© 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. 151
    Prev
    Next
    INDEX
    Explanations

    words that are often followed by something specific

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

    The neuron fires on the main predicate words of sentences—especially forms of “be,” “have,” “show,” “work,” and other principal verbs (often in passive or analytic constructions).

    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
    maß
    -1.05
     gua
    -0.96
    ionar
    -0.93
     puramente
    -0.91
     ￥
    -0.91
    INAL
    -0.90
     Krat
    -0.90
    】：
    -0.90
     pipa
    -0.90
     exacer
    -0.89
    POSITIVE LOGITS
    ŷ
    0.84
    chung
    0.82
     November
    0.82
     aliviar
    0.79
     Hinweis
    0.79
    чили
    0.79
     it
    0.79
    回戦
    0.76
     ноября
    0.76
    ̦
    0.76
    Activations Density 0.121%

    No Known Activations