© 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
    2. Gemma-2-27B
    3. 10-GEMMASCOPE-RES-131K
    4. 14888
    Prev
    Next
    INDEX
    Explanations

    holds or holds up

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

    The neuron is primarily triggered by occurrences of the verb “hold” (especially “hold” or “holds”) in present‐tense contexts.

    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
     ukra
    -1.64
    獁
    -1.63
    놋
    -1.60
    ⢓
    -1.59
     kuli
    -1.59
     boty
    -1.56
     dedans
    -1.52
     插图
    -1.52
     Anſ
    -1.51
    
    -1.51
    POSITIVE LOGITS
    ).
    2.16
     =
    1.51
     kommen
    1.50
     principalmente
    1.48
    );
    1.48
     trough
    1.47
    ”;
    1.40
     sorta
    1.37
     sağ
    1.36
    –
    1.36
    Activations Density 0.012%

    No Known Activations