© 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. 10-GEMMASCOPE-RES-131K
    7. 71871
    Prev
    Next
    INDEX
    Explanations

    preposition + noun/verb

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

    The neuron activates on English prepositions (e.g. “of,” “from,” “with,” “to,” “for”) when they appear as capitalized connectors in names, titles, or code identifiers.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/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
     and
    -2.73
     in
    -2.16
     two
    -2.05
     after
    -2.03
     when
    -1.98
     by
    -1.95
     as
    -1.92
     now
    -1.84
     first
    -1.83
     on
    -1.82
    POSITIVE LOGITS
    熥
    1.98
     feroit
    1.98
     réellement
    1.93
    2
    1.77
    ed
    1.72
     incroy
    1.70
    蹀
    1.70
     compréhen
    1.66
     verhindert
    1.61
     acheter
    1.60
    Activations Density 0.013%

    No Known Activations