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

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 84316
    Prev
    Next
    INDEX
    Explanations

    roles and characters

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

    The neuron detects mentions of someone performing or being cast in a role (e.g. “plays the part of,” “role,” “character,” “villain,” “played,” “starred as”).

    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
     increase
    -1.20
     from
    -1.17
     when
    -1.13
     increases
    -1.10
     can
    -1.08
    並且
    -1.08
     sufficient
    -1.07
     用于
    -1.04
    操作系统
    -1.04
     where
    -1.04
    POSITIVE LOGITS
     role
    1.84
     roles
    1.68
     personnage
    1.46
     personnages
    1.45
     zei
    1.34
     rekening
    1.34
     scènes
    1.33
    +"_
    1.30
     роль
    1.29
    roles
    1.27
    Activations Density 0.023%

    No Known Activations