© 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. 79734
    Prev
    Next
    INDEX
    Explanations

    IP, codes, email, copyright, address, call letters

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

    The neuron fires on structured identifier tokens and labels—e.g. “IP,” “email,” “hex” (as in hex‐color)—marking bits of text that denote technical addresses or codes.

    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
    стасия
    -1.14
    /',
    -1.04
    팎
    -1.02
    で使
    -1.01
     beginnetje
    -0.97
    liputi
    -0.97
    ilerini
    -0.96
    raza
    -0.96
     ..."
    -0.94
    /";
    -0.94
    POSITIVE LOGITS
    ielded
    0.99
     цвят
    0.98
     to
    0.98
     aforementioned
    0.98
     mentioned
    0.95
     ruddy
    0.95
    したのが
    0.94
     that
    0.94
    ването
    0.93
     again
    0.93
    Activations Density 0.002%

    No Known Activations