© 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. 72851
    Prev
    Next
    INDEX
    Explanations

    what, why, how, q

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

    This neuron is primarily tracking the boundaries of quoted speech—punctuation like quotation marks (and the first emphasized word in dialogue) that mark direct quotes.

    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
    funding
    -0.86
     andere
    -0.86
     essayer
    -0.82
    searching
    -0.81
    べり
    -0.80
    ámara
    -0.79
     other
    -0.79
    otyp
    -0.78
    ricle
    -0.77
    の部分
    -0.77
    POSITIVE LOGITS
     wha
    0.96
     what
    0.95
    WHAT
    0.88
    listbox
    0.88
     aton
    0.88
    どういう
    0.86
    ibilité
    0.85
     Wha
    0.84
    原来
    0.84
     infatti
    0.82
    Activations Density 0.002%

    No Known Activations