© 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. 129008
    Prev
    Next
    INDEX
    Explanations

    The neuron activates on numeric tokens—especially multi‐digit sequences like years, version numbers, or other digits in copyright/license headers.

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

    copyright notices

    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
     agradável
    -1.07
    衒
    -1.05
    erobic
    -1.05
     hadapan
    -1.04
     companhia
    -1.04
     хочет
    -1.03
     administração
    -1.03
     encantador
    -1.03
     Buchung
    -1.03
     gemeinsamen
    -1.02
    POSITIVE LOGITS
    p
    1.09
    w
    1.05
    から
    0.98
    ö
    0.96
    к
    0.94
     харак
    0.94
    ax
    0.92
    ke
    0.91
    el
    0.91
    on
    0.91
    Activations Density 0.389%

    No Known Activations