© 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. 107198
    Prev
    Next
    INDEX
    Explanations

    last word or final thing

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

    The neuron activates strongly on the specific token “Last,” especially when it appears in titles or names.

    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
     last
    -0.92
     osią
    -0.80
    Brien
    -0.80
     MISCELLANEOUS
    -0.80
    最后
    -0.79
     Personalis
    -0.79
     feel
    -0.72
    สุด
    -0.72
    Pav
    -0.72
    seido
    -0.72
    POSITIVE LOGITS
    tifice
    0.86
    och
    0.83
     first
    0.83
    first
    0.81
    inetic
    0.81
    dddd
    0.81
     temptation
    0.77
    Tem
    0.77
     Moh
    0.77
    lably
    0.76
    Activations Density 0.023%

    No Known Activations