Dontopedia
Explore

Prompt Architecture

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

Prompt Architecture has 21 facts recorded in Dontopedia across 2 references, with 3 live disagreements.

21 facts·14 predicates·2 sources·3 in dispute

Mostly:has feature(4), based on research from(3), implements(3)

Maturity scale raw canonical shape-checked rule-derived certified

Based on Research Fromin disputebasedOnResearchFrom

Implementsin disputeimplements

Has Featurein disputehasFeature

Has ParameterhasParameter

  • maxOutputTokens 1024[1]all time · Part 6

Includesincludes

Causes Improved JudgingcausesImprovedJudging

  • better reasoning and less bias[1]all time · Part 6

Commits to Unbiased EvaluationcommitsToUnbiasedEvaluation

  • null[1]all time · Part 6

Evaluated PositivelyevaluatedPositively

Hedges With FallbackhedgesWithFallback

Known to Work WellknownToWorkWell

Preferred OverpreferredOver

  • shorter token limits[1]all time · Part 6

Presupposes Existence ofpresupposesExistenceOf

Inbound mentions (3)

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.

describesDescribes(1)

mentionsTopicMentions Topic(1)

performsExpositionPerforms Exposition(1)

Other facts (2)

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.

2 facts
PredicateValueRef
Targets AI Judgesnull[1]
Has Parameter ChangeMax Output Tokens[2]

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.

basedOnResearchFromblah/training-and-evals/part-6
ex:alpaca-eval
basedOnResearchFromblah/training-and-evals/part-6
ex:arena-hard
basedOnResearchFromblah/training-and-evals/part-6
ex:mt-bench
causesImprovedJudgingblah/training-and-evals/part-6
better reasoning and less bias
commitsToUnbiasedEvaluationblah/training-and-evals/part-6
null
evaluatedPositivelyblah/training-and-evals/part-6
ex:mt-bench
hasFeatureblah/training-and-evals/6
ex:anti-bias-instructions
hasFeatureblah/training-and-evals/6
ex:chain-of-thought
hasFeatureblah/training-and-evals/6
ex:rubric-first-authority
hasFeatureblah/training-and-evals/6
ex:system-user-separation
hasParameterblah/training-and-evals/part-6
maxOutputTokens 1024
hasParameterChangeblah/training-and-evals/6
ex:max-output-tokens
hedgesWithFallbackblah/training-and-evals/part-6
ex:general-guidelines
implementsblah/training-and-evals/part-6
ex:chain-of-thought-before-scoring
implementsblah/training-and-evals/part-6
ex:rubric-first-authority
implementsblah/training-and-evals/part-6
ex:system-user-message-separation
includesblah/training-and-evals/part-6
ex:anti-bias-instructions
knownToWorkWellblah/training-and-evals/part-6
ex:arena-hard
preferredOverblah/training-and-evals/part-6
shorter token limits
presupposesExistenceOfblah/training-and-evals/part-6
ex:eval-questions
targetsAiJudgesblah/training-and-evals/part-6
null

References (2)

2 references
  1. [1]Part 616 facts
    customctx:discord/blah/training-and-evals/part-6
  2. customctx:discord/blah/training-and-evals/6
    • full texttraining-and-evals-6
      text/plain3 KBdoc:agent/training-and-evals-6/09d7a12a-230e-40d3-a120-15b7fbb9ef5e
      Show excerpt
      [2026-02-19 02:10] ajaxdavis: ``` Prompt architecture (based on research from MT-Bench, Arena-Hard, AlpacaEval):

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.