© 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. 31-GEMMASCOPE-2-RES-16K
    7. 11015
    Prev
    Next
    INDEX
    Explanations

    characters and letters

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

    The neuron is especially sensitive to occurrences of rune‐name tokens (e.g. “Ansuz,” “Gibor”) and other esoteric proper‐noun labels.

    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_31_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
     эксперти
    0.56
     മു
    0.54
     फडण
    0.54
     পারি
    0.53
    曈
    0.52
     १८४
    0.51
    涯
    0.50
     fadeInLeft
    0.49
     긴
    0.49
     रविदास
    0.49
    POSITIVE LOGITS
     letter
    1.84
     char
    1.81
     letters
    1.81
     alphabet
    1.80
    字母
    1.80
    letter
    1.80
     Buchstaben
    1.79
    字符
    1.77
     chars
    1.77
    char
    1.73
    Activations Density 0.105%

    No Known Activations