© 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. 34175
    Prev
    Next
    INDEX
    Explanations

    excessive, very, strong, undue

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

    The neuron detects extreme modifiers—adjectives and adverbs that intensify or emphasize degree (e.g. “very,” “excessive,” “harsh,” “ultra,” “exclusively,” “long,” “solitary”).

    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
     pourraient
    -1.24
    縈
    -1.03
    极为
    -1.02
     leta
    -0.99
    もあり
    -0.98
    astify
    -0.97
     exécu
    -0.97
    もあって
    -0.97
     arrivent
    -0.96
    镲
    -0.96
    POSITIVE LOGITS
     or
    1.52
     сильно
    1.02
     too
    1.02
     excessive
    0.96
     heavily
    0.96
     trying
    0.94
     ebenfalls
    0.92
     verschillen
    0.90
     Endlich
    0.90
    或者
    0.89
    Activations Density 0.040%

    No Known Activations