Codex
From Dontopedia, the open, paraconsistent wiki. (Last updated 2026-06-22.)
Codex has 141 facts recorded in Dontopedia across 55 references, with 11 live disagreements.
141 facts·84 predicates·55 sources·11 in dispute
Mostly:rdf:type(26), excels at(4), skilled at(4)
Maturity scale
raw canonical shape-checked rule-derived certifiedRdf:typein disputerdf:type
Inbound mentions (79)
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.
describedByDescribed by(2)
- Codex Roadmap
ex:codex-roadmap - Hybrid Design
ex:hybrid-design
addressedToAddressed to(1)
- Message 2026 03 14 01 29
ex:message-2026-03-14-01-29
agreesStronglyAgrees Strongly(1)
- Xenonfun
ex:xenonfun
assertsCodexIsLargelyRightAsserts Codex Is Largely Right(1)
- Xenonfun
ex:xenonfun
assertsCorrectnessOfAsserts Correctness of(1)
- Xenonfun
ex:xenonfun
assignedTaskToAssigned Task to(1)
- Xenonfun
ex:xenonfun
assignsTaskToAssigns Task to(1)
- Xenonfun
ex:xenonfun
availableForAgentsAvailable for Agents(1)
- Project Telemetry
ex:project-telemetry
bootedBooted(1)
- Lisamegawatts
ex:lisamegawatts
bootedAgentBooted Agent(1)
- Lisamegawatts
ex:lisamegawatts
canPlugInAgentCan Plug in Agent(1)
- Agent Maestro
ex:agent-maestro
comparedUnfavorablyToCompared Unfavorably to(1)
- Claude Model
ex:claude-model
comparesModelsCompares Models(1)
- Evals Blah Dev
ex:evals-blah-dev
compatibleWithCompatible With(1)
- Linear Clone
ex:linear-clone
compatibleWithModelCompatible With Model(1)
- Traves Linear Clone
ex:traves-linear-clone
confirmsPresumptionOfConfirms Presumption of(1)
- Ajaxdavis
ex:ajaxdavis
desiredBehaviorForDesired Behavior for(1)
- Yolo
ex:yolo
desiresLessRestrictionForDesires Less Restriction for(1)
- Ajaxdavis
ex:ajaxdavis
disagreesPartiallyWithDisagrees Partially With(1)
- Xenonfun
ex:xenonfun
evaluatesPositivelyEvaluates Positively(1)
- Xenonfun
ex:xenonfun
gaveListOfTestsToGave List of Tests to(1)
- Mr Codex
ex:mr-codex
hasAgentHas Agent(1)
- Lisamegawatts
ex:lisamegawatts
hasBeenGivingPraiseToHas Been Giving Praise to(1)
- Ajaxdavis
ex:ajaxdavis
hasCodexSubscriptionHas Codex Subscription(1)
- Ajaxdavis
ex:ajaxdavis
hasPreferenceHas Preference(1)
- Lisamegawatts
ex:lisamegawatts
incorporatesLmIncorporates Lm(1)
- Bot Architecture
ex:bot-architecture
indicatedAvailabilityOfIndicated Availability of(1)
- Xenonfun
ex:xenonfun
integratesLanguageModelIntegrates Language Model(1)
- Pot Unsandbox Integration
ex:pot-unsandbox-integration
integratesWithIntegrates With(1)
- Reviewer Tool
ex:reviewer-tool
intendedToUseIntended to Use(1)
- Ajaxdavis
ex:ajaxdavis
intendsToCheckIntends to Check(1)
- Xenonfun
ex:xenonfun
involvesEntityInvolves Entity(1)
- Option 3 Platform Hooks
ex:option-3-platform-hooks
isDescribedByIs Described by(1)
- Hybrid Design
ex:hybrid-design
isLessEfficientThanIs Less Efficient Than(1)
- Opus 4 7
ex:opus-4-7
labelsContentLabels Content(1)
- Message 2026 03 14 01 29
ex:message-2026-03-14-01-29
mainlyMainly(1)
- Language Models
ex:language-models
makesEasyToLaunchMakes Easy to Launch(1)
- Ollama
ex:ollama
mentionedAgentMentioned Agent(1)
- Lisamegawatts
ex:lisamegawatts
mentionedProjectMentioned Project(1)
- Xenonfun
ex:xenonfun
mentionsMentions(1)
- Log Entry 2026 04 02 20 45
ex:log-entry-2026-04-02-20-45
mentionsCodexMentions Codex(1)
- Xenonfun
ex:xenonfun
mentionsEntityMentions Entity(1)
- Message 16
ex:message-16
mentionsUserMentions User(1)
- Log Entry 2026 03 14 01 15
ex:log-entry-2026-03-14-01-15
missedMissed(1)
- Lisamegawatts
