© 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. 128295
    Prev
    Next
    INDEX
    Explanations

    The neuron fires on tokens that signal attribution or introduction of entities—things like “introduce,” “says,” “according to,” and the relative-clause markers that set up reported or identified information.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    people and pronouns

    np_acts-logits-general · gemini-2.5-flash-lite
    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
     öpp
    -1.00
     собрать
    -0.99
    ätz
    -0.96
     모든
    -0.95
    میزی
    -0.94
    вайте
    -0.94
     Smile
    -0.93
     atriz
    -0.93
     ⓒ
    -0.93
     vanligt
    -0.92
    POSITIVE LOGITS
     who
    2.28
     whom
    1.93
     кого
    1.38
    who
    1.26
    whom
    1.23
     him
    1.20
    คนที่
    1.12
    姓名
    1.09
    Who
    1.08
     quien
    1.06
    Activations Density 0.095%

    No Known Activations