© 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-2-27B
    3. 34-GEMMASCOPE-RES-131K
    4. 86886
    Prev
    Next
    INDEX
    Explanations

    __ followed by keywords

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

    The neuron consistently fires on code-like tokens—especially those containing digits or underscores—such as numeric literals, macro/constant names, and “dunder” identifiers (e.g. __init__, __get__).

    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
     morbidity
    -0.76
     Nt
    -0.75
     baker
    -0.75
     Kid
    -0.74
    loth
    -0.71
     PIG
    -0.71
     finder
    -0.71
    okan
    -0.71
     KID
    -0.70
     metallurgy
    -0.69
    POSITIVE LOGITS
    __(
    0.86
    __::
    0.79
    __.
    0.75
    Recher
    0.74
    __,
    0.73
    __:
    0.73
     통해
    0.72
    %,
    0.71
     хотите
    0.70
    吼
    0.68
    Activations Density 0.047%

    No Known Activations