© 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. 32982
    Prev
    Next
    INDEX
    Explanations

    text formatting (case, special characters)

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

    This neuron is specialized to pick out non‐lexical, code or structure markers—tokens like underscores, brackets, backslashes, parse‐tree labels, and other punctuation/symbols.

    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
     
    -1.16
    </h3>
    -1.02
     also
    -1.00
    том
    -0.91
    -
    -0.90
    親
    -0.90
     only
    -0.90
    </h5>
    -0.87
     anderen
    -0.87
    ning
    -0.87
    POSITIVE LOGITS
     algumas
    1.03
     мировой
    1.02
    ሂ
    1.01
    èvent
    1.00
    このまま
    0.99
     Francji
    0.99
     recours
    0.99
     কি
    0.98
    咾
    0.97
     utilizzata
    0.97
    Activations Density 0.002%

    No Known Activations