© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 27797
    Prev
    Next
    INDEX
    Explanations

    code and file artifacts

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

    The neuron strongly activates on long uninterrupted sequences of uppercase letters, digits, “+”, “/” (and similar characters) typical of Base64‐encoded blocks such as those in RSA/DH keys.

    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
     this
    -2.36
     because
    -2.34
     while
    -2.28
     might
    -2.27
     on
    -2.25
     could
    -2.13
     over
    -2.13
     in
    -2.11
     before
    -2.11
     during
    -2.08
    POSITIVE LOGITS
     ayur
    2.03
    primarily
    1.97
    nearly
    1.94
    particularly
    1.91
     Dinas
    1.88
     dando
    1.88
     shimmering
    1.86
     unbelievably
    1.85
     flamboyant
    1.83
    specifically
    1.81
    Activations Density 0.017%

    No Known Activations