© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-12B
    5. 31-GEMMASCOPE-2-RES-16K
    7. 13961
    Prev
    Next
    INDEX
    Explanations

    exploitation and abuse

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

    The neuron detects words that denote wrongdoing or abusive behavior (e.g. fraud, abuse, coercion, exploitation, lying).

    oai_token-act-pair · o4-mini
    Triggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-12b-pt/resid_post/layer_31_width_16k_l0_medium
    Prompts (Dashboard)
    392,802 prompts, 256 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Negative Logits
    Violence
    0.62
     Violence
    0.60
    暴力
    0.58
    ֶ
    0.57
    ihad
    0.57
     उन्
    0.56
    angs
    0.54
    ั่ว
    0.54
    恐惧
    0.53
     সহিংস
    0.52
    POSITIVE LOGITS
     majest
    0.77
     committed
    0.67
     কেননা
    0.66
     perpetrated
    0.65
     detection
    0.64
     transparen
    0.64
     detectable
    0.63
    нические
    0.63
     magní
    0.62
     detector
    0.61
    Activations Density 0.246%

    No Known Activations