© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW
    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Gemma-2-27B
    3. 22-GEMMASCOPE-RES-131K
    4. 9496
    Prev
    Next
    INDEX
    Explanations

    information and guides

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

    The neuron fires on user‐directed instructional phrasing—especially second‐person “you” plus verbs or modals (you can, will, allows you to) that guide the reader’s actions.

    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
     摄影
    -1.22
     or
    -1.18
    lorious
    -1.05
    sediaan
    -1.03
    竇
    -0.99
    tablissements
    -0.99
    晌
    -0.96
    ği
    -0.94
     one
    -0.93
    хіво
    -0.91
    POSITIVE LOGITS
    Anything
    1.18
    usions
    1.18
     further
    1.16
     INGRED
    1.13
    erunner
    1.05
     supli
    1.03
     everything
    1.02
     bune
    1.01
    Everything
    0.99
     يساعد
    0.99
    Activations Density 0.064%

    No Known Activations