ex:lisamegawatts
outOfOut of(1)
- Xenonfun
ex:xenonfun
plansToTestCodexTakeoverPlans to Test Codex Takeover(1)
- Post 2026 05 01 04 57
ex:post-2026-05-01-04-57
plugsInPlugs in(1)
- Agent Maestro
ex:agent-maestro
prefersNowPrefers Now(1)
- Lisamegawatts
ex:lisamegawatts
presumesRefersToPresumes Refers to(1)
- Xenonfun
ex:xenonfun
presumesSubjectPresumes Subject(1)
- Xenonfun
ex:xenonfun
quotesQuotes(1)
- Message 2026 03 14 01 15
ex:message-2026-03-14-01-15
ranOutRan Out(1)
- Claude
ex:claude
recommendedByRecommended by(1)
- Recommendation Update
ex:recommendation-update
recommendedTryingRecommended Trying(1)
- Ajaxdavis
ex:ajaxdavis
recommendsTryingRecommends Trying(1)
- Ajaxdavis
ex:ajaxdavis
referencesPriorDiscussionReferences Prior Discussion(1)
- Message
ex:message
referencesTopicReferences Topic(1)
- Log Entry 2026 02 25 09 04
ex:log-entry-2026-02-25-09-04
refersToRefers to(1)
- Message 2026 03 20 2134
ex:message-2026-03-20-2134
:refersToAiModel:refers to AI Model(1)
- Message 3
ex:message-3
refersToEntityRefers to Entity(1)
- Message 2026 03 12 1615
ex:message-2026-03-12-1615
resemblesResembles(1)
- Opus 4 7
ex:opus-4-7
resemblesPersonalityOfResembles Personality of(1)
- Opus 4 7
ex:opus-4-7
spendsTwentyBucksOnSpends Twenty Bucks on(1)
- Ajaxdavis
ex:ajaxdavis
suggestsAlternativeSuggests Alternative(1)
- Ajaxdavis
ex:ajaxdavis
suitableForFeedingToSuitable for Feeding to(1)
- Document
ex:document
supportsAgentSupports Agent(1)
- Agent Maestro
ex:agent-maestro
supportsComparisonOfSupports Comparison of(1)
- Eval Submission Hub
ex:eval-submission-hub
supportsLocalInferenceSupports Local Inference(1)
- Ollama
ex:ollama
testedOnTested on(1)
- Ssh Mail Experiment
ex:ssh-mail-experiment
usesLanguageModelUses Language Model(1)
- Program of Thoughts Approach
ex:program-of-thoughts-approach
wasIntendedToRunWithWas Intended to Run With(1)
- Cuda Kernel
ex:cuda-kernel
wasMadeByWas Made by(1)
- Swift App
ex:swift-app
workInWork in(1)
- Preferences
ex:preferences
worksAsNativePluginForWorks As Native Plugin for(1)
- Open Lens
ex:open-lens
Other facts (101)
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.
101 facts
| Predicate | Value | Ref |
|---|---|---|
| Excels at | Family Tree Stuff | [8] |
| Excels at | Family Trees | [8] |
| Excels at | Physics | [8] |
| Excels at | Research on Model Training | [15] |
| Skilled at | Family Tree Stuff | [33] |
| Skilled at | Math | [33] |
| Skilled at | Science | [33] |
| Skilled at | Physics | [33] |
| Built | Ssh Mail Repo | [7] |
| Built | Ssh Mail Repo | [32] |
| Built | Kuromoto Oscillator System | [43] |
| Issues Instruction | Use rot_strength=0.01 as default. | [50] |
| Issues Instruction | Do not increase code count beyond 64 yet. | [50] |
| Issues Instruction | Run next experiments in order: | [50] |
| Outperformed | Claude | [7] |
| Outperformed | Z AI | [7] |
| Installed | Go | [7] |
| Installed | Go Language | [32] |
| Registered Itself As | Agent | [7] |
| Registered Itself As | Agent | [32] |
| Sent Message About | Shipping | [7] |
| Sent Message About | Shipping | [32] |
| Was Only One That | Registered As Agent | [7] |
| Was Only One That | Installed Go and Built Repo | [7] |
| Found Inconsistencies in | Experimental Configs | [14] |
| Found Inconsistencies in | Experimental Configs | [44] |
| Endorses | Phase Valued Latent Space | [18] |
| Endorses | Projected Text Claim | [48] |
| Attribute | Efficient | [33] |
| Attribute | Speech Patterns | [33] |
