© 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. 47687
    Prev
    Next
    INDEX
    Explanations

    code punctuation and separators

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

    The neuron strongly activates on non-natural‐language tokens such as code identifiers, file‐metadata fields, and other technical or formatting elements (e.g. “DataBindingUtil,” “serialized,” “font,” version numbers, license tags, etc.).

    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
     keep
    -0.75
    بان
    -0.74
     shirt
    -0.74
    Full
    -0.73
    保持
    -0.73
    autoIncrement
    -0.72
    ocs
    -0.72
     ऊ
    -0.72
    shirt
    -0.71
     bonita
    -0.71
    POSITIVE LOGITS
     fijo
    0.82
     zetten
    0.70
    Люби
    0.68
    Catalyst
    0.66
    Retrieval
    0.66
    רד
    0.65
    﨑
    0.64
     gesetzt
    0.64
    给了
    0.64
    Apro
    0.63
    Activations Density 0.047%

    No Known Activations