© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 25795
    Prev
    Next
    INDEX
    Explanations

    roles and their actions

    np_acts-logits-general · gemini-2.5-flash-lite

    The neuron strongly activates on words naming people’s roles or titles (e.g. manager, moderator, cook, owner, Senator, Director, officers).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_22/width_131k
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    ALREADY
    -1.07
    δη
    -1.05
    almost
    -0.98
    andidates
    -0.97
    already
    -0.96
    και
    -0.95
    nearly
    -0.94
    ᠂
    -0.94
     détect
    -0.94
    formatics
    -0.93
    POSITIVE LOGITS
     or
    1.43
     who
    1.20
     himself
    1.13
     whose
    1.11
     this
    1.10
     if
    1.05
     whom
    1.03
     directly
    1.02
     which
    0.98
     personally
    0.98
    Activations Density 0.011%

    No Known Activations