Interactive example · {{ demo.title }}Interactive

{{ demo.hintTeknik }}

{{ learnHead }}

{{ learnBody }}

{{ b.label }}

{{ learnQuestion }}

{{ t.name }}

{{ intel.name }} Illustrative AI summary: {{ intel.level }}

{{ intelSkillLabel }}

{{ intel.desc }}

{{ intelCondition }}

The indicator is illustrative; it is not measured performance.

{{ b.v }} {{ b.place }}

{{ binContribLabel }}

{{ binStr }} = {{ binVal }}

{{ binSumLabel }}: {{ binSumExpr }}{{ binZeroLabel }}

{{ binLast }}

{{ c.v }} ▲

State: {{ tStateLabel }}Completed result{{ tTemporaryLabel }}: {{ tVal }}

{{ tLastLine }}

{{ tLastCell }}

{{ tCompare }}

Start: {{ tStartFixed }}. {{ tSimpleStory }}{{ tStory }}

{{ tAutoLabel }} Step{{ tGuidedLabel }}Restart Start: {{ s.v }}

{{ cycleInstr }}

{{ cycleActiveText }}

{{ p.label }}{{ p.badge }} →

Phase: {{ cyclePhaseLabel }} · Information card: {{ cyclePart.label }}

{{ cyclePhaseNote }}

{{ cyclePart.desc }}

{{ cycleResult }}

{{ cycleNextLabel }}{{ cyclePlayLabel }}

{{ classifyScope }}

{{ it.label }}

{{ g.label }}

{{ b.label }}

{{ it.mark }}

{{ classifyReasonLabel }}: {{ it.fb }}

{{ classifyNote }}

{{ n.label }} {{ n.rel }} ▸

{{ q.label }}

{{ chainNextLabel }} →

{{ chainVerdict }}

{{ f.glyph }} {{ f.label }}

{{ expertLegend }}

{{ r.id }} IF {{ r.cond }} → THEN {{ r.then }} {{ r.fire }}

{{ expertChain }}

{{ expertFactsLabel }}: {{ expertFacts }}

{{ expertRulesLabel }}: {{ expertRules }}

SYSTEM ADVICE

{{ advice }}

{{ g.label }} Search

{{ c.label }}

Startgoalwallexamined cellfound route

{{ gridLast }}

Examined cells: {{ gridExplored }} Cells in route: {{ gridPathLen }} (start and goal included)

{{ gridCompareLabel }}

{{ r.line }}

{{ mkAssumption }}

{{ s.label }}

{{ mkNextRowLabel }} · {{ mkCurrent }}

{{ p.label }}

{{ p.pct }}

Next day {{ mkAutoLabel }} {{ mkResetLabel }}

{{ mkLast }}

{{ mkRound }}

Days generated: {{ mkDays }} · observed proportions in this run: {{ mkDist }}

{{ e.text }}

{{ spamCluesLabel }}

{{ g.t }}

{{ spamPrepared }}

FEATURES (INPUT)

{{ f.glyph }}{{ f.label }}

SUPPLIED CORRECT LABEL

{{ spamLabel }}

{{ t.label }}

{{ scMini }}

{{ scScaleLabel }}

{{ b.label }}

{{ scActualLabel }}: {{ scSel.y0 }}

{{ scPredLabel }}: {{ scSel.p0 }}

{{ scGapLabel }}: {{ scSel.g0 }}

{{ scScore }}

{{ scFitLabel }}

Centers (✕) are fixed; this panel does not learn them. The red outlier was also preselected; no outlier threshold is computed here.

{{ b.label }}

{{ kmFirstLabel }}: {{ kmD0 }}

{{ kmSecondLabel }}: {{ kmD1 }}

{{ kmNearLabel }}: {{ kmNear }}

{{ kmLabel }}

Step size: {{ o.label }}

Step {{ descN }} · model setting: {{ descXVal }} · error score: {{ lossVal }}

{{ descDirLabel }}

{{ descCompareLabel }}: {{ descPrevLoss }} → {{ lossVal }}

Step {{ descAutoLabel }} restart

{{ o.label }}

All three curves are predefined illustrations; none is fitted to the data.

