© 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
    2. Gemma-3-12B
    3. 31-GEMMASCOPE-2-RES-16K
    4. 6751
    Prev
    Next
    INDEX
    Explanations

    code statements ending in semicolon

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

    The neuron strongly fires on numeric literals and adjacent tokens in code or text (e.g. standalone numbers, “+” concatenations with numbers, and numeric return values).

    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
    ());
    1.12
     ();
    1.12
    ");
    1.11
     {});
    1.11
    );
    1.09
     "");
    1.05
    ();
    1.04
    ');
    1.01
     '');
    0.99
    ];
    0.97
    POSITIVE LOGITS
    പ്പിക്കുന്ന
    0.60
    赋
    0.55
     ಮಾಡಿದ್ದಾರೆ
    0.55
     vaut
    0.53
     केल्या
    0.53
    ാവുന്ന
    0.53
     रस्त
    0.52
    ,"\
    0.52
     সমৃদ্ধ
    0.52
     adaptés
    0.52
    Activations Density 0.000%

    No Known Activations