© 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. 128101
    Prev
    Next
    INDEX
    Explanations

    The neuron activates on numerical tokens, especially multi‐digit numbers or numeric sequences.

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

    percentages and monetary amounts

    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
     two
    -2.80
     at
    -2.78
     people
    -2.67
     all
    -2.63
     three
    -2.59
     as
    -2.56
     this
    -2.48
     four
    -2.48
     only
    -2.45
     have
    -2.41
    POSITIVE LOGITS
     constater
    2.17
    ܆
    2.16
     corriger
    2.13
    たまに
    2.05
     maniere
    2.05
     effectuer
    2.00
     praticamente
    2.00
     ridiculously
    2.00
     ludicrous
    2.00
     AMAZING
    1.97
    Activations Density 0.092%

    No Known Activations