{{ fitVerdict }}

{{ n.caption }}{{ neuronWeightLabel }}: {{ n.w }}

{{ neuronContribLabel }}: {{ n.val }} × {{ n.w }} = {{ n.c }}

{{ neuronConst }}

{{ neuronZLabel }} {{ neuronZ }}

{{ neuronActNow }}: {{ neuronActRule }}

{{ neuronAvalLabel }} {{ neuronAval }}

{{ o.label }}

{{ neuronThrLabel }}: {{ neuronFire }}

{{ learnToggleLabel }} {{ neuronRestartLabel }}

Inputs: {{ ffInputsNow }} {{ b.v }} {{ ffFeedLabel }} {{ learnToggleLabel }}

{{ ffBright }}
{{ ffThick }}

{{ b.label }}

{{ ffSelName }} · {{ ffContribLabel }}

{{ ffSelLine }}

{{ bpStatus }}

{{ bpInputsLabel }}

{{ bpGuidedConnection }}

{{ bpSelectedEffect }}

{{ bpSelectedUpdate }}

{{ m.t }}

{{ bpSetup }}

{{ bpLegend }}

{{ b.label }}

{{ bpExplorationSelection }}

{{ s.name }}
{{ s.values }}
{{ s.gradient }}

{{ bpExplorationFormula }}

{{ bpTrainLabel }}

{{ learnToggleLabel }} {{ bpResetLabel }}

Filter: {{ f.label }} Slide

Image (7×7)

→

Feature map (5×5)

{{ convRespLabel }}: {{ convResp }}

{{ convSum }}

{{ convLegend }}

{{ convRestart }}

{{ rnnHeading }}

{{ r.name }}

{{ rnnComponentLabel }}: {{ r.focus }}

{{ rnnPreviousLabel }}: {{ r.before }}

{{ rnnAllLabel }}: {{ r.all }}

{{ rnnInputContributionLabel }}: {{ rnnInputContribution }}

{{ rnnNextLabel }} {{ rnnCarryLabel }} {{ rnnDetailsLabel }} {{ rnnResetLabel }}

{{ rnnBounds }}

{{ ganBounds }}

{{ ganStageText }}

{{ c.name }}

{{ p.caption }}

{{ ganSourceLabel }}: {{ c.source }}

{{ c.answer }}

{{ c.comparison }}

{{ ganSendLabel }}

{{ ganRoleQuestion }}

{{ r.name }}

{{ ganFeedback }}

{{ ganPath }}

{{ ganResetLabel }}

Year {{ expYear }} in the assumed series

{{ expCount }}

Transistors · doublings: {{ expDoublings }} (assumed series)

{{ expPrevLabel }}: {{ expPrev }} → {{ expNext }}

{{ expBarLabel }}

{{ expTransistor }}

+2 years → double restart

{{ e.label }}

{{ w.text }}

{{ tokSelWord }} → {{ tokSelPieces }}

{{ tokRule }}

{{ t.text }}

{{ tokCount }} illustrative pieces · {{ tokWords }} words

In this splitter, “##” marks a continuation of the same word.

The points were placed by hand; they are not embeddings learned by a real model.

{{ embedPosLabel }}: {{ embedPos }}

{{ embedCaption }}

Illustrative shares: the information shares for the selected piece, including its own share, total 100%. Earlier and later pieces can be used; these shares are not outputs from a trained model.

{{ t.text }}

{{ attnBarHead }}

{{ attnVNote }}

{{ h }}Σ {{ c.t }}

{{ attnCaption }}

{{ r.name }}{{ r.pre }}{{ r.neu }}{{ r.post }}

{{ g.t }}{{ g.mark }}

{{ r.uText }}

{{ genStepLabel }}

{{ genBasicNote }}
{{ genFixedNote }}

{{ genToggleLabel }}

{{ genBridge }}

{{ c.word }}

{{ c.pct }}

{{ genNote }}

{{ genMainLabel }} restart

{{ t.label }}

TYPE OF TRAINING EXAMPLE

{{ trainData }}

↓

LEARNING OBJECTIVE

{{ trainLearns }}

↓

PREPARED ANSWER TO “WHAT IS THE CAPITAL OF TÜRKİYE?”

{{ trainOutput }}

{{ trainNote }}

noisestep progress {{ diffPct }} · revealed pixels {{ diffClean }}/64

{{ diffCaption }}

Context window (a sliding window used in this demo) · {{ ctxUsed }} / {{ ctxMax }} words

{{ ctxOutLabel }}

{{ w.text }}

{{ ctxInLabel }}

{{ w.text }}

{{ ctxCaption }}

Add word restart

{{ p.label }}

PROMPT

{{ l.text }}

{{ promptCheckLabel }}:{{ c.g }} {{ c.t }}

Completeness: {{ promptQualityLabel }} (not a quality score){{ promptTier }}

PREPARED ANSWER FOR COMPARISON no model runs in this panel

{{ g.label }}

{{ g.t }}^{← {{ g.tag }}}

Predefined source–answer pairs · no real search or model call

{{ q.label }}

{{ ragToggleLabel }}

{{ ragStateLabel }}: {{ b.label }}

📄 EXAMPLE SOURCE (FICTIONAL)

{{ g.t }}

{{ ragHint }}

{{ ragAnswerLabel }}

{{ g.t }}

{{ ragCheckLabel }}

⚠ Unverified; it may sound right but be wrong.

