© 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. 25287
    Prev
    Next
    INDEX
    Explanations

    on behalf of

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

    This neuron fires on the phrase “on behalf of,” i.e. instances where someone is acting or speaking for another party.

    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
     after
    -2.14
     seems
    -1.95
     indicate
    -1.94
     has
    -1.91
     $\
    -1.83
     makes
    -1.83
     resistir
    -1.79
     rates
    -1.77
     suggest
    -1.76
     utilize
    -1.75
    POSITIVE LOGITS
    鹟
    2.23
     dij
    2.11
     zoude
    2.03
     of
    1.98
     lasci
    1.98
    襌
    1.95
     inan
    1.93
     kristal
    1.91
     própri
    1.85
    沵
    1.85
    Activations Density 0.006%

    No Known Activations