© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-12B
    5. 24-GEMMASCOPE-2-RES-16K
    7. 8116
    Prev
    Next
    INDEX
    Explanations

    algebraic operations or formatting

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

    This neuron strongly activates on the first meaningful word of new sections or headings (e.g. the leading content token that begins a paragraph, title, or block).

    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
    ཌ
    0.86
     aberr
    0.84
     عاما
    0.82
    𝗎
    0.82
    ュー
    0.73
    ровали
    0.73
    eof
    0.73
    ↣
    0.73
    ☹
    0.72
    ⛓
    0.72
    POSITIVE LOGITS
     western
    0.74
     steamed
    0.68
    ocal
    0.68
    ní
    0.67
     Philosophical
    0.66
     anim
    0.64
    n
    0.64
     southwestern
    0.64
     candy
    0.63
     Library
    0.63
    Activations Density 0.009%

    No Known Activations