© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    2. Gemma-3-27B-IT
    3. 40-GEMMASCOPE-2-RES-262K
    4. 7743
    Prev
    Next
    INDEX
    Explanations

    how questions and greetings

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

    This neuron strongly detects the token "How" (i.e., the beginning of "how" questions/greetings such as "How" or "Howdy").

    oai_token-act-pair · gpt-5-miniTriggered by @vetterc0
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-27b-it/resid_post/layer_40_width_262k_l0_medium
    Prompts (Dashboard)
    238,145 prompts, 512 tokens each
    Dataset (Dashboard)
    lmsys + oasst1
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     تړل
    0.41
     والك
    0.34
    слары
    0.34
     multiuser
    0.34
    muster
    0.34
     لیګ
    0.34
    Multiplier
    0.33
     ല
    0.32
     multic
    0.32
    ジャパン
    0.32
    POSITIVE LOGITS
     does
    0.60
     did
    0.59
     do
    0.57
    beit
    0.54
    soever
    0.50
     can
    0.45
     ironic
    0.45
    ?
    0.44
     else
    0.44
     dare
    0.44
    Activations Density 0.106%

    No Known Activations