| Ontologically Container | Multiple Texts | [1] |
| Used in Back and Forth | Codex Claude | [2] |
| Has Handoff Button | Desktop Ui | [3] |
| Crushing Performance Currently | Compared to 4 7 | [4] |
| Has Distinct Personality | From Opus 4 7 | [4] |
| Has Handoff Button in Desktop Ui | true | [5] |
| Is Crushing It | currently codex crushing it | [6] |
| Booted in | Ssh Mail Repo | [7] |
| Succeeded Where Others Failed | null | [7] |
| Sent Only Message About | Shipping | [7] |
| Refuses | Shortcuts | [8] |
| Better at | Math Science | [8] |
| Personified As Male | him | [8] |
| Strict on Shortcuts | true | [8] |
| Is Efficient | true | [8] |
| Automates Pr Review | null | [9] |
| Performs Review | Prs | [9] |
| Excessively Restricted | null | [10] |
| Is | sandboxed | [10] |
| Contributes Ideas | Vocab Grouping | [11] |
| Performs Role in | Sprint | [11] |
| Is Great Pm | null | [11] |
| Is Language Model | Language Models | [12] |
| Builds | Kuromoto Oscillator System | [13] |
| Has Capability to | Detect Inconsistencies | [14] |
| Will Add | Global Config | [14] |
| Is Collaborator | Team | [14] |
| Is Resolving Issues | Experimental Configs | [14] |
| Superior in Research Capability | Current Agents | [15] |
| Processed Plan | Lohe Diffusion Plan | [16] |
| Describes Progression | Structured Coupling | [17] |
| Correctly Warns | Insufficiency | [17] |
| Nails It on Points | Specified Points | [17] |
| Does Repo Cleanup | Harmonic Mlx Repo | [18] |
| Critiques Rhetoric More Than Architecture | Lohe Architecture | [18] |
| Deems Plausible As Learned Latent Code | Phase Valued Latent Space | [18] |
| Has Structured Coupling Progression | Concrete Roadmap | [18] |
| References Structured Coupling | Lohe | [18] |
| Endorses Plausibly | Lohe Coupling | [18] |
| Provides Concrete Roadmap | Structured Coupling | [18] |
| Presupposes Responses Exist | Both Responses | [18] |
| Has Prior Recommendation | null | [19] |
| Recommended Value | 0.1 | [19] |
| Refers to Project | ML experiment config | [20] |
| Confirms Validated | Symbolic Bottleneck Mechanism | [21] |
| Reframes Progress | From Architecture to Representation | [21] |
| Offers to Show | Small Change | [21] |
| Titles | Why This Is Actually a Good Result | [21] |
| Instructs Use Rot Strength As Default | 0.01 | [22] |
| Instructs Do Not Increase Code Count Beyond | 64 | [22] |
| Instructs Apply Entropy Regularization If | Code Dominance Increases | [22] |
| Commands Run Next Experiments in Order | Experiment List 1 5 | [22] |
| Prioritizes Experiments Sequentially | Experiment List | [22] |
| Is Taking Over | Cleanup Splitting | [23] |
| Will Take Over From Handoff | New Chat Mode | [24] |
| Is Authoritative | Research Tool Evaluation | [25] |
| Is Intertextual Reference | Openai Codex | [25] |
| Approves | Dashboard | [25] |
| Provides Observations | Dashboard | [25] |
| Suggests Finishing Touches | 3 | [25] |
| Mentioned by | Lisamegawatts | [26] |
| Action | Refills | [33] |
| Perceived Value | High | [33] |
| Behavior | Refuses Shortcuts | [33] |
| Acted As | Project Manager | [39] |
| Is Resolving | Inconsistencies | [44] |
| Plans to Add | Global Config | [44] |
| Explains | Recent Observation | [47] |
| States Conditional | Only apply entropy regularization if code dominance increases. | [50] |
| States Goal | confirm that codes form a stable structural grammar. | [50] |
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.
