© 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
    2. Gemma-2-27B
    3. 10-GEMMASCOPE-RES-131K
    4. 22827
    Prev
    Next
    INDEX
    Explanations

    dealing with conditions

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

    The neuron fires strongly on the head nouns of high-level framing phrases (e.g. “story” in “this whole story,” “part” in “the worst part of working at Google,” “trick” in “the trick to dealing with this,” “secret” in “all this success,” etc.) used to introduce or summarize key points.

    oai_token-act-pair · o4-mini
    Triggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/width_131k
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Negative Logits
    🛇
    -1.48
    FunctionFlags
    -1.39
    меты
    -1.38
    forChild
    -1.36
     població
    -1.33
    __':
    
    -1.30
    ihnachts
    -1.29
    &(
    -1.28
    //});
    -1.27
    mby
    -1.27
    POSITIVE LOGITS
     is
    2.94
     are
    2.20
     would
    2.00
     will
    1.83
     you
    1.66
     being
    1.55
     must
    1.47
     was
    1.45
     tấn
    1.44
     then
    1.39
    Activations Density 0.153%

    No Known Activations