© 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. 129079
    Prev
    Next
    INDEX
    Explanations

    The neuron fires on semantically rich “open‐class” words—nouns, verbs, adjectives, and adverbs—rather than on function words or punctuation.

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

    parents wanted it

    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
     there
    -1.15
     perros
    -0.90
     at
    -0.89
    чта
    -0.88
    LECT
    -0.86
     Unter
    -0.85
    forma
    -0.84
     informazioni
    -0.84
     people
    -0.83
     person
    -0.83
    POSITIVE LOGITS
     wakil
    1.06
    GNUC
    1.05
    ulosa
    1.05
    加減
    1.00
    ęgn
    0.98
    Slik
    0.98
    alnya
    0.97
    ⾞
    0.97
    imagem
    0.96
    Klem
    0.95
    Activations Density 0.020%

    No Known Activations