© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 86622
    Prev
    Next
    INDEX
    Explanations

    very much and very well

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

    This neuron specifically detects the adverb “very,” i.e. the common intensifier “very” in text.

    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
    鱨
    -2.39
     composta
    -2.16
     dando
    -2.06
     WHICH
    -2.06
    ⮜
    -2.00
    ти
    -1.99
    茈
    -1.98
    shown
    -1.95
     dritte
    -1.94
    圌
    -1.93
    POSITIVE LOGITS
    This
    3.16
    Although
    2.97
    呦
    2.75
    Despite
    2.72
    These
    2.69
    When
    2.61
    Perhaps
    2.50
     Tieren
    2.50
    However
    2.41
    It
    2.38
    Activations Density 0.028%

    No Known Activations