© 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. 128547
    Prev
    Next
    INDEX
    Explanations

    The neuron is strongly triggered by code‐style identifiers and other multi‐word, capitalized tokens (e.g. class or module names, XML/JSON keys, and other CamelCase or uppercase labels).

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

    code delimiters and data tags

    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
     has
    -1.75
     without
    -1.73
     epitom
    -1.70
     Despite
    -1.67
     Taking
    -1.63
     before
    -1.62
     using
    -1.62
     Getting
    -1.61
     does
    -1.56
     took
    -1.56
    POSITIVE LOGITS
    甗
    1.67
    mainly
    1.62
    嘮
    1.60
    킁
    1.60
     SPECIFIC
    1.58
     dando
    1.57
    specifically
    1.56
     スリーブ
    1.52
     unbelievably
    1.52
    apparently
    1.52
    Activations Density 0.047%

    No Known Activations