© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 10520
    Prev
    Next
    INDEX
    Explanations

    harassment

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

    The neuron signals on words referring to harassment or stalking (e.g. “harassed,” “harassment,” “stalking,” “terrorised”).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/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
     guerrero
    -1.84
     beneficiario
    -1.69
     profeta
    -1.63
    話は
    -1.59
    The
    -1.54
    </em>
    -1.51
     turista
    -1.48
     alza
    -1.48
     halted
    -1.43
     arbeitete
    -1.43
    POSITIVE LOGITS
    }
    1.69
    ging
    1.65
     kısm
    1.53
     harass
    1.52
     भी
    1.52
     olur
    1.49
     witte
    1.45
    几乎
    1.45
     nuisance
    1.43
     ثلاثة
    1.41
    Activations Density 0.039%

    No Known Activations