© 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. 31-GEMMASCOPE-2-RES-16K
    7. 12499
    Prev
    Next
    INDEX
    Explanations

    in: action, one, Christ, Philipp

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

    The neuron fires on words that form the titles or headings of sections (i.e. title-cased or headline-style phrases).

    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
    Either
    0.53
    (_.
    0.51
     Bxd
    0.51
    iflower
    0.51
    umbia
    0.49
    🎍
    0.49
     બીજી
    0.49
     मुताब
    0.48
     －
    0.48
    .$,
    0.48
    POSITIVE LOGITS
    พิมพ์
    0.57
    мни
    0.56
     করেছিল
    0.55
     al
    0.54
    كس
    0.53
     fruits
    0.53
     전혀
    0.52
    我覺得
    0.51
     kug
    0.51
    </em>
    0.50
    Activations Density 0.176%

    No Known Activations