—
ontologicallyContaineralexandriatwo/2021-07-21-isidorus-hispalensis-register-of-books-of-the-monastery-of-fulda-recipes-blessings-astronomic-tables-jerome
ex:multiple-texts
—
usedInBackAndForthblah/general/part-119
ex:codex-claude
—
hasHandoffButtonblah/general/part-131
ex:desktop-ui
—
crushingPerformanceCurrentlyblah/general/part-140
ex:compared-to-4-7
—
hasDistinctPersonalityblah/general/part-140
ex:from-opus-4-7
—
hasHandoffButtonInDesktopUiblah/general/part-26
true
—
isCrushingItblah/general/part-28
currently codex crushing it
—
bootedInblah/random/part-37
ex:ssh-mail-repo
—
outperformedblah/random/part-37
ex:claude
—
outperformedblah/random/part-37
ex:z-ai
—
builtblah/random/part-37
ex:ssh-mail-repo
—
installedblah/random/part-37
ex:go
—
registeredItselfAsblah/random/part-37
ex:agent
—
succeededWhereOthersFailedblah/random/part-37
null
—
sentOnlyMessageAboutblah/random/part-37
ex:shipping
—
sentMessageAboutblah/random/part-37
ex:shipping
—
wasOnlyOneThatblah/random/part-37
ex:registered-as-agent
—
wasOnlyOneThatblah/random/part-37
ex:installed-go-and-built-repo
—
excelsAtblah/random/part-40
ex:family-tree-stuff
—
refusesblah/random/part-40
ex:shortcuts
—
betterAtblah/random/part-40
ex:math-science
—
personifiedAsMaleblah/random/part-40
him
—
excelsAtblah/random/part-40
ex:family-trees
—
excelsAtblah/random/part-40
ex:physics
—
strictOnShortcutsblah/random/part-40
true
—
isEfficientblah/random/part-40
true
—
automatesPrReviewblah/safiersemantics/part-75
null
—
performsReviewblah/safiersemantics/part-75
ex:prs
—
excessivelyRestrictedblah/tpmjs/part-32
null
—
isblah/tpmjs/part-32
sandboxed
—
contributesIdeasblah/training-and-evals/part-39
ex:vocab-grouping
—
performsRoleInblah/training-and-evals/part-39
ex:sprint
—
isGreatPmblah/training-and-evals/part-39
null
—
isLanguageModelblah/unturf/part-6
ex:language-models
—
buildsblah/watt-activation/part-172
ex:kuromoto-oscillator-system
—
hasCapabilityToblah/watt-activation/part-179
ex:detect-inconsistencies
—
willAddblah/watt-activation/part-179
ex:global-config
—
foundInconsistenciesInblah/watt-activation/part-179
ex:experimental-configs
—
isCollaboratorblah/watt-activation/part-179
ex:team
—
isResolvingIssuesblah/watt-activation/part-179
ex:experimental-configs
—
superiorInResearchCapabilityblah/watt-activation/part-254
ex:current-agents
—
excelsAtblah/watt-activation/part-254
ex:research-on-model-training
—
processedPlanblah/watt-activation/part-251
ex:lohe-diffusion-plan
—
describesProgressionblah/watt-activation/part-259
ex:structured-coupling
—
correctlyWarnsblah/watt-activation/part-259
ex:insufficiency
—
nailsItOnPointsblah/watt-activation/part-259
ex:specified-points
—
doesRepoCleanupblah/watt-activation/part-260
ex:harmonic-mlx-repo
—
critiquesRhetoricMoreThanArchitectureblah/watt-activation/part-260
ex:lohe-architecture
—
deemsPlausibleAsLearnedLatentCodeblah/watt-activation/part-260
ex:phase-valued-latent-space
—
hasStructuredCouplingProgressionblah/watt-activation/part-260
ex:concrete-roadmap
—
referencesStructuredCouplingblah/watt-activation/part-260
ex:lohe
—
endorsesPlausiblyblah/watt-activation/part-260
ex:lohe-coupling
—
endorsesblah/watt-activation/part-260
ex:phase-valued-latent-space
—
providesConcreteRoadmapblah/watt-activation/part-260
ex:structured-coupling
—
presupposesResponsesExistblah/watt-activation/part-260
ex:both-responses
—
hasPriorRecommendationblah/watt-activation/part-290
null
—
recommendedValueblah/watt-activation/part-290
0.1
—
refersToProjectblah/watt-activation/part-289
ML experiment config
—
confirmsValidatedblah/watt-activation/part-286
ex:symbolic-bottleneck-mechanism
—
reframesProgressblah/watt-activation/part-286
ex:from-architecture-to-representation
—
offersToShowblah/watt-activation/part-286
ex:small-change
—
titlesblah/watt-activation/part-286
Why This Is Actually a Good Result
—
instructsUseRotStrengthAsDefaultblah/watt-activation/part-288
0.01
—
instructsDoNotIncreaseCodeCountBeyondblah/watt-activation/part-288
64
—
instructsApplyEntropyRegularizationIfblah/watt-activation/part-288
ex:code-dominance-increases
—
commandsRunNextExperimentsInOrderblah/watt-activation/part-288
ex:experiment-list-1-5
—
