© 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. 26168
    Prev
    Next
    INDEX
    Explanations

    seemed confused

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

    The neuron fires on language that segments or describes groups of people (e.g. “people across the … spectrum,” “many continue to …,” “others …”), i.e. phrases framing group distribution or opinions.

    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
     or
    -1.66
     will
    -1.41
     now
    -1.39
     has
    -1.13
     can
    -1.12
     since
    -1.10
     conector
    -1.09
     pengalaman
    -1.02
     tindakan
    -0.98
     then
    -0.98
    POSITIVE LOGITS
    seemed
    1.48
    っていました
    1.25
     Ży
    1.22
     mostaza
    1.22
     tinham
    1.20
     embreagem
    1.19
     Mě
    1.16
    OOTD
    1.15
     extremely
    1.15
    fled
    1.14
    Activations Density 0.021%

    No Known Activations