© 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. 100170
    Prev
    Next
    INDEX
    Explanations

    taking care of

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

    The neuron fires on phrases expressing care or treatment—especially “take care of”/“care of” and related “look after” or “good to” constructions.

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

    phrases about caring for or looking after someone or something (e.g., “take care of,” “look after”).

    oai_token-act-pair · gpt-5Triggered by @jyhe0408

    the phrase "take care of" or similar constructions involving "care of".

    oai_token-act-pair · claude-4-5-sonnetTriggered 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
     deberían
    -1.78
    andafter
    -1.58
     originalen
    -1.57
    mpagne
    -1.55
     Frankrijk
    -1.51
    xvii
    -1.50
    amsung
    -1.49
    xiii
    -1.48
     mancher
    -1.43
     embalaje
    -1.39
    POSITIVE LOGITS
    at
    1.94
    as
    1.70
    While
    1.55
    Although
    1.54
    Хотя
    1.52
    Any
    1.48
    If
    1.48
    縐
    1.47
    ’
    1.45
    Only
    1.44
    Activations Density 0.007%

    No Known Activations