© 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. 34-GEMMASCOPE-RES-131K
    7. 30815
    Prev
    Next
    INDEX
    Explanations

    roller combinations (skating, derby, coaster, etc.)

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

    The neuron is strongly activated by occurrences of the word “roller” (in any context).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/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
     existencia
    -0.79
     Erwä
    -0.76
    有無
    -0.75
     Aufwand
    -0.75
     隔
    -0.73
     Protects
    -0.72
     awe
    -0.71
     horrors
    -0.70
    ぼう
    -0.69
     erkennt
    -0.69
    POSITIVE LOGITS
     coaster
    1.48
    Roller
    1.45
     Roll
    1.36
     rolling
    1.33
     roller
    1.28
     Roller
    1.24
    ROLLER
    1.23
    Roll
    1.21
     derby
    1.19
    Rolling
    1.18
    Activations Density 0.007%

    No Known Activations