© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 35140
    Prev
    Next
    INDEX
    Explanations

    working version

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

    This neuron is a “partitive/quantifier” detector – it fires on the preposition “of” (and its surrounding tokens) in constructions like “one of our projects” or “gets out of the sixth inning.”

    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
    immediately
    -1.09
    pretty
    -1.06
    cended
    -0.94
    ното
    -0.93
     cuadrada
    -0.91
    Drei
    -0.90
    essentially
    -0.89
    almost
    -0.88
     VERY
    -0.88
    earlier
    -0.88
    POSITIVE LOGITS
     one
    1.50
     someday
    1.24
     since
    1.24
     algún
    1.18
    真正
    1.14
     alguna
    1.13
     ever
    1.13
     truly
    1.04
    真正的
    0.99
     fully
    0.98
    Activations Density 0.067%

    No Known Activations