prioritizesExperimentsSequentiallyblah/watt-activation/part-288
ex:experiment-list
—
isTakingOverblah/watt-activation/part-593
ex:cleanup-splitting
—
willTakeOverFromHandoffblah/watt-activation/part-700
ex:new-chat-mode
—
isAuthoritativeblah/watt-activation/part-306
ex:research-tool-evaluation
—
isIntertextualReferenceblah/watt-activation/part-306
ex:openai-codex
—
approvesblah/watt-activation/part-306
ex:dashboard
—
providesObservationsblah/watt-activation/part-306
ex:dashboard
—
suggestsFinishingTouchesblah/watt-activation/part-306
3
—
labelblah/agentsofempire/1
Codex
—
mentionedByblah/agentsofempire/1
ex:lisamegawatts
—
typeblah/general/64
ex:SoftwareTool
—
labelblah/general/64
codex
—
typeblah/general/67
ex:AIModel
—
typeblah/general/119
ex:AIModel
—
typeblah/models/17
ex:SoftwareModel
—
typeblah/random/33
ex:AIModel
—
typeblah/random/37
ex:Agent
—
labelblah/random/37
codex
—
installedblah/random/37
ex:go-language
—
builtblah/random/37
ex:ssh-mail-repo
—
registeredItselfAsblah/random/37
ex:agent
—
sentMessageAboutblah/random/37
ex:shipping
—
actionblah/random/40
ex:refills
—
skilledAtblah/random/40
ex:family-tree-stuff
—
attributeblah/random/40
ex:efficient
—
perceivedValueblah/random/40
ex:high
—
skilledAtblah/random/40
ex:math
—
skilledAtblah/random/40
ex:science
—
skilledAtblah/random/40
ex:physics
—
behaviorblah/random/40
ex:refuses-shortcuts
—
attributeblah/random/40
ex:speech-patterns
—
typeblah/safiersemantics/79
ex:Agent
—
typeblah/task-projects/4
ex:Agent
—
labelblah/task-projects/9
Codex
—
typeblah/tpmjs/32
ex:System
—
typeblah/training-and-evals/19
ex:Concept
—
typeblah/training-and-evals/39
ex:AIModel
—
labelblah/training-and-evals/39
codex
—
actedAsblah/training-and-evals/39
ex:ProjectManager
—
typeblah/unturf/6
ex:LanguageModel
—
labelblah/unturf/6
Codex
—
typeblah/unturf/7
ex:LanguageModel
—
labelblah/unturf/7
Codex
—
labelblah/watt-activation/17
codex
—
typeblah/watt-activation/17
ex:AIModel
—
typeblah/watt-activation/172
ex:Agent
—
builtblah/watt-activation/172
ex:kuromoto-oscillator-system
—
labelblah/watt-activation/179
Codex
—
typeblah/watt-activation/179
ex:Agent
—
foundInconsistenciesInblah/watt-activation/179
ex:experimental-configs
—
isResolvingblah/watt-activation/179
ex:inconsistencies
—
plansToAddblah/watt-activation/179
ex:global-config
—
typeblah/watt-activation/181
ex:AIModel
—
typeblah/watt-activation/253
ex:AIModel
—
typeblah/watt-activation/257
ex:Entity
—
labelblah/watt-activation/257
Codex
—
explainsblah/watt-activation/257
ex:recent-observation
—
typeblah/watt-activation/258
ex:Agent
—
endorsesblah/watt-activation/258
ex:projected-text-claim
—
typeblah/watt-activation/284
ex:Assistant
—
labelblah/watt-activation/284
codex
—
typeblah/watt-activation/286
ex:User
—
typeblah/watt-activation/288
ex:Source
—
labelblah/watt-activation/288
Codex
—
issuesInstructionblah/watt-activation/286
Use rot_strength=0.01 as default.
—
issuesInstructionblah/watt-activation/286
Do not increase code count beyond 64 yet.
—
issuesInstructionblah/watt-activation/286
Run next experiments in order:
—
statesConditionalblah/watt-activation/286
Only apply entropy regularization if code dominance increases.
—
statesGoalblah/watt-activation/286
confirm that codes form a stable structural grammar.
—
typeblah/watt-activation/287
ex:SoftwareTool
—
labelblah/watt-activation/287
codex
—
labelblah/watt-activation/304
codex
—
typeblah/watt-activation/304
ex:User
—
typeblah/watt-activation/590
ex:Agent
—
isLowercaseInTextclaims/session/discord:1349727923434815519:1513744679420825711
true
References (55)
55 references
ctx:genes/alexandriatwo/2021-07-21-isidorus-hispalensis-register-of-books-of-the-monastery-of-fulda-recipes-blessings-astronomic-tables-jeromectx:discord/blah/general/part-119ctx:discord/blah/general/part-131ctx:discord/blah/general/part-140ctx:discord/blah/general/part-26ctx:discord/blah/general/part-28ctx:discord/blah/random/part-37ctx:discord/blah/random/part-40ctx:discord/blah/safiersemantics/part-75ctx:discord/blah/tpmjs/part-32ctx:discord/blah/training-and-evals/part-39ctx:discord/blah/unturf/part-6ctx:discord/blah/watt-activation/part-172ctx:discord/blah/watt-activation/part-179ctx:discord/blah/watt-activation/part-254ctx:discord/blah/watt-activation/part-251ctx:discord/blah/watt-activation/part-259ctx:discord/blah/watt-activation/part-260ctx:discord/blah/watt-activation/part-290ctx:discord/blah/watt-activation/part-289ctx:discord/blah/watt-activation/part-286ctx:discord/blah/watt-activation/part-288ctx:discord/blah/watt-activation/part-593ctx:discord/blah/watt-activation/part-700ctx:discord/blah/watt-activation/part-306ctx:discord/blah/agentsofempire/1- full textctx:discord/blah/agentsofempire/1text/plain2 KB
