage	Neat in Unit	[12]
Is Slightly Ahead on	Final Combined Accuracy	[12]
Is Faster by Factor	357x	[12]
Processes Num Tasks	5	[12]
Starts Condition	Fast Phase	[12]
Is Fusion Based Evolutionary Search	null	[13]
Is Often Stronger Than Differentiable Nas Baselines	null	[13]
Used Less Wall Clock Time on Average Than	Tensorneat Hybrid Search	[13]

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.

207 undated

—

inNeuralNetworkslisa-watts/resume-professional-profile

ex:neural-networks

—

commitsTolisa-watts/resume-professional-profile

ex:bio-inspired-ai-ontology

—

hasGitHublisa-watts/resume-professional-profile

github.com/MonumentalSystems/Symbiogenesis

—

exploreslisa-watts/resume-professional-profile

ex:independent-neural-modules-merging

—

addresseslisa-watts/resume-professional-profile

ex:catastrophic-forgetting

—

isNovelFrameworkForlisa-watts/resume-professional-profile

ex:continual-learning

—

isNovellisa-watts/resume-professional-profile

true

—

preservesPreviouslyLearnedCapabilitieslisa-watts/resume-professional-profile

ex:co-adapting-neural-modules

—

presupposeslisa-watts/resume-professional-profile

ex:biological-symbiogenesis-exists

—

inspiredBylisa-watts/resume-professional-profile

ex:biological-symbiogenesis

—

isEvaluatedAslisa-watts/research-catastrophic-forgetting

novel metaphor

—

isNovelMetaphorForlisa-watts/research-catastrophic-forgetting

ex:solving-catastrophic-interference

—

embodiesTeleologicalPreservationlisa-watts/research-catastrophic-forgetting

ex:function

—

islisa-watts/research-catastrophic-forgetting

ex:continual-learning

—

providesBiologicalInspirationForlisa-watts/research-catastrophic-forgetting

ex:evolving-ai-project

—

involveslisa-watts/research-catastrophic-forgetting

ex:merger-of-organisms

—

analogousTolisa-watts/research-catastrophic-forgetting

ex:continual-learning-in-ai

—

presupposesExistenceOflisa-watts/research-symbiogenesis

ex:endosymbiosis-events

—

superiorTolisa-watts/research-symbiogenesis

ex:natural-selection-alone

—

drawsFromHistoricalContextOflisa-watts/research-symbiogenesis

ex:early-20th-century-biology

—

contrastsWithlisa-watts/research-symbiogenesis

ex:gradual-mutations-and-individual-competition

—

demonstratesThatlisa-watts/research-symbiogenesis

ex:major-evolutionary-advancements-from-symbiotic-mergers

—

inspiresContinualLearningSolutionlisa-watts/research-symbiogenesis

ex:lisa-watts-ai-research

—

isPossibleForlisa-watts/research-symbiogenesis

ex:other-organelles

—

isEvaluatedAsLeadinglisa-watts/research-symbiogenesis

true

—

hasAlternativeNamelisa-watts/research-symbiogenesis

ex:serial-endosymbiotic-theory

—

hasAlternativeNamelisa-watts/research-symbiogenesis

ex:endosymbiotic-theory

—

positivelyEvaluatedBylisa-watts/research-symbiogenesis

ex:text

—

isLeadingEvolutionaryTheoryOflisa-watts/research-symbiogenesis

ex:origin-of-eukaryotic-cells-from-prokaryotic-organisms

—

ontologicallyCommitsTolisa-watts/research-symbiogenesis

ex:symbiosis-as-evolutionary-driver

—

axiologicalValueOflisa-watts/research-symbiogenesis

ex:cooperation-over-competition

—

isMetaphorForlisa-watts/research-symbiogenesis

ex:continual-learning-in-ai

—

modelsBiologicalProcessblah/training-and-evals/part-16

symbiogenesis

—

commitsToEvolutionaryParadigmblah/training-and-evals/part-20

null

—

isOngoingProcessblah/training-and-evals/part-22

null

—

achievesConvergenceInLessTimeblah/watt-activation/part-1

true

—

usesFusionInDistillationblah/watt-activation/part-1

true

—

usesDynamicLrblah/watt-activation/part-1

true

—

biologicallyInspiredblah/watt-activation/part-1

ex:endosymbiosis

—

lacksOptimizationsblah/watt-activation/part-1

true

—

achievesConvergenceInLessComputeblah/watt-activation/part-1

true

—

neverOverfitsDueToblah/watt-activation/part-1

ex:gelation-signal

—

performsBetterWideThanDeepblah/watt-activation/part-1

true

—

superiorToStandardTrainingOnSomeTasksblah/watt-activation/part-1

true

—

closeReplicationOnCifar10blah/watt-activation/part-1

~CLOSE

—

isNasMethodblah/watt-activation/part-1

true

—

achievesConvergenceInLessParamsblah/watt-activation/part-1

true

—

replicatesPaperOnMnistblah/watt-activation/part-1

YES

—

usesFusionInMoeblah/watt-activation/part-1

true

—

relatedToblah/watt-activation/part-2

ex:symbio

—

hasBranchblah/watt-activation/part-2

ex:metalverificationreplication

—

connectsTasksAtblah/watt-activation/part-12

ex:boundary

—

impliesModularArchitectureblah/watt-activation/part-12

ex:structural-territory

—

teleologicallySeparatesTasksblah/watt-activation/part-12

ex:task

—

isKeyInsightblah/watt-activation/part-12

confirmed

—

analogousToblah/watt-activation/part-12

ex:mitochondria

—

protectsKnowledgeBySeparationblah/watt-activation/part-12

ex:old-knowledge

—

givesToEachTaskblah/watt-activation/part-12

