© 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. 15851
    Prev
    Next
    INDEX
    Explanations

    journal articles and reports

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

    The neuron strongly activates on document‐structure and header tokens—particularly punctuation marks (commas) and section‐heading words or numbers used in article metadata and outlines.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    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
     should
    -1.36
     it
    -1.22
     only
    -1.19
     everything
    -1.16
     even
    -1.16
     autobús
    -1.13
     things
    -1.13
    家は
    -1.11
     enfans
    -1.06
     can
    -1.05
    POSITIVE LOGITS
     groovy
    1.13
    まれて
    1.12
    hline
    1.09
    unked
    1.06
    veröffentlichung
    1.05
    }$,
    1.04
    DOES
    1.04
    ocusing
    1.02
    bitro
    1.02
    rophilic
    1.02
    Activations Density 0.002%

    No Known Activations