© 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. 104598
    Prev
    Next
    INDEX
    Explanations

    crime and violence

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

    The main thing this neuron does is detect words naming violent or criminal actions.

    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
     censored
    -0.94
     embal
    -0.94
    라이
    -0.93
    Writer
    -0.92
    Automatically
    -0.88
    许久
    -0.84
     sudor
    -0.84
    PEZ
    -0.83
    alright
    -0.81
    例文帳に追加
    -0.81
    POSITIVE LOGITS
     who
    1.04
     leider
    1.02
     attacks
    1.02
     attack
    0.98
     violence
    0.90
    ますね
    0.89
     someone
    0.88
    attack
    0.87
    attacks
    0.87
     crime
    0.84
    Activations Density 0.083%

    No Known Activations