© 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. 3115
    Prev
    Next
    INDEX
    Explanations

    the/an followed by noun

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

    This neuron fires strongly on words that begin new sections or instructions—typically imperative verbs or gerunds used as headings or commands.

    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
     Mohamed
    0.60
     Overwatch
    0.59
     Natal
    0.57
     Subway
    0.57
     Wale
    0.57
     Miley
    0.56
     [
    0.56
     Aeros
    0.56
     Mikey
    0.55
     Anniversary
    0.55
    POSITIVE LOGITS
    ocsát
    0.57
     aiuta
    0.55
    òi
    0.55
    includes
    0.52
    ๓
    0.51
    requires
    0.49
    aminar
    0.49
    мога
    0.49
     ہیں
    0.49
    ène
    0.48
    Activations Density 0.332%

    No Known Activations