© 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
    2. Gemma-2-27B
    3. 34-GEMMASCOPE-RES-131K
    4. 57436
    Prev
    Next
    INDEX
    Explanations

    used phrases involving Roman numerals or specific orders

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

    The neuron consistently activates on numeric tokens or digit sequences, i.e. it spotlights numbers in the text.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/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
     westward
    -0.94
    ward
    -0.93
     eastward
    -0.90
     southward
    -0.89
    WARD
    -0.88
     northward
    -0.79
     purpose
    -0.79
    Mostly
    -0.79
    Ặ
    -0.78
     MongoDB
    -0.77
    POSITIVE LOGITS
     purchase
    0.88
    Purchase
    0.87
     Purchase
    0.85
     buy
    0.84
     purchases
    0.82
    采
    0.81
     buys
    0.79
     Anschluss
    0.77
     Buy
    0.77
     purchasing
    0.75
    Activations Density 0.011%

    No Known Activations