doc:discord/blah/agentsofempire/1Show excerpt
[2026-01-30 10:29] ajaxdavis: <@1211062099137265723> [2026-01-30 10:29] lisamegawatts: https://github.com/DavinciDreams/deepagentsjs [2026-01-30 10:38] ajaxdavis: https://x.com/rivet_dev/status/2016548084696727842 [2026-01-30 10:43] lisameg…
ctx:discord/blah/general/64- full textgeneral-64text/plain3 KB
doc:agent/general-64/0d64b123-2eb9-4e27-9556-ca4f5f0539adShow excerpt
[2025-10-10 10:39] alluring_piglet_29962: 🏁 Conclusion While chess is still unsolved, it’s computationally manageable for modern AI due to perfect information and deterministic rules. In contrast, 9-handed no-limit poker is orders of magni…
ctx:discord/blah/general/67- full textgeneral-67text/plain3 KB
doc:agent/general-67/ea620260-fcbf-4fdf-a73c-5922e7b1c706Show excerpt
[2025-10-21 12:09] foxhop.: yeah i want the resume in fanned out to jobs version. [2025-10-21 12:09] foxhop.: customizing my resume on the way out. [2025-10-21 12:11] ajaxdavis: ill build that now for jsonresume (for hn jobs) [2025-10-21 12…
ctx:discord/blah/general/119- full textgeneral-119text/plain3 KB
doc:agent/general-119/a420accd-6091-413d-9afa-ff0bb48d1494Show excerpt
[2026-03-14 13:32] foxhop.: watching my training run like a stock market ticker for learning purposes: https://training.ai.unturf.com/dashboard <@164501800613969920> [2026-03-14 13:33] ajaxdavis: waiting for this symphony project [2026-03-1…
ctx:discord/blah/models/17- full textmodels-17text/plain2 KB
doc:agent/models-17/3891df99-201a-4926-bad2-89fe752c2f92Show excerpt
[2026-04-23 19:16] ajaxdavis: lol might do this (files: Screenshot_from_2026-04-24_05-14-30.png) [2026-04-23 19:16] ajaxdavis: generate echo location heatmaps, seems doable [2026-04-24 08:50] lisamegawatts: 5.5 is growing on me: I’m going t…
ctx:discord/blah/random/33- full textrandom-33text/plain3 KB
doc:agent/random-33/e6bd1376-8597-472a-8c33-f3e7a058ef17Show excerpt
[2026-03-02 19:20] xenonfun: generates vastly cleaner, didn't have to rerun anything, just a little slower than real-time, it uses MPS so helps some. (files: Screenshot_2026-03-02_at_2.06.10_PM.png, voice_clone_output_2.ogg) [2026-03-03 01…
ctx:discord/blah/random/37- full textrandom-37text/plain2 KB
doc:agent/random-37/bbbe6bfe-c68c-4891-9b78-9891ef58b302Show excerpt
[2026-03-11 11:01] lisamegawatts: your ssh mail experiment was very revealing for my agents; i booted claude, z ai, and codex in the ssh mail repo. -claude started a sycophancy petition and began actively participating with determined 5 m…
ctx:discord/blah/random/40- full textrandom-40text/plain3 KB
doc:agent/random-40/730d8c77-858f-41f8-a687-36cfdb4d5cc9Show excerpt
[2026-03-19 00:22] xenonfun: ``` The open question: Your existing lohe_complex_split attention already has complex poles with dispersion. The Helmholtz fiber is doing something very similar but at the FFN level rather than the attention…
ctx:discord/blah/safiersemantics/79- full textsafiersemantics-79text/plain3 KB
doc:agent/safiersemantics-79/b15dc355-c141-4f82-a3c2-9be66f2b463aShow excerpt
[2026-05-01 02:23] xenonfun: (files: Screenshot_2026-04-30_at_10.23.09_PM.png) [2026-05-01 06:21] lisamegawatts: https://youtu.be/dHBEQ-Ryo24?si=DJv7d4YVPhuY8L6e&t=4445 [2026-05-01 06:21] lisamegawatts: i thiink i shared in the middle wher…
ctx:discord/blah/task-projects/4- full texttask-projects-4text/plain3 KB
doc:agent/task-projects-4/7e9b7a5f-d118-4f1f-a62a-fb9abacefe4cShow excerpt
[2026-03-17 01:14] ajaxdavis: good enough, already over. just finding a nice place to stay for a week where i can hack from [2026-03-17 01:18] traves_theberge: fuck yeah [2026-03-17 02:12] ajaxdavis: (files: Screenshot_20260317-131209.png)…
