© 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. 5077
    Prev
    Next
    INDEX
    Explanations

    The neuron strongly activates on content words that begin a sentence—i.e. sentence-initial tokens right after a sentence boundary.

    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
    إ
    0.77
     wed
    0.76
    і
    0.70
    мі
    0.68
     la
    0.68
     й
    0.67
    jug
    0.67
    Categoria
    0.67
    І
    0.67
    чно
    0.67
    POSITIVE LOGITS
    að
    0.79
    িশালী
    0.78
     sekali
    0.77
     აღმასრულ
    0.77
    âteaux
    0.77
     zudem
    0.75
     politiche
    0.75
     vaginale
    0.75
     smells
    0.74
    sticky
    0.74
    Activations Density 0.001%

    No Known Activations