© 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-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 70634
    Prev
    Next
    INDEX
    Explanations

    listed buildings

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

    The neuron selectively activates on standalone numeric tokens that are part of arithmetic/math problem expressions (i.e. sequences of digits in calculation questions and answers).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    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
     included
    -0.90
     mairie
    -0.88
     opcional
    -0.87
    ﻹ
    -0.83
    юль
    -0.83
    してない
    -0.82
     this
    -0.82
     részt
    -0.79
    AspectRatio
    -0.78
     cilindros
    -0.77
    POSITIVE LOGITS
     unſ
    1.02
    atically
    0.97
    IZED
    0.94
     Manus
    0.93
    <0xE0>
    0.91
     derelict
    0.91
     ento
    0.91
     ATLAS
    0.90
     eſſ
    0.90
    atize
    0.90
    Activations Density 0.018%

    No Known Activations