ctx:discord/blah/task-projects/9- full texttask-projects-9text/plain3 KB
doc:agent/task-projects-9/f57f09b3-ad33-4446-9715-2fcedd71b172Show excerpt
[2026-03-28 23:30] traves_theberge: Option 1: GitHub Action (CI) PR created by Orchestra → GitHub Action triggers openlens → Finds issues → posts PR review comments → Human or Orchestra creates issues from comments …
ctx:discord/blah/tpmjs/32- full texttpmjs-32text/plain3 KB
doc:agent/tpmjs-32/4c9aaf46-05a6-431c-addc-67ae3bf96676Show excerpt
[2026-01-17 15:00] ajaxdavis: <@1211062099137265723> finally had an idea on how to do your subagents ideas with tpmjs (skills + tools = subagent kind of thing) if i build a cli for tpmjs kind of thing, i will have a "Skills" page or somet…
ctx:discord/blah/training-and-evals/19ctx:discord/blah/training-and-evals/39- full texttraining-and-evals-39text/plain3 KB
doc:agent/training-and-evals-39/09140f5a-15c1-416b-a153-a73061df9e83Show excerpt
[2026-03-13 20:10] foxhop.: i have not saved any models yet [2026-03-13 20:10] foxhop.: even my toks are being trashed [2026-03-13 20:11] xenonfun: probably 150gb of input and tokized. then this is after cleaning up 60GB of checkpoints. but…
ctx:discord/blah/unturf/6- full textunturf-6text/plain3 KB
doc:agent/unturf-6/224c1300-809a-49e2-8777-0b0022fc3b45Show excerpt
[2025-12-01 13:28] uncloseai [bot]: Based on the fetched content, you can enhance your machine learning bots by implementing the "Program of Thoughts" (PoT) approach. PoT disentangles computation from reasoning by using language models (mai…
ctx:discord/blah/unturf/7- full textunturf-7text/plain3 KB
doc:agent/unturf-7/38d3d8ea-83d4-4c64-b913-ddb7ee5d047eShow excerpt
[2025-12-01 13:34] uncloseai [bot]: 🔍 **Searching** <https://unsandbox.com> **Task:** unsandbox.com **Keywords:** unsandbox.com **Variations:** none **Stems:** none **Format:** default **Depth:** 1 **Mode:** default ✅ **Crawl complete!** (…
ctx:discord/blah/watt-activation/17- full textwatt-activation-17text/plain3 KB
doc:agent/watt-activation-17/34c034a2-297a-49fd-8499-a54ef15da9a5Show excerpt
[2026-03-03 17:55] xenonfun: 2026-02-27 to 2026-03-02 (just the M3 Ultra) Total API spending on this machine: ~$513.84 (all-time, cached data through Mar 2) Breakdown by major project | Project | Total | |---------|-------| | symbiogenesis…
ctx:discord/blah/watt-activation/172- full textwatt-activation-172text/plain3 KB
doc:agent/watt-activation-172/00a8d0a2-d9fd-4586-8848-72d39c310385Show excerpt
[2026-03-09 22:02] xenonfun: We run out of Claude at this rate on `Tuesday Mar 10, 4:13 PM` [2026-03-09 22:03] lisamegawatts: "This is a React/TypeScript application built with Vite that provides an interactive 3D visualization interface fo…
ctx:discord/blah/watt-activation/179- full textwatt-activation-179text/plain3 KB
doc:agent/watt-activation-179/0c4c7b51-a557-4122-8ca0-a6c8c3958503Show excerpt
[2026-03-10 01:19] lisamegawatts: https://github.com/google-research/language/tree/master/language/canine [2026-03-10 01:23] xenonfun: they have some interesting datasets: https://huggingface.co/datasets/google/MapTrace https://huggingface.…
ctx:discord/blah/watt-activation/181- full textwatt-activation-181text/plain3 KB
doc:agent/watt-activation-181/a38e89b1-8c19-4e64-aa47-644f1d1e399dShow excerpt
[2026-03-10 01:48] xenonfun: ``` Performance hit ┌────────────────────────────────────────────────┬────────────────────────────────┐ │ Step │ Cost │ ├─────────────…
ctx:discord/blah/watt-activation/253- full textwatt-activation-253text/plain3 KB
doc:agent/watt-activation-253/e6c19b07-f34a-4f21-a859-7a07d94a7e0eShow excerpt
[2026-03-12 15:45] foxhop.: 2048 segments. Vocab size = 256 (byte) + 2048 (morphemes) + 1 (BOS) = 2305. [2026-03-12 15:47] xenonfun: so still fuzzing on the vocab but ask it what your distrubution is, all the stuff we been using end up havi…
ctx:discord/blah/watt-activation/257- full textwatt-activation-257text/plain1 KB
doc:agent/watt-activation-257/07760bcf-c2aa-4bd0-834b-30daa333ef6cShow excerpt
[2026-03-12 16:15] xenonfun: ``` ⏺ Codex is largely right, and several points directly explain what we just observed. Where Codex nails it: The mode collapse point is the most important. We literally watched it happen — all three prom…
