© 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
    2. Gemma-3-12B
    3. 24-GEMMASCOPE-2-RES-16K
    4. 8324
    Prev
    Next
    INDEX
    Explanations

    private class constructor

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

    This neuron strongly activates on code identifier tokens—especially variable or member names with underscores (e.g. “m_frame”)—i.e. parts of source‐code identifiers.

    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
    Without
    0.71
    while
    0.69
    Taking
    0.68
    While
    0.66
    ($
    0.65
    ליך
    0.64
    (
    0.64
    :])
    0.63
    throw
    0.63
    :)
    0.62
    POSITIVE LOGITS
     BEEN
    0.74
     costante
    0.72
     alumnos
    0.71
     olnud
    0.71
     ollut
    0.70
     genellikle
    0.70
     telah
    0.70
     właścic
    0.70
     atualmente
    0.69
     gerente
    0.68
    Activations Density 0.012%

    No Known Activations