© 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
    2. Gemma-3-12B
    3. 31-GEMMASCOPE-2-RES-16K
    4. 5220
    Prev
    Next
    INDEX
    Explanations

    limitations, restrictions, style, flexibility

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

    This neuron fires on headline‐style or list‐item lead-in words (typically title-cased or uppercase tokens) that mark section headings or bullet-point topics.

    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_31_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
    </th>
    0.70
    </h1>
    0.67
    ();}
    0.61
     ١
    0.59
    יי
    0.59
    1
    0.57
    ៕
    0.55
    </a>
    0.54
    ቼ
    0.53
    in
    0.53
    POSITIVE LOGITS
     Although
    1.03
    Although
    0.97
     Whenever
    0.91
     Though
    0.90
     Because
    0.90
     While
    0.89
     Aunque
    0.88
     Despite
    0.87
     Throughout
    0.84
     Undoubtedly
    0.84
    Activations Density 0.221%

    No Known Activations