© 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. 10-GEMMASCOPE-RES-131K
    7. 104351
    Prev
    Next
    INDEX
    Explanations

    processing items

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

    The neuron specifically fires on occurrences of the word "through," especially in contexts like “read through,” “loop through,” or “scrolled through.”

    oai_token-act-pair · o4-miniTriggered 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
    Not in Any Lists

    No Comments

    Negative Logits
    .
    -1.96
     When
    -1.85
     Neither
    -1.75
     Those
    -1.73
     Extremely
    -1.73
    s
    -1.64
     Perhaps
    -1.62
     because
    -1.60
     Despite
    -1.59
     kept
    -1.55
    POSITIVE LOGITS
    雠
    1.98
     butik
    1.93
     dezelve
    1.91
    釤
    1.86
    ization
    1.79
     ſy
    1.77
    隳
    1.77
    formatie
    1.71
    kende
    1.71
     yyj
    1.70
    Activations Density 0.013%

    No Known Activations