© 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. 12854
    Prev
    Next
    INDEX
    Explanations

    statements of past actions

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

    This neuron detects the initial few tokens immediately following an opening quotation mark—that is, the start of a quoted phrase.

    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
     this
    -2.77
     only
    -1.99
     from
    -1.98
     their
    -1.98
     at
    -1.96
     several
    -1.95
     first
    -1.91
     as
    -1.89
     before
    -1.86
     three
    -1.84
    POSITIVE LOGITS
     maniere
    1.93
     offrir
    1.86
     ayur
    1.63
     colossal
    1.60
     murale
    1.60
     DIFFERENT
    1.58
     goofy
    1.56
     flamboyant
    1.56
     ſou
    1.55
     decidedly
    1.54
    Activations Density 0.035%

    No Known Activations