© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 25996
    Prev
    Next
    INDEX
    Explanations

    phrase structure

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

    The neuron lights up on common two‐word phrases that pair a content word (often a noun) with a following preposition (e.g. “head of,” “terms of,” “changes in,” “along the,” “eye contact”).

    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
    卵
    -0.95
     cez
    -0.94
     Grüßen
    -0.90
    牛仔
    -0.88
    -0.87
     idee
    -0.84
     Ihr
    -0.83
     Ihre
    -0.83
    紺
    -0.83
    Greet
    -0.82
    POSITIVE LOGITS
     disfrutando
    0.94
    📛
    0.91
     versátil
    0.88
    ⼈
    0.86
    %>
    0.86
     widać
    0.85
    對象
    0.85
     verkau
    0.85
     him
    0.84
     giugno
    0.84
    Activations Density 0.162%

    No Known Activations