ex:structural-territory

—

contrastsWithblah/watt-activation/part-12

ex:standard-methods

—

providesGaResultsblah/watt-activation/part-34

ex:per-layer-lr-multipliers

—

comparedToblah/watt-activation/part-33

ex:symbiogpt

—

superiorInSpeedblah/watt-activation/part-308

ex:neat-in-unit

—

hasMeanWallTimePerSeedblah/watt-activation/part-308

2.7s

—

isMuchFasterPerSeedThanblah/watt-activation/part-308

ex:neat-in-unit

—

completesTasksInTimeblah/watt-activation/part-308

1-2 seconds of log time

—

hasEfficiencyAdvantageblah/watt-activation/part-308

ex:neat-in-unit

—

isSlightlyAheadOnblah/watt-activation/part-308

ex:final-combined-accuracy

—

isFasterByFactorblah/watt-activation/part-308

357x

—

processesNumTasksblah/watt-activation/part-308

—

startsConditionblah/watt-activation/part-308

ex:fast-phase

—

isFusionBasedEvolutionarySearchblah/watt-activation/part-307

null

—

isOftenStrongerThanDifferentiableNasBaselinesblah/watt-activation/part-307

null

—

usedLessWallClockTimeOnAverageThanblah/watt-activation/part-307

ex:tensorneat-hybrid-search

—

outperformedDartsByOnMnistblah/watt-activation/part-307

2.8

—

outperformsblah/watt-activation/part-307

ex:neat-style-nas-baseline

—

commitsToOntologicalSuperiorityOverblah/watt-activation/part-307

ex:neat

—

superiorInWallClockTimeblah/watt-activation/part-307

null

—

consistentlyOutperformedblah/watt-activation/part-307

ex:standard-baselines

—

hasAverageTimeInSecondsblah/watt-activation/part-307

34.3

—

wonAllFiveSeedsAgainstblah/watt-activation/part-307

ex:tensorneat-hybrid-search

—

hasMeanTestGapOverNeatblah/watt-activation/part-307

0.0990

—

winsAll5SeedsAgainstblah/watt-activation/part-307

ex:neat

—

achievedTestAccuracyOnCifar10blah/watt-activation/part-307

34.66 ± 2.32 %

—

beatsblah/watt-activation/part-307

ex:darts

—

isCompetitiveWithNeuroevolutionBaselinesblah/watt-activation/part-307

null

—

isFasterOnAverageThanblah/watt-activation/part-307

ex:neat

—

isProjectNameblah/watt-activation/part-347

ex:wandb

—

cooccursWithFusionblah/watt-activation/part-441

null

—

drawsFromBiologyblah/watt-activation/part-461

horizontal gene transfer

—

canRunNowblah/watt-activation/part-466

ex:cnn-benchmark

—

readyForCnnBenchmarkblah/watt-activation/part-466

now

—

partOfReplicationblah/watt-activation/part-466

ex:cnn-benchmark

—

isIntertextualSubjectblah/watt-activation/part-469

ex:original-paper

—

involvesGelationblah/watt-activation/part-469

null

—

beingPortedToblah/watt-activation/part-488

ex:rust

—

implementedInblah/watt-activation/part-488

ex:python

—

involvesEvolutionblah/watt-activation/part-488

true

—

integratesCliffordModulesblah/watt-activation/part-507

true

—

postAdditionTestsblah/watt-activation/part-507

916

—

hadPriorTestsblah/watt-activation/part-507

872

—

rustProjectblah/watt-activation/part-507

true

—

previouslyExistedblah/watt-activation/part-507

true

—

wasPreviouslyCoupledblah/watt-activation/part-537

null

—

hasStandardVersionblah/watt-activation/part-584

ex:standard-symbiogenesis

—

priorArtInDomainblah/watt-activation/part-584

ex:chimera-governor

—

achievedScoreblah/watt-activation/part-306

0.537

—

achievedScoreOnSeed123blah/watt-activation/part-306

0.528

—

outperformsblah/watt-activation/part-306

ex:tensorneat-hybrid

—

biologicalTermAppliedblah/watt-activation/part-438

ex:proposed-approach

—

analogousToProposedApproachblah/watt-activation/part-438

ex:proposed-approach

—

involvesOrganismsWithDifferentSpectralCapabilitiesblah/watt-activation/part-438

ex:organisms

—

isStatisticallySuperiorByWilcoxonblah/watt-activation/part-434

ex:null

—

avoidsblah/watt-activation/part-434

multi-round synchronization

—

avoidsblah/watt-activation/part-434

waiting for stragglers

—

avoidsblah/watt-activation/part-434

proximal term requiring the global model broadcast back

—

avoidsNeedToblah/watt-activation/part-434

restart training

—

avoidsNeedToblah/watt-activation/part-434

synchronize rounds

—

benefitsFromblah/watt-activation/part-434

ex:client-diversity

—

embracesNonIidDiversityblah/watt-activation/part-434

ex:null

—

enablesClientsToblah/watt-activation/part-434

upload their model once and go offline

—

ensuresZeroDataSharingblah/watt-activation/part-434

ex:null

—

exceedsCapabilitiesOfblah/watt-activation/part-434

ex:fedprox

—

exceedsCapabilitiesOfblah/watt-activation/part-434

ex:scaffold

—

feedsOnblah/watt-activation/part-434

ex:heterogeneity

—

fusesModelsViablah/watt-activation/part-434

block-diagonal + compression

—

hasBiggestStrengthblah/watt-activation/part-434

diverse population for evolutionary search

—

hasMaximumMarginVsFedproxFtblah/watt-activation/part-434

+2.7pp

—

hasMeanAccuracyIn50ClientFinalblah/watt-activation/part-434

87.5%

—

hasMinimumMarginVsFedproxFtblah/watt-activation/part-434

+1.1pp

—