© 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
    2. Gemma-2-27B
    3. 22-GEMMASCOPE-RES-131K
    4. 92948
    Prev
    Next
    INDEX
    Explanations

    qualifier verb combinations

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

    The neuron fires on modal auxiliaries and adverbial intensifiers—that is, words like can/could/would and emphasis‐or‐certainty adverbs such as also, indeed, almost, necessarily, totally, and undoubtedly.

    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
    
    -2.27
    PENDAHULUAN
    -2.13
     biela
    -2.06
    
    -2.05
     rése
    -2.03
     sorprendió
    -1.97
    consejos
    -1.96
     meleg
    -1.95
    saucony
    -1.88
    danke
    -1.87
    POSITIVE LOGITS
    m
    2.67
     be
    2.67
    S
    2.64
    C
    2.63
    i
    2.63
    J
    2.56
    A
    2.55
    a
    2.47
    l
    2.41
    R
    2.39
    Activations Density 0.101%

    No Known Activations