© 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
    3. Gemma-2-27B
    5. 34-GEMMASCOPE-RES-131K
    7. 90294
    Prev
    Next
    INDEX
    Explanations

    word sequences or common phrases

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

    The neuron fires on common function words and morphological prefixes—e.g. English auxiliaries and pronouns like “can,” “be,” “are,” “we” and Indonesian verb or prepositional prefixes such as “Di,” “Meng,” “Mem.”

    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
     would
    -1.07
     on
    -1.07
     now
    -1.04
    があり
    -1.03
     in
    -0.98
     could
    -0.98
     to
    -0.98
     can
    -0.95
    などが
    -0.91
    つまり
    -0.90
    POSITIVE LOGITS
     debout
    1.09
    のですか
    1.07
     vrais
    1.05
     vermeld
    1.03
    Ք
    1.02
     courir
    1.00
    に挑戦
    0.98
    amię
    0.97
     biens
    0.97
    ほら
    0.97
    Activations Density 0.001%

    No Known Activations