TASK · SCRIPTED SCENARIO

{{ agentTask }}

No steps taken yet.

{{ agentLegend }}

{{ s.num }}STEP

Scripted next task: {{ s.thought }}

🔧 {{ s.tool }} · {{ s.action }}

{{ s.calc }}

→ observation: {{ s.obs }}

✓ {{ s.final }}

↓ {{ s.obs }}

{{ agentDoneLabel }}

Next stepNext step restart

{{ p.label }}

{{ archName }}

{{ archDesc }}

“{{ archQ }}”

{{ archAccessLabel }}: {{ b.label }}

📄 {{ archSource }}

{{ h.n }}. {{ h.from }} → {{ h.to }}{{ h.carry }}

{{ archAnswer }}

{{ archOptNote }}

{{ archFullNote }}

{{ archNextLabel }} {{ archResetLabel }}

{{ d.label }}

{{ sectorName }}

▸{{ e.text }}

{{ c.k }}

{{ c.v }}

{{ biasNote }}

Illustrative bias setting{{ biasPct }}

{{ b.label }}

{{ biasFormula }}

{{ biasKeyText }}

Group A approval rate{{ biasA }}

Group B approval rate{{ biasB }}

{{ biasChange }}

{{ biasCaption }}

{{ a.label }}

{{ explainResult }}

{{ explainBase }}

{{ explainStripLabel }}: {{ explainStrip }}

{{ f.label }}

{{ f.val }} · {{ f.dir }}

{{ explainNextLabel }}{{ explainAllLabel }}

{{ explainCaption }}

Explain →

{{ dfBoundary }}

{{ a.t }}

Case {{ dfIdxNum }} / {{ dfTotal }} · fictional example · matched: event {{ dfScE }} · origin {{ dfScO }} · channel {{ dfScC }}

{{ dfText }}

{{ dfQNum }} / 3 · {{ dfQ }}

{{ o.label }}

{{ r.head }} {{ r.text }}

{{ dfNote }}

Next → restart

{{ u.label }}

{{ r.k }}: {{ r.v }}

{{ o.label }}

{{ u.mark }}

{{ u.why }}

{{ regCaption }}

restart

{{ alignBoundary }}

🎯 {{ g.label }}

{{ alignGoalLabel }}

{{ alignGoal }}

{{ alignIntentLabel }}

{{ alignIntent }}

{{ alignMet }}

{{ alignMissed }}

POSSIBLE UNWANTED BEHAVIOR

{{ alignBehavior }}

💡 {{ alignLesson }}

Show →

Exchange {{ turIdxNum }} / {{ turTotal }} · fictional example

{{ turPrompt }}

{{ turReply }}

{{ turVerdict }}

💡 {{ turTell }}

Next → restart

Human 🧑 Machine 🤖 Cannot tell from this text

{{ crTableHead }}

{{ r.in }}→{{ r.out }}{{ r.mark }}

Note {{ crIdxNum }} / {{ crTotal }} · fictional assumption: the person in the room knows no Chinese

INCOMING CHARACTER SEQUENCE

{{ crIn }}

→

YOUR SELECTED REPLY

{{ crOut }}

{{ crFeedback }}

Meanings of the correct table pair: “{{ crInMeaning }}” → “{{ crOutMeaning }}”. If you assume the role of a person who knows no Chinese, you can match this reply without knowing its meaning; this says nothing about your actual knowledge of Chinese. Is asking whether the person understands the same as asking whether the person, rules and processes together understand? These three pairs do not settle the debate.

Next → restart

Send your matched reply →

{{ t.label }}

{{ capName }}

{{ r.k }}: {{ r.v }}

Three distinct definitions · not a measured intelligence ranking or an AGI progress percentage

{{ capDesc }}

{{ capPrompt }}

{{ singPrompt }}

{{ b.label }}

{{ singLegend }}

{{ singName }}

{{ singSchema }}

{{ singNote }}

{{ s.label }}

Whose role would you investigate? You can pick more than one party.

{{ p.label }}

{{ respEmptyText }}

Show the role explanation →

{{ respVerdict }}

{{ respWho }}

{{ respCase }}

{{ respNote }}

{{ respRights }}

{{ rfHead }}

{{ rfIntro }}

{{ rfReasonQ }}

{{ o.g }} {{ o.label }}

{{ rfFactQ }}

{{ o.g }} {{ o.label }}

{{ rfCompareLabel }}

{{ f }}

{{ a.label }}. {{ a.text }}

{{ rfLimit }}

{{ learnLocalText }}

Calculations and symbols

{{ n.t }}

What’s happening?

{{ neOluyor }}

Check answers {{ quizLegend }} {{ quizScore }} / {{ quizTotal }} correct