© 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. 65070
    Prev
    Next
    INDEX
    Explanations

    states definitions or intentions

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

    This neuron strongly activates on subject pronouns (especially “I” and “we”) and common linking/helper verbs (forms of “to be” and modals like “should”).

    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
     on
    -0.95
    Å
    -0.84
     first
    -0.84
    เหรียญ
    -0.82
    まず
    -0.81
     phú
    -0.79
    MessageTagHelper
    -0.79
    เป็น
    -0.79
     fermented
    -0.79
    ท่าน
    -0.79
    POSITIVE LOGITS
    カチ
    1.10
     sometimes
    1.08
     many
    0.94
    sometimes
    0.93
     hábiles
    0.93
    standigheden
    0.92
    斯克
    0.92
     něk
    0.90
     parfois
    0.90
    viens
    0.90
    Activations Density 0.175%

    No Known Activations