© 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. 5822
    Prev
    Next
    INDEX
    Explanations

    Consistent is

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

    The neuron strongly activates on verbs used in second‐person commands or calls to action—that is, imperative forms instructing the reader what to do.

    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
     sämt
    0.88
    സ്ത
    0.86
     ৩৪
    0.85
    ╾
    0.82
     расстоя
    0.80
     GMO
    0.80
     mínima
    0.80
     escritos
    0.80
     малы
    0.79
     Bhagavata
    0.79
    POSITIVE LOGITS
    ac
    0.90
    دا
    0.78
    can
    0.76
    <0x81>
    0.73
    حت
    0.71
    オン
    0.67
    يكم
    0.66
    ي
    0.66
     confinement
    0.66
    ab
    0.65
    Activations Density 0.003%

    No Known Activations