© 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. 63575
    Prev
    Next
    INDEX
    Explanations

    end of sentence punctuation followed by word

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

    This neuron strongly responds to sentence-punctuation tokens (commas, periods, exclamation marks), effectively marking sentence or clause boundaries.

    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
     pittores
    -1.80
    鳚
    -1.79
     signifikan
    -1.76
     simpel
    -1.65
     fokus
    -1.64
     illustre
    -1.59
    glück
    -1.59
    -¡
    -1.58
     charmant
    -1.58
    
    -1.57
    POSITIVE LOGITS
     would
    2.50
     will
    2.45
     only
    2.38
     didn
    2.28
     could
    2.16
     wouldn
    1.94
     have
    1.93
     always
    1.91
     can
    1.89
     lots
    1.89
    Activations Density 0.010%

    No Known Activations