© 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. 10904
    Prev
    Next
    INDEX
    Explanations

    start of description

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

    The neuron strongly activates on the final content word (the terminal noun or material) just before the period or punctuation at the end of short product‐listing or feature sentences.

    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
     ორ
    0.65
    ovým
    0.63
     новым
    0.61
    ophylline
    0.59
    bicyclo
    0.59
     explorations
    0.58
     корректи
    0.57
     exploration
    0.57
     рассмотре
    0.57
     வகையான
    0.57
    POSITIVE LOGITS
     horribly
    0.58
     décider
    0.55
     nicer
    0.55
     citizenry
    0.54
    ஒரு
    0.53
     courteous
    0.53
     victimes
    0.53
    سين
    0.53
    に使
    0.51
     dringend
    0.50
    Activations Density 0.001%

    No Known Activations