© 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. 22-GEMMASCOPE-RES-131K
    7. 33527
    Prev
    Next
    INDEX
    Explanations

    phrases requiring translation

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

    The neuron flags tokens that aren’t English words—i.e. it activates on foreign‐language vocabulary.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_22/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
     after
    -1.91
     for
    -1.88
     an
    -1.87
     at
    -1.87
     große
    -1.70
     with
    -1.66
     as
    -1.66
     sehr
    -1.63
     first
    -1.62
     had
    -1.59
    POSITIVE LOGITS
     delightfully
    1.95
     magnific
    1.90
     deliciously
    1.88
     soooo
    1.80
     wacky
    1.80
     delectable
    1.79
     insanely
    1.79
     accentu
    1.77
     praticamente
    1.76
     !!!!
    1.73
    Activations Density 0.005%

    No Known Activations