© 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-2-27B
    5. 34-GEMMASCOPE-RES-131K
    7. 31550
    Prev
    Next
    INDEX
    Explanations

    de, di, du, dib names

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

    The neuron strongly activates on subword tokens that begin with the letter “D” (e.g. “Dee,” “Div,” “Di,” “Dait,” “Du,” etc.), i.e. word‐initial “D” fragments.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/width_131k
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     leksikon
    -0.81
    搅拌
    -0.81
    ESSMENT
    -0.80
    ểm
    -0.80
    esterno
    -0.79
    糞
    -0.79
    HASAN
    -0.79
    żym
    -0.78
     atmosfer
    -0.78
    について
    -0.78
    POSITIVE LOGITS
    んですよね
    0.79
     aislado
    0.78
     timeval
    0.77
     However
    0.77
    usakan
    0.77
     overrun
    0.76
     millón
    0.76
     altavoz
    0.75
    duplicates
    0.75
    裴
    0.75
    Activations Density 0.008%

    No Known Activations