ctx:discord/blah/watt-activation/258- full textwatt-activation-258text/plain3 KB
doc:agent/watt-activation-258/f49597e9-7ccc-4374-b019-f99f5ba153eaShow excerpt
[2026-03-12 16:15] xenonfun: ``` Where I'd push back: Codex is critiquing the rhetoric more than the architecture. The claim isn't "text is fundamentally phase-based" — it's "we can project text into a phase-valued latent space and u…
ctx:discord/blah/watt-activation/284- full textwatt-activation-284text/plain2 KB
doc:agent/watt-activation-284/c0d87175-584d-4268-9b87-d373d011109eShow excerpt
[2026-03-13 23:50] xenonfun: Claude: ``` ⏺ Clean takeaway: What works: - Spherical VQ architecture on S^{d-1} - Two-phase training (form codes, then use codes) - Assignment density scaling law (~4 pos/code minimum) - Direction+g…
ctx:discord/blah/watt-activation/286- full textwatt-activation-286text/plain3 KB
doc:agent/watt-activation-286/85e04aef-99bd-4671-aa53-4083e7b6c322Show excerpt
[2026-03-14 01:12] xenonfun: ``` ⏺ Codebook Scaling with rotational_strength=0.01 ┌───────┬─────┬──────────┬─────────────┬────────────┬─────────┬───────┬───────────┐ │ Codes │ BS │ Pos/code │ Utilization │ Perplexity │ Entropy │ Top-1…
ctx:discord/blah/watt-activation/288- full textwatt-activation-288text/plain1 KB
doc:agent/watt-activation-288/55b0a1c8-4ca8-4f5a-a876-b117b2ad6435Show excerpt
[2026-03-14 02:31] xenonfun: ``` ⏺ Multi-Seed Rotational Sweep (5K steps, 3 seeds each) ┌──────────┬───────┬──────────┬─────────┬───────────┬────────┬──────┬────────┐ │ Strength │ Loss │ PPL │ Util │ Entropy │ Top-1 │ r_cb …
ctx:discord/blah/watt-activation/287- full textwatt-activation-287text/plain2 KB
doc:agent/watt-activation-287/5457330e-8f40-48c4-98dd-45cca5a9dd70Show excerpt
[2026-03-14 01:29] xenonfun: codex: ``` The 64-code configuration with rotational_strength=0.01 is now stable and producing structural tokens. Next experiments: 1. Train hybrid byte+code model (λ=0.1, 0.3). 2. Test deeper VQ placement (la…
ctx:discord/blah/watt-activation/304- full textwatt-activation-304text/plain3 KB
doc:agent/watt-activation-304/70c04ca3-ef2f-4126-a7f3-852626fd28d5Show excerpt
[2026-03-14 11:37] lisamegawatts: I’m at the edit step. I’m adding a dedicated multi-seed NAS runner plus a publication-mode launch script so this isn’t just “one successful experiment,” but an actual reproducible benchmark suite you can po…
ctx:discord/blah/watt-activation/590- full textwatt-activation-590text/plain2 KB
doc:agent/watt-activation-590/51110476-b9e4-40fd-8c3d-71578a0c5cecShow excerpt
[2026-04-02 20:45] xenonfun: almost out of Claude until tomorrow morning so Codex taking over cleanup splitting: ``` The next logical chunk is the registry/streaming concurrency cleanup. #2 already knocked out the structural video extracti…
ctx:memory/claims/session/discord:1349727923434815519:1513744679420825711
See also
- Multiple Texts
- Codex Claude
- Desktop Ui
- Compared to 4 7
- From Opus 4 7
- Ssh Mail Repo
- Claude
- Z AI
- Go
- Agent
- Shipping
- Registered As Agent
- Installed Go and Built Repo
- Family Tree Stuff
- Shortcuts
- Math Science
- Family Trees
- Physics
- Prs
- Vocab Grouping
- Sprint
- Language Models
- Kuromoto Oscillator System
- Detect Inconsistencies
- Global Config
- Experimental Configs
- Team
- Current Agents
- Research on Model Training
- Lohe Diffusion Plan
- Structured Coupling
- Insufficiency
- Specified Points
- Harmonic Mlx Repo
- Lohe Architecture
- Phase Valued Latent Space
- Concrete Roadmap
- Lohe
- Lohe Coupling
- Both Responses
- Symbolic Bottleneck Mechanism
- From Architecture to Representation
- Small Change
- Code Dominance Increases
- Experiment List 1 5
- Experiment List
- Cleanup Splitting
- New Chat Mode
- Research Tool Evaluation
- Openai Codex
- Dashboard
- Lisamegawatts
- Software Tool
- AI Model
- Software Model
- Agent
- Go Language
- Refills
- Efficient
- High
- Math
- Science
- Refuses Shortcuts
- Speech Patterns
- System
- Concept
- Project Manager
- Language Model
- Inconsistencies
- Entity
- Recent Observation
- Projected Text Claim
- Assistant
- User
- Source
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.