© 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. 58525
    Prev
    Next
    INDEX
    Explanations

    step followed by prepositions

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

    The neuron fires strongly on occurrences of the verb “step” (and its inflected forms like “stepped” or “stepping”) when used in common multiword expressions such as “step up,” “step in,” or “step forward.”

    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
    0
    -2.73
    (
    -2.17
    >
    -2.08
    .
    -2.02
    _
    -1.87
     of
    -1.85
     —
    -1.83
    )$
    -1.81
    4
    -1.78
    ات
    -1.76
    POSITIVE LOGITS
    इटम
    2.75
    नलोड
    2.58
    鼉
    2.27
    囖
    2.25
    
    2.17
     consi
    2.09
    おわりに
    2.08
     стоковая
    2.06
     ſte
    2.02
    %;
    1.99
    Activations Density 0.008%

    No Known Activations