© 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. 25567
    Prev
    Next
    INDEX
    Explanations

    contrasting terms or alternative states

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

    The neuron fires on parts of multi-word or hyphenated technical terms and quoted phrases, effectively detecting subword segments in compound expressions.

    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
    ///<
    -0.90
     etc
    -0.87
     artesanales
    -0.86
     including
    -0.80
     Including
    -0.77
     gese
    -0.75
    ARA
    -0.75
    ctl
    -0.74
    和其他
    -0.74
    Nullable
    -0.74
    POSITIVE LOGITS
     whereas
    1.33
    Whereas
    1.17
     Whereas
    1.17
    whereas
    1.02
     vs
    0.97
     contrasted
    0.94
     Gegensatz
    0.93
    に対し
    0.91
    kező
    0.90
     rather
    0.88
    Activations Density 0.172%

    No Known Activations