© 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. 128502
    Prev
    Next
    INDEX
    Explanations

    The neuron flags strongly emotive or evaluative words—especially high-impact adjectives and intensifiers used to express praise, emphasis, or strong sentiment.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    following descriptive words

    np_acts-logits-general · gemini-2.5-flash-lite
    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
    స్తు
    -0.93
    ppling
    -0.91
    IALS
    -0.91
     langit
    -0.91
     sonuç
    -0.90
    ঈ
    -0.90
     результатов
    -0.87
     maintenant
    -0.86
    ggable
    -0.86
     并
    -0.84
    POSITIVE LOGITS
     when
    1.36
     again
    1.20
     at
    1.04
    嗱
    1.02
     any
    1.02
    に戻
    0.94
    device
    0.93
     on
    0.93
    Ź
    0.92
     تريد
    0.91
    Activations Density 0.023%

    No Known Activations