© 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. 12600
    Prev
    Next
    INDEX
    Explanations

    The neuron strongly activates on the very first substantive word of a new section or paragraph (i.e. it marks section-header or heading tokens).

    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
     beams
    0.56
     Zol
    0.55
    Золо
    0.54
     Orioles
    0.54
     beam
    0.52
     نف
    0.52
    👯
    0.51
    Foo
    0.50
     Foo
    0.48
    WordArray
    0.48
    POSITIVE LOGITS
    attrib
    0.57
    ambarkan
    0.55
     unix
    0.55
     怪
    0.54
    怪
    0.54
    cepción
    0.54
    akkhati
    0.52
     mille
    0.52
     Ше
    0.52
     piş
    0.52
    Activations Density 0.000%

    No Known Activations