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

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 61375
    Prev
    Next
    INDEX
    Explanations

    can do

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

    The neuron activates on the modal verb “can,” flagging occurrences of “can” (often in “you can” or similar) that appear in suggestions or instructions.

    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
     what
    -1.82
     they
    -1.73
     which
    -1.66
     where
    -1.65
     whose
    -1.57
     when
    -1.55
     how
    -1.49
    重要
    -1.46
     will
    -1.45
    石垣
    -1.41
    POSITIVE LOGITS
     maniere
    1.64
     gants
    1.50
     plafon
    1.48
    なー
    1.48
    waiting
    1.45
    adding
    1.43
     chaussures
    1.43
    els
    1.43
    combined
    1.41
     bascul
    1.41
    Activations Density 0.072%

    No Known Activations