© 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. 24-GEMMASCOPE-2-RES-16K
    7. 2610
    Prev
    Next
    INDEX
    Explanations

    The neuron reliably spikes on the first “content” word that begins a new paragraph or segment, regardless of its part of speech.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    compound nouns or technical terms where "front" or other position/orientation words appear in specific contexts.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @jyhe0408

    short, formulaic multiword collocations and discourse-marker phrases (common two- to three-word chunks like verb+preposition, determiner+noun/adjective, or other fixed expressions signaling relations or transitions).

    oai_token-act-pair · gpt-5Triggered 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
     Họ
    1.00
     Profes
    0.78
     चाहती
    0.75
    are
    0.72
    k
    0.71
     profes
    0.71
     ihnen
    0.71
     Medical
    0.71
     ellos
    0.70
     ဒါ
    0.70
    POSITIVE LOGITS
     ²
    0.73
    ‚‚
    0.73
    ³,
    0.73
    ONOM
    0.72
    껭
    0.72
    AN
    0.71
    dört
    0.70
    ²,
    0.69
    位数
    0.69
    টসঅ্যাপ
    0.68
    Activations Density 0.001%

    No Known Activations