© 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. 57818
    Prev
    Next
    INDEX
    Explanations

    phrases separating clauses

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

    This neuron responds to high-frequency function words and punctuation that structure discourse—especially sentence boundaries (periods, commas, question marks) and connectors like “but,” “and,” “or,” and common short prepositions.

    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
    worst
    -0.95
    🔑
    -0.93
    Ď
    -0.93
    nextProps
    -0.92
     utveckling
    -0.90
     kunst
    -0.88
    ėms
    -0.87
    ழ்
    -0.85
     我
    -0.85
     uku
    -0.85
    POSITIVE LOGITS
     let
    1.46
     along
    1.25
     explaining
    1.13
     Along
    1.13
     now
    1.13
     explain
    1.13
     not
    1.08
     talking
    1.07
     Let
    1.02
    Let
    1.02
    Activations Density 0.002%

    No Known Activations