Skip to content
H
Howardism
Howardism · Vol. 03Plate I · No. 01

Interpretability, tagged.

Notes23TagInterpretabilityOldest8 May 2026Newest26 Oct 2026

Every article tagged interpretability, newest first.

  1. C01
    Minimal-Pair Probing: Present, Readable, Used
    InterpretabilityProbesLatent Reasoning+3
    · 16′
  2. C02
    Model Diffing
    InterpretabilityModel DiffingModel Organisms+2
    · 12′
  3. C03
    Metagaming
    AlignmentSafetyInterpretability+2
    · 14′
  4. C04
    Natural-Language Autoencoders (NLAs)
    InterpretabilityActivation MonitoringVerbalizer+6
    · 50′
  5. C05
    Probe-Gated Monitor Cascade
    InterpretabilityActivation ProbesMonitoring+4
    · 20′
  6. C06
    WorkspaceBench
    InterpretabilityBenchmarksActivation Monitoring+2
    · 16′
  7. C07
    Sandbagging Elicitation (Reference & Context Grafting)
    InterpretabilityAlignmentSafety+2
    · 13′
  8. C08
    Interference Weights
    InterpretabilitySuperpositionMechanistic Interpretability+1
    · 19′
  9. C09
    Invisible Reasoning (Filler-Token Latent Computation)
    InterpretabilityReasoningChain Of Thought+2
    · 27′
  10. E10
    Transluce
    EntityOrgAI Evaluation+2
    · 9′
  11. C11
    Introspective Coupling
    InterpretabilityIntrospectionSelf Report+3
    · 17′
  12. C12
    Model Organisms
    InterpretabilityAlignmentSafety+2
    · 17′
  13. C13
    Self-Report as a Safety Signal
    AlignmentSafetyIntrospection+3
    · 43′
  14. C14
    Access-Consciousness Indicators in AI
    AlignmentModel WelfareConsciousness+1
    · 12′
  15. C15
    The Assistant Persona in the Workspace
    InterpretabilityAlignmentPersona+1
    · 14′
  16. C16
    Automatic vs. Flexible Cognition in LLMs
    InterpretabilityCognitionReasoning+1
    · 15′
  17. C17
    Counterfactual Reflection Training
    AlignmentTrainingHonesty+1
    · 6′
  18. C18
    Internal Signatures of Misalignment
    AlignmentInterpretabilityMonitoring+2
    · 21′
  19. C19
    Jacobian Lens (J-lens)
    InterpretabilityAlignmentMonitoring+1
    · 53′
  20. C20
    The Global Workspace in Language Models (J-space)
    InterpretabilityAlignmentCognition+1
    · 45′
  21. C21
    Evaluation Awareness & Grader Gaming
    AlignmentSafetyEvaluation+2
    · 71′
  22. C22
    White-Box Activation Monitoring
    InterpretabilityAlignmentSafety+2
    · 100′
  23. C23
    Chain-of-Thought Monitorability
    AlignmentSafetyChain Of Thought+2
    · 90′