© 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. 10-GEMMASCOPE-RES-131K
    7. 43947
    Prev
    Next
    INDEX
    Explanations

    political party or media reports

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

    The neuron is primarily picking out content‐bearing “named” or technical terms—proper nouns, acronyms, API/class names, programming keywords, and other standout domain‐specific words.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/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
     what
    -1.55
     who
    -1.55
     either
    -1.40
     mesmas
    -1.34
    тельностью
    -1.34
    accéder
    -1.31
    ？』
    -1.30
    犼
    -1.30
     viejas
    -1.25
     would
    -1.24
    POSITIVE LOGITS
    aruhi
    1.55
     efektif
    1.34
     Еще
    1.34
    嘻嘻
    1.33
    nehå
    1.30
    batik
    1.30
    udsman
    1.28
    衩
    1.28
     he
    1.27
     может
    1.27
    Activations Density 0.215%

    No Known Activations