© 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. 22-GEMMASCOPE-RES-131K
    7. 128617
    Prev
    Next
    INDEX
    Explanations

    The neuron strongly activates on punctuation tokens used in code and regex syntax—especially backslashes, square‐bracket character classes, and angle brackets.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    celebratory emojis

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_22/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
     イヤ
    -0.92
     lana
    -0.92
     états
    -0.91
     deko
    -0.91
     HIER
    -0.89
    biamo
    -0.89
    ásticos
    -0.88
     Loa
    -0.88
     résister
    -0.87
     lâng
    -0.86
    POSITIVE LOGITS
     sebe
    0.90
    īts
    0.85
    torta
    0.84
     others
    0.83
    ſen
    0.82
    цтва
    0.81
     terus
    0.81
     Os
    0.81
     Gebiete
    0.81
     Fiji
    0.80
    Activations Density 0.023%

    No Known Activations