© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 127914
    Prev
    Next
    INDEX
    Explanations

    This neuron spikes on the first content-bearing word of each new passage or section—i.e. the “topic” word that kicks off a block of text.

    oai_token-act-pair · o4-mini
    Triggered by @jyhe0408

    time periods or states

    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
    Negative Logits
     occuper
    -1.00
     gela
    -0.98
     harta
    -0.97
     maniere
    -0.94
    lish
    -0.94
     víc
    -0.93
     おしゃれ
    -0.93
     rivi
    -0.92
     horloge
    -0.92
    ...\
    -0.91
    POSITIVE LOGITS
     nor
    1.81
     because
    1.76
     that
    1.52
     but
    1.33
     until
    1.30
     unless
    1.29
     it
    1.29
     which
    1.27
     remain
    1.18
     Nor
    1.14
    Activations Density 0.080%

    No Known Activations