© 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-3-12B
    5. 24-GEMMASCOPE-2-RES-16K
    7. 2769
    Prev
    Next
    INDEX
    Explanations

    he said or she said

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

    This neuron detects speaker attribution phrases—names and reporting verbs (e.g. “said,” “added,” “asserted”) that introduce or follow quoted speech.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-12b-pt/resid_post/layer_24_width_16k_l0_medium
    Prompts (Dashboard)
    392,802 prompts, 256 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    EF
    0.64
    unts
    0.61
    :'',
    0.60
    :
    0.58
    )、
    0.55
    ange
    0.54
    ef
    0.54
     ধন্যবাদ
    0.54
    文件中
    0.52
    值为
    0.52
    POSITIVE LOGITS
    .¹
    0.74
    ​.
    0.73
    ¹.
    0.73
    ‌.
    0.70
     ”.
    0.70
    ².
    0.68
    .​​
    0.66
     ».
    0.66
    ሉ።
    0.65
     ذریع
    0.64
    Activations Density 0.